Federal Register, Volume 78 Issue 10 (Tuesday, January 15, 2013)

[Federal Register Volume 78, Number 10 (Tuesday, January 15, 2013)]
[Rules and Regulations]
[Pages 3085-3287]
From the Federal Register Online via the Government Printing Office [www.gpo.gov]
[FR Doc No: 2012-30946]

[[Page 3085]]

Vol. 78

Tuesday,

No. 10

January 15, 2013

Part II

Environmental Protection Agency

-----------------------------------------------------------------------

40 CFR Parts 50, 51, 52 et al.

National Ambient Air Quality Standards for Particulate Matter; Final
Rule

Federal Register / Vol. 78 , No. 10 / Tuesday, January 15, 2013 /
Rules and Regulations

[[Page 3086]]

-----------------------------------------------------------------------

ENVIRONMENTAL PROTECTION AGENCY

40 CFR Parts 50, 51, 52, 53 and 58

[EPA-HQ-OAR-2007-0492; FRL-9761-8]
RIN 2060-AO47

National Ambient Air Quality Standards for Particulate Matter

AGENCY: Environmental Protection Agency (EPA).

ACTION: Final rule.

-----------------------------------------------------------------------

SUMMARY: Based on its review of the air quality criteria and the
national ambient air quality standards (NAAQS) for particulate matter
(PM), the EPA is making revisions to the suite of standards for PM to
provide requisite protection of public health and welfare and to make
corresponding revisions to the data handling conventions for PM and to
the ambient air monitoring, reporting, and network design requirements.
The EPA also is making revisions to the prevention of significant
deterioration (PSD) permitting program with respect to the NAAQS
revisions.
With regard to primary (health-based) standards for fine particles
(generally referring to particles less than or equal to 2.5 micrometers
([mu]m) in diameter, PM2.5), the EPA is revising the annual
PM2.5 standard by lowering the level to 12.0 micrograms per
cubic meter ([mu]g/m\3\) so as to provide increased protection against
health effects associated with long- and short-term exposures
(including premature mortality, increased hospital admissions and
emergency department visits, and development of chronic respiratory
disease), and to retain the 24-hour PM2.5 standard at a
level of 35 [mu]g/m\3\. The EPA is revising the Air Quality Index (AQI)
for PM2.5 to be consistent with the revised primary
PM2.5 standards. With regard to the primary standard for
particles generally less than or equal to 10 [micro]m in diameter
(PM10), the EPA is retaining the current 24-hour
PM10 standard to continue to provide protection against
effects associated with short-term exposure to thoracic coarse
particles (i.e., PM10-2.5). With regard to the secondary
(welfare-based) PM standards, the EPA is generally retaining the
current suite of secondary standards (i.e., 24-hour and annual
PM2.5 standards and a 24-hour PM10 standard).
Non-visibility welfare effects are addressed by this suite of secondary
standards, and PM-related visibility impairment is addressed by the
secondary 24-hour PM2.5 standard.

DATES: The final rule is effective on March 18, 2013.

ADDRESSES: Section X.B requests comments on an information collection
request regarding changes to the monitoring requirements. Submit your
comments, identified by Docket ID No. EPA-HQ-OAR-2007-0492, to the EPA
by one of the following methods:
www.regulations.gov: Follow the on-line instructions for
submitting comments.
Email: a-and-r-Docket@epa.gov.
Fax: 202-566-9744.
Mail: Docket No. EPA-HQ-OAR-2007-0492, Environmental
Protection Agency, Mail code 6102T, 1200 Pennsylvania Ave. NW.,
Washington, DC 20460. Please include a total of two copies.
Hand Delivery: Docket No. EPA-HQ-OAR-2007-0492,
Environmental Protection Agency, EPA West, Room 3334, 1301 Constitution
Ave. NW., Washington, DC. Such deliveries are only accepted during the
Docket's normal hours of operation, and special arrangements should be
made for deliveries of boxed information.
Instructions: Direct your comments to Docket ID No. EPA-HQ-OAR-
2007-0492. The EPA's policy is that all comments received will be
included in the public docket without change and may be made available
online at www.regulations.gov, including any personal information
provided, unless the comment includes information claimed to be
Confidential Business Information (CBI) or other information whose
disclosure is restricted by statute. Do not submit information that you
consider to be CBI or otherwise protected through www.regulations.gov
or email. The www.regulations.gov Web site is an ``anonymous access''
system, which means the EPA will not know your identity or contact
information unless you provide it in the body of your comment. If you
send an email comment directly to the EPA without going through
www.regulations.gov your email address will be automatically captured
and included as part of the comment that is placed in the public docket
and made available on the Internet. If you submit an electronic
comment, the EPA recommends that you include your name and other
contact information in the body of your comment and with any disk or
CD-ROM you submit. If the EPA cannot read your comment due to technical
difficulties and cannot contact you for clarification, the EPA may not
be able to consider your comment. Electronic files should avoid the use
of special characters, any form of encryption, and be free of any
defects or viruses. For additional information about EPA's public
docket visit the EPA Docket Center homepage at http://www.epa.gov/epahome/dockets.htm. Comments on this information collection request
should also be sent to the Office of Management and Budget (OMB). See
section X.B below for additional information regarding submitting
comments to OMB.
Docket: The EPA has established a docket for this action under
Docket No. EPA-HQ-OAR-2007-0492. All documents in the docket are listed
on the www.regulations.gov Web site. This includes documents in the
rulemaking docket (Docket ID No. EPA-HQ-OAR-2007-0492) and a separate
docket, established for 2009 Integrated Science Assessment (Docket No.
EPA-HQ-ORD-2007-0517), that has have been incorporated by reference
into the rulemaking docket. All documents in these dockets are listed
on the www.regulations.gov Web site. Although listed in the index, some
information is not publicly available, e.g., CBI or other information
whose disclosure is restricted by statute. Certain other material, such
as copyrighted material, is not placed on the Internet and may be
viewed, with prior arrangement, at the EPA Docket Center. Publicly
available docket materials are available either electronically in
www.regulations.gov or in hard copy at the Air and Radiation Docket and
Information Center, EPA/DC, EPA West, Room 3334, 1301 Constitution Ave.
NW., Washington, DC. The Public Reading Room is open from 8:30 a.m. to
4:30 p.m., Monday through Friday, excluding legal holidays. The
telephone number for the Public Reading Room is (202) 566-1744 and the
telephone number for the Air and Radiation Docket and Information
Center is (202) 566-1742. For additional information about EPA's public
docket visit the EPA Docket Center homepage at: http://www.epa.gov/epahome/dockets.htm.

FOR FURTHER INFORMATION CONTACT: Ms. Beth M. Hassett-Sipple, Health and
Environmental Impacts Division, Office of Air Quality Planning and
Standards, U.S. Environmental Protection Agency, Mail code C504-06,
Research Triangle Park, NC 27711; telephone: (919) 541-4605; fax: (919)
541-0237; email: hassett-sipple.beth@epa.gov.

SUPPLEMENTARY INFORMATION:

General Information

Availability of Related Information

A number of the documents that are relevant to this rulemaking are
available through the EPA's Office of Air Quality Planning and
Standards (OAQPS) Technology Transfer Network (TTN) Web site at http://www.epa.gov/ttn/naaqs/standards/pm/s_pm_index.html.

[[Page 3087]]

These documents include the Plan for Review of the National Ambient Air
Quality Standards for Particulate Matter (U.S. EPA, 2008a), available
at http://www.epa.gov/ttn/naaqs/standards/pm/s_pm_2007_pd.html, the
Integrated Science Assessment for Particulate Matter (U.S. EPA, 2009a),
available at http://www.epa.gov/ttn/naaqs/standards/pm/s_pm_2007_isa.html, the Quantitative Health Risk Assessment for Particulate
Matter (U.S. EPA, 2010a), available at http://www.epa.gov/ttn/naaqs/standards/pm/s_pm_2007_risk.html, the Particulate Matter Urban-
Focused Visibility Assessment (U.S. EPA 2010b), available at http://www.epa.gov/ttn/naaqs/standards/pm/s_pm_2007_risk.html, and the
Policy Assessment for the Review of the Particulate Matter National
Ambient Air Quality Standards (U.S. EPA, 2011a), available at http://www.epa.gov/ttn/naaqs/standards/pm/s_pm_2007_pa.html. These and
other related documents are also available for inspection and copying
in the EPA docket identified above.

Table of Contents

The following topics are discussed in this preamble:
I. Executive Summary
A. Purpose of This Regulatory Action
B. Summary of Major Provisions
C. Costs and Benefits
II. Background
A. Legislative Requirements
B. Review of the Air Quality Criteria and Standards for PM
1. Previous PM NAAQS Reviews
2. Litigation Related to the 2006 PM Standards
3. Current PM NAAQS Review
C. Related Control Programs To Implement PM Standards
D. Summary of Proposed Revisions to the PM NAAQS
E. Organization and Approach to Final PM NAAQS Decisions
III. Rationale for Final Decisions on the Primary PM2.5
Standards
A. Background
1. General Approach Used in Previous Reviews
2. Remand of Primary Annual PM2.5 Standard
3. General Approach Used in the Policy Assessment for the
Current Review
B. Overview of Health Effects Evidence
C. Overview of Quantitative Characterization of Health Risks
D. Conclusions on the Adequacy of the Current Primary
PM2.5 Standards
1. Introduction
a. Evidence- and Risk-based Considerations in the Policy
Assessment
b. CASAC Advice
c. Administrator's Proposed Conclusions Concerning the Adequacy
of the Current Primary PM2.5 Standards
2. Comments on the Need for Revision
3. Administrator's Final Conclusions Concerning the Adequacy of
the Current Primary PM2.5 Standards
E. Conclusions on the Elements of the Primary Fine Particle
Standards
1. Indicator
2. Averaging Time
3. Form
a. Annual Standard
b. 24-Hour Standard
4. Level
a. General Approach for Considering Standard Levels
b. Proposed Decisions on Level
i. Consideration of Alternative Standard Levels in the Policy
Assessment
ii. CASAC Advice
iii. Administrator's Proposed Decisions on the Primary
PM2.5 Standard Levels
c. Comments on Standard Levels
i. Annual Standard Level
ii. 24-Hour Standard Level
d. Administrator's Final Conclusions on the Primary
PM2.5 Standard Levels
F. Administrator's Final Decisions on the Primary
PM2.5 Standards
IV. Rationale for Final Decision on Primary PM10 Standard
A. Background
1. Previous Reviews of the PM NAAQS
a. Reviews Completed in 1987 and 1997
b. Review Completed in 2006
2. Litigation Related to the 2006 Primary PM10
Standards
3. General Approach Used in the Current Review
B. Health Effects Related to Exposure to Thoracic Coarse
Particles
C. Consideration of the Current and Potential Alternative
Standards in the Policy Assessment
1. Consideration of the Current Standard in the Policy
Assessment
2. Consideration of Potential Alternative Standards in the
Policy Assessment
D. CASAC Advice
E. Administrator's Proposed Conclusions Concerning the Adequacy
of the Current Primary PM10 Standard
F. Public Comments on the Administrator's Proposed Decision To
Retain the Primary PM10 Standard
G. Administrator's Final Decision on the Primary PM10
Standard
V. Communication of Public Health Information
VI. Rationale for Final Decisions on the Secondary PM Standards
A. Background
1. Approaches Used in Previous Reviews
2. Remand of 2006 Secondary PM2.5 Standards
3. General Approach Used in the Policy Assessment for the
Current Review
B. Proposed Decisions on Secondary PM Standards
1. PM-related Visibility Impairment
a. Nature of PM-related Visibility Impairment
i. Relationship Between Ambient PM and Visibility
ii. Temporal Variations of Light Extinction
iii. Periods During the Day of Interest for Assessment of
Visibility
iv. Exposure Durations of Interest
v. Periods of Fog and Rain
b. Public Perception of Visibility Impairment
c. Summary of Proposed Conclusions
i. Adequacy
ii. Indicator
iii. Averaging Time
iv. Form
v. Level
vi. Administrator's Proposed Conclusions
vii. Related Technical Analysis
2. Other (Non-Visibility) PM-related Welfare Effects
a. Evidence of Other Welfare Effects Related to PM
b. CASAC Advice
c. Summary of Proposed Decisions Regarding Other Welfare Effects
C. Comments on Proposed Rule
1. Comments on Proposed Secondary PM Standard for Visibility
Protection
a. Overview of Comments
b. Indicator
i. Comments on Calculated vs. Directly Measured Light Extinction
ii. Comments on Specific Aspects of Calculated Light Extinction
Indicator
c. Averaging Time
d. Form
e. Level
i. Comments on Visibility Preference Studies
ii. Specific Comments on Level
f. Need for a Distinct Secondary Standard
g. Legal Issues
h. Relationship With Regional Haze Program
2. Comments on the Proposed Decision Regarding Non-Visibility
Welfare Effects
D. Conclusions on Secondary PM Standards
1. Conclusions Regarding Secondary PM Standards To Address Non-
Visibility Welfare Effects
2. Conclusions Regarding Secondary PM Standards for Visibility
Protection
E. Administrator's Final Decisions on Secondary PM Standards
VII. Interpretation of the NAAQS for PM
A. Amendments to Appendix N: Interpretation of the NAAQS for
PM2.5
1. General
2. Monitoring Considerations
3. Requirements for Data Use and Reporting for Comparison With
the NAAQS for PM2.5
4. Comparisons with the PM2.5 NAAQS
B. Exceptional Events
C. Updates for Data Handling Procedures for Reporting the Air
Quality Index
VIII. Amendments to Ambient Monitoring and Reporting Requirements
A. Issues Related to 40 CFR Part 53 (Reference and Equivalent
Methods)
1. PM2.5 and PM10-2.5 Federal Equivalent
Methods
2. Use of Chemical Speciation Network (CSN) Methods to Support
the Proposed New Secondary PM2.5 Visibility Index NAAQS
B. Changes to 40 CFR Part 58 (Ambient Air Quality Surveillance)
1. Terminology Changes
2. Special Considerations for Comparability of PM2.5
Ambient Air Monitoring Data to the NAAQS

[[Page 3088]]

a. Revoking Use of Population-Oriented as a Condition for
Comparability of PM2.5 Monitoring Sites to the NAAQS
b. Applicability of Micro- and Middle-scale Monitoring Sites to
the Annual PM2.5 NAAQS
3. Changes to Monitoring for the National Ambient Air Monitoring
System
a. Background
b. Primary PM2.5 NAAQS
i. Addition of a Near-road Component to the PM2.5
Monitoring Network
ii. Use of PM2.5 Continuous FEMs at SLAMS
c. Revoking PM10-2.5 Speciation Requirements at NCore
Sites
d. Measurements for the Proposed New PM2.5 Visibility
Index NAAQS
4. Revisions to the Quality Assurance Requirements for SLAMS,
SPMs, and PSD
a. Quality Assurance Weight of Evidence
b. Quality Assurance Requirements for the Chemical Speciation
Network
c. Waivers for Maximum Allowable Separation of Collocated
PM2.5 Samplers and Monitors
5. Revisions To Probe and Monitoring Path Siting Criteria
a. Near-road Component to the PM2.5 Monitoring
Network
b. CSN Network
c. Reinsertion of Table E-1 to Appendix E
6. Additional Ambient Air Monitoring Topics
a. Annual Monitoring Network Plans and Periodic Assessment
b. Operating Schedules
c. Data Reporting and Certification for CSN and IMPROVE Data
d. Requirements for Archiving Filters
IX. Clean Air Act Implementation Requirements for the PM NAAQS
A. Designation of Areas
1. Overview of Clean Air Act Designations Requirements
2. Proposed Designations Schedules
3. Comments and Responses
4. Final Intended Designations Schedules
B. Section 110(a)(2) Infrastructure SIP Requirements
C. Implementing the Revised Primary Annual PM2.5 NAAQS in
Nonattainment Areas
D. Prevention of Significant Deterioration and Nonattainment New
Source Review Programs for the Revised Primary Annual PM2.5 NAAQS
1. Prevention of Significant Deterioration
a. Transition Provision (Grandfathering)
i. Proposal
ii. Comments and Responses
iii. Final Action
b. Modeling Tools and Guidance Applicable to the Revised Primary
Annual PM2.5 NAAQS
c. PSD Screening Tools: Significant Emissions Rates, Significant
Impact Levels, and Significant Monitoring Concentration
d. PSD Increments
e. Other PSD Transition Issues
2. Nonattainment New Source Review
E. Transportation Conformity Program
F. General Conformity Program
X. Statutory and Executive Order Reviews
A. Executive Order 12866: Regulatory Planning and Review and
Executive Order 13563: Improving Regulation and Regulatory Review
B. Paperwork Reduction Act
C. Regulatory Flexibility Act
D. Unfunded Mandates Reform Act
E. Executive Order 13132: Federalism
F. Executive Order 13175: Consultation and Coordination With
Indian Tribal Governments
G. Executive Order 13045: Protection of Children From
Environmental Health and Safety Risks
H. Executive Order 13211: Actions that Significantly Affect
Energy Supply, Distribution, or Use
I. National Technology Transfer and Advancement Act
J. Executive Order 12898: Federal Actions To Address
Environmental Justice in Minority Populations and Low-Income
Populations
K. Congressional Review Act
References

I. Executive Summary

A. Purpose of This Regulatory Action

Sections 108 and 109 of the Clean Air Act (CAA) govern the
establishment, review, and revision, as appropriate, of the national
ambient air quality standards (NAAQS) to protect public health and
welfare. The CAA requires periodic review of the air quality criteria--
the science upon which the standards are based--and the standards
themselves. This rulemaking is being done pursuant to these statutory
requirements. The schedule for completing this review is established by
a court order.
In 2006, the EPA completed its last review of the PM NAAQS. In that
review, the EPA took three principal actions: (1) With regard to fine
particles (generally referring to particles less than or equal to 2.5
micrometers ([mu]m) in diameter, PM2.5), at that time, the
EPA revised the level of the primary 24-hour PM2.5 standard
from 65 to 35 [mu]g/m\3\ and retained the level of the primary annual
PM2.5 standard; (2) With regard to the primary standards for
particles less than or equal to 10 [micro]m in diameter
(PM10), the EPA retained the primary 24-hour PM10
standard to continue to provide protection against effects associated
with short-term exposure to thoracic coarse particles (i.e.,
PM10-2.5) and revoked the primary annual PM10
standard; and (3) the EPA also revised the secondary standards to be
identical in all respects to the primary standards.
In subsequent litigation, the U.S. Court of Appeals for the
District of Columbia Circuit remanded the primary annual
PM2.5 standard to the EPA because the Agency had failed to
explain adequately why the standard provided the requisite protection
from both short- and long-term exposures to fine particles, including
protection for at-risk populations such as children. The court remanded
the secondary PM2.5 standards to the EPA because the Agency
failed to explain adequately why setting the secondary standards
identical to the primary standards provided the required protection for
public welfare, including protection from PM-related visibility
impairment.
The EPA initiated this review in June 2007. Between 2007 and 2011,
the EPA prepared draft and final Integrated Science Assessments, Risk
and Exposure Assessments, and Policy Assessments. Multiple drafts of
all of these documents were subject to review by the public and were
peer reviewed by the EPA's Clean Air Scientific Advisory Committee
(CASAC). The EPA proposed revisions to the primary and secondary PM
NAAQS on June 29, 2012 (77 FR 38890). This final rulemaking is the
final step in the review process.
In this rulemaking, the EPA is revising the suite of standards for
PM to provide requisite protection of public health and welfare. The
EPA is revising the PSD permitting regulations to address the changes
in the PM NAAQS. In addition, the EPA is updating the AQI for
PM2.5 and making changes in the data handling conventions
for PM and ambient air monitoring, reporting, and network design
requirements to correspond with the changes to the PM NAAQS.

B. Summary of Major Provisions

With regard to the primary standards for fine particles, the EPA is
revising the annual PM2.5 standard by lowering the level
from 15.0 to 12.0 [mu]g/m\3\ so as to provide increased protection
against health effects associated with long-and short-term exposures.
The EPA is retaining the level (35 [mu]g/m\3\) and the form (98th
percentile) of the 24-hour PM2.5 standard to continue to
provide supplemental protection against health effects associated with
short-term exposures. This action provides increased protection for
children, older adults, persons with pre-existing heart and lung
disease, and other at-risk populations against an array of
PM2.5-related adverse health effects that include premature
mortality, increased hospital admissions and emergency department
visits, and development of chronic respiratory disease. The EPA also is
eliminating spatial averaging provisions as part of the form of the
annual standard to avoid potential disproportionate impacts on at-risk
populations.
The final decisions for the primary annual and 24-hour
PM2.5 standards are

[[Page 3089]]

within the ranges that CASAC advised the Agency to consider. These
decisions are based on an integrative assessment of an extensive body
of new scientific evidence, which substantially strengthens what was
known about PM2.5-related health effects in the last review,
including extended analyses of key epidemiological studies, and
evidence of health effects observed at lower ambient PM2.5
concentrations, including effects in areas that likely met the current
standards. The revised suite of PM2.5 standards also
reflects consideration of a quantitative risk assessment that estimates
public health risks likely to remain upon just meeting the current and
various alternative standards. Based on this information, the
Administrator concludes that the current primary PM2.5
standards are not requisite to protect public health with an adequate
margin of safety, as required by the CAA, and that these revisions are
warranted to provide the appropriate degree of increased public health
protection.
With regard to the primary standard for thoracic coarse particles
(PM10-2.5), the EPA is retaining the current 24-hour
PM10 standard, with a level of 150 [mu]g/m\3\ and a one-
expected exceedance form, to continue to provide protection against
effects associated with short-term exposure to PM10-2.5
including premature mortality and increased hospital admissions and
emergency department visits. In reaching this decision, the
Administrator concludes that the available health evidence and air
quality information for PM10-2.5, taken together with the
considerable uncertainties and limitations associated with that
information, suggests that a standard is needed to protect against
short-term exposure to all types of PM10-2.5 and that the
degree of public health protection provided against short-term
exposures to PM10-2.5 does not need to be increased beyond
that provided by the current PM10 standard.
With regard to the secondary PM standards, the Administrator is
retaining the current suite of secondary PM standards, except for a
change to the form of the annual PM2.5 standard.
Specifically, the EPA is retaining the current secondary 24-hour
PM2.5 and PM10 standards, and is revising only
the form of the secondary annual PM2.5 standard to remove
the option for spatial averaging consistent with this change to the
primary annual PM2.5 standard. This suite of secondary
standards addresses PM-related non-visibility welfare effects including
ecological effects, effects on materials, and climate impacts. With
respect to PM-related visibility impairment, the Administrator has
identified a target degree of protection, defined in terms of a
PM2.5 visibility index (based on speciated PM2.5
mass concentrations and relative humidity data to calculate
PM2.5 light extinction), a 24-hour averaging time, and a
90th percentile form, averaged over 3 years, and a level of 30
deciviews (dv), which she judges to be requisite to protect public
welfare with regard to visual air quality (VAQ). The EPA's analysis of
monitoring data provides the basis for concluding that the current
secondary 24-hour PM2.5 standard would provide sufficient
protection, and in some areas greater protection, relative to this
target protection level. Adding a distinct secondary standard to
address visibility would not affect this protection. Since sufficient
protection from visibility impairment will be provided for all areas of
the country without adoption of a distinct secondary standard, and
adoption of a distinct secondary standard will not change the degree of
over-protection of VAQ provided for some areas of the country by the
secondary 24-hour PM2.5 standard, the Administrator judges
that adoption of a distinct secondary standard, in addition to the
current suite of secondary standards, is not needed to provide
requisite protection for both visibility and non-visibility related
welfare effects.
The revisions to the PM NAAQS trigger a process under which states
(and tribes, if they choose) will make recommendations to the
Administrator regarding designations, identifying areas of the country
that either meet or do not meet the revised NAAQS. States will also
review, modify and supplement their existing state implementation plans
(SIPs), as needed. With regard to these implementation-related
activities, the EPA intends to promulgate a separate implementation
rule on a schedule that provides timely clarity to the states, tribes,
and other parties responsible for NAAQS implementation. The NAAQS
revisions also affect the applicable air permitting requirement, but
cause no significant change to the transportation conformity and
general conformity processes. The EPA is revising its PSD regulations
to provide limited grandfathering from the requirements that result
from the revised PM NAAQS.
On other topics, the EPA is changing the AQI for PM2.5
to be consistent with the revised primary PM2.5 NAAQS. The
EPA also is revising the data handling procedures for PM2.5
consistent with the revised PM2.5 NAAQS including the
computations necessary for determining when the standards are met and
the measurement data that are appropriate for comparison to the
standards. With regard to monitoring-related activities, the EPA is
updating several aspects of the monitoring regulations and specifically
requiring that a small number of PM2.5 monitors be relocated
to be collocated with measurements of other pollutants (e.g., nitrogen
dioxide, carbon monoxide) in the near-road environment.

C. Costs and Benefits

In setting the NAAQS, the EPA may not consider the costs of
implementing the standards. This was confirmed by the United States
Supreme Court in Whitman v. American Trucking Associations, 531 U.S.
457, 465-472, 475-76 (2001), as noted in section II.A of this rule. As
has traditionally been done in NAAQS rulemaking, the EPA has conducted
a Regulatory Impact Analysis (RIA) to provide the public with
information on the potential costs and benefits of attaining several
alternative PM2.5 standards. In NAAQS rulemaking, the RIA is
done for informational purposes only, and the final decisions on the
NAAQS in this rulemaking are not in any way based on consideration of
the information or analyses in the RIA. The RIA fulfills the
requirements of Executive Orders 13563 and 12866. The summary of the
RIA, which is discussed in more detail below in section X.A, estimates
benefits ranging from $4,000 million to $9,100 million at a 3 percent
discount rate and $3,600 million to $8,200 million at a 7 percent
discount rate in 2020 and costs ranging from $53 million to $350
million per year at a 7 percent discount rate.

II. Background

A. Legislative Requirements

Two sections of the CAA govern the establishment, review and
revision of the NAAQS. Section 108 (42 U.S.C. 7408) directs the
Administrator to identify and list certain air pollutants and then to
issue air quality criteria for those pollutants. The Administrator is
to list those air pollutants that in her ``judgment, cause or
contribute to air pollution which may reasonably be anticipated to
endanger public health or welfare;'' ``the presence of which in the
ambient air results from numerous or diverse mobile or stationary
sources;'' and ``for which * * * [the Administrator] plans to issue air
quality criteria * * *'' Air quality criteria are intended to
``accurately reflect the latest scientific knowledge useful in
indicating the kind and extent of all identifiable effects on public
health or

[[Page 3090]]

welfare which may be expected from the presence of [a] pollutant in the
ambient air * * *'' 42 U.S.C. 7408(b). Section 109 (42 U.S.C. 7409)
directs the Administrator to propose and promulgate ``primary'' and
``secondary'' NAAQS for pollutants for which air quality criteria are
issued. Section 109(b)(1) defines a primary standard as one ``the
attainment and maintenance of which in the judgment of the
Administrator, based on such criteria and allowing an adequate margin
of safety, are requisite to protect the public health.'' \1\ A
secondary standard, as defined in section 109(b)(2), must ``specify a
level of air quality the attainment and maintenance of which, in the
judgment of the Administrator, based on such criteria, is requisite to
protect the public welfare from any known or anticipated adverse
effects associated with the presence of [the] pollutant in the ambient
air.'' \2\
---------------------------------------------------------------------------

\1\ The legislative history of section 109 indicates that a
primary standard is to be set at ``the maximum permissible ambient
air level * * * which will protect the health of any [sensitive]
group of the population,'' and that for this purpose ``reference
should be made to a representative sample of persons comprising the
sensitive group rather than to a single person in such a group.'' S.
Rep. No. 91-1196, 91st Cong., 2d Sess. 10 (1970).
\2\ Welfare effects as defined in section 302(h) (42 U.S.C.
7602(h)) include, but are not limited to, ``effects on soils, water,
crops, vegetation, man-made materials, animals, wildlife, weather,
visibility and climate, damage to and deterioration of property, and
hazards to transportation, as well as effects on economic values and
on personal comfort and well-being.''
---------------------------------------------------------------------------

The requirement that primary standards provide an adequate margin
of safety was intended to address uncertainties associated with
inconclusive scientific and technical information available at the time
of standard setting. It was also intended to provide a reasonable
degree of protection against hazards that research has not yet
identified. See Lead Industries Association v. EPA, 647 F.2d 1130, 1154
(D.C. Cir 1980); American Petroleum Institute v. Costle, 665 F.2d 1176,
1186 (D.C. Cir. 1981); American Farm Bureau Federation v. EPA, 559 F.
3d 512, 533 (D.C. Cir. 2009); Association of Battery Recyclers v. EPA,
604 F. 3d 613, 617-18 (D.C. Cir. 2010). Both kinds of uncertainties are
components of the risk associated with pollution at levels below those
at which human health effects can be said to occur with reasonable
scientific certainty. Thus, in selecting primary standards that provide
an adequate margin of safety, the Administrator is seeking not only to
prevent pollution levels that have been demonstrated to be harmful but
also to prevent lower pollutant levels that may pose an unacceptable
risk of harm, even if the risk is not precisely identified as to nature
or degree. The CAA does not require the Administrator to establish a
primary NAAQS at a zero-risk level or at background concentration
levels, see Lead Industries v. EPA, 647 F.2d at 1156 n.51, but rather
at a level that reduces risk sufficiently so as to protect public
health with an adequate margin of safety.
In addressing the requirement for an adequate margin of safety, the
EPA considers such factors as the nature and severity of the health
effects involved, the size of at-risk population(s), and the kind and
degree of the uncertainties that must be addressed. The selection of
any particular approach to providing an adequate margin of safety is a
policy choice left specifically to the Administrator's judgment. See
Lead Industries Association v. EPA, 647 F.2d at 1161-62; Whitman v.
American Trucking Associations, 531 U.S. 457, 495 (2001).
In setting standards that are ``requisite'' to protect public
health and welfare, as provided in section 109(b), the EPA's task is to
establish standards that are neither more nor less stringent than
necessary for these purposes. In so doing, the EPA may not consider the
costs of implementing the standards. See generally, Whitman v. American
Trucking Associations, 531 U.S. 457, 465-472, 475-76 (2001). Likewise,
``[a]ttainability and technological feasibility are not relevant
considerations in the promulgation of national ambient air quality
standards.'' American Petroleum Institute v. Costle, 665 F. 2d at 1185.
Section 109(d)(1) requires that ``not later than December 31, 1980,
and at 5-year intervals thereafter, the Administrator shall complete a
thorough review of the criteria published under section 108 and the
national ambient air quality standards * * * and shall make such
revisions in such criteria and standards and promulgate such new
standards as may be appropriate * * *'' Section 109(d)(2) requires that
an independent scientific review committee ``shall complete a review of
the criteria * * * and the national primary and secondary ambient air
quality standards * * * and shall recommend to the Administrator any
new * * * standards and revisions of existing criteria and standards as
may be appropriate. * * *'' Since the early 1980's, this independent
review function has been performed by the CASAC.\3\
---------------------------------------------------------------------------

\3\ The CASAC PM Review Panel is comprised of the seven members
of the chartered CASAC, supplemented by fifteen subject-matter
experts appointed by the Administrator to provide additional
scientific expertise relevant to this review of the PM NAAQS. Lists
of current CASAC members and review panels are available at: http://yosemite.epa.gov/sab/sabproduct.nsf/WebCASAC/CommitteesandMembership?OpenDocument. Members of the CASAC PM Review
Panel are listed in the CASAC letters providing advice on draft
assessment documents (Samet, 2009a-f, 2012a-d).
---------------------------------------------------------------------------

B. Review of the Air Quality Criteria and Standards for PM

1. Previous PM NAAQS Reviews
The EPA initially established NAAQS for PM under section 109 of the
CAA in 1971. Since then, the Agency has made a number of changes to
these standards to reflect continually expanding scientific
information, particularly with respect to the selection of indicator\4\
and level. Table 1 provides a summary of the PM NAAQS that have been
promulgated to date. These decisions are briefly discussed below.
---------------------------------------------------------------------------

\4\ Particulate matter is the generic term for a broad class of
chemically and physically diverse substances that exist as discrete
particles (liquid droplets or solids) over a wide range of sizes,
such that the indicator for a PM NAAQS has historically been defined
in terms of particle size ranges.

Table 1--Summary of National Ambient Air Quality Standards Promulgated for PM 1971-2006 a
----------------------------------------------------------------------------------------------------------------
Averaging
Final rule Indicator time Level Form
----------------------------------------------------------------------------------------------------------------
1971--36 FR 8186 April 30, TSP.......... 24-hour...... 260 [mu]g/m\3\ Not to be exceeded more than
1971. (primary). once per year.
150 [mu]g/m\3\......
(secondary).........
Annual....... 75 [mu]g/m\3\....... Annual average.
(primary)...........
1987--52 FR 24634 July 1, PM10......... 24-hour...... 150 [mu]g/m\3\...... Not to be exceeded more than
1987. once per year on average over
a 3-year period.
Annual....... 50 [mu]g/m\3\....... Annual arithmetic mean,
averaged over 3 years.

[[Page 3091]]

1997--62 FR 38652 July 18, PM2.5........ 24-hour...... 65 [mu]g/m\3\....... 98th percentile, averaged over
1997. 3 years.\b\
Annual....... 15.0 [mu]g/m\3\..... Annual arithmetic mean,
averaged over 3 years.c d
PM10......... 24-hour...... 150 [mu]g/m\3\...... Initially promulgated 99th
percentile, averaged over 3
years; when 1997 standards
for PM10 were vacated, the
form of 1987 standards
remained in place (not to be
exceeded more than once per
year on average over a 3-year
period).
Annual....... 50 [mu]g/m\3\....... Annual arithmetic mean,
averaged over 3 years.
2006--71 FR 61144 October PM2.5........ 24-hour...... 35 [mu]g/m\3\....... 98th percentile, averaged over
17, 2006. Annual....... 15.0 [mu]g/m\3\..... 3 years.\b\
Annual arithmetic mean,
averaged over 3 years.c e
PM10......... 24-hour...... 150 [mu]g/m\3\...... Not to be exceeded more than
once per year on average over
a 3-year period.
----------------------------------------------------------------------------------------------------------------
\a\ When not specified, primary and secondary standards are identical.
\b\ The level of the 24-hour standard is defined as an integer (zero decimal places) as determined by rounding.
For example, a 3-year average 98th percentile concentration of 35.49 [mu]g/m\3\ would round to 35 [mu]g/m\3\
and thus meet the 24-hour standard and a 3-year average of 35.50 [mu]g/m\3\ would round to 36 and, hence,
violate the 24-hour standard (40 CFR part 50, appendix N).
\c\ The level of the annual standard is defined to one decimal place (i.e., 15.0 [mu]g/m\3\) as determined by
rounding. For example, a 3-year average annual mean of 15.04 [mu]g/m\3\ would round to 15.0 [mu]g/m\3\ and,
thus, meet the annual standard and a 3-year average of 15.05 [mu]g/m\3\ would round to 15.1 [mu]g/m\3\ and,
hence, violate the annual standard (40 CFR part 50, appendix N).
\d\ The level of the standard was to be compared to measurements made at sites that represent ``community-wide
air quality'' recording the highest level, or, if specific requirements were satisfied, to average
measurements from multiple community-wide air quality monitoring sites (``spatial averaging'').
\e\ The EPA tightened the constraints on the spatial averaging criteria by further limiting the conditions under
which some areas may average measurements from multiple community-oriented monitors to determine compliance
(See 71 FR 61165 to 61167, October 17, 2006).

In 1971, the EPA established NAAQS for PM based on the original air
quality criteria document (DHEW, 1969; 36 FR 8186, April 30, 1971). The
reference method specified for determining attainment of the original
standards was the high-volume sampler, which collects PM up to a
nominal size of 25 to 45 [mu]m (referred to as total suspended
particles or TSP). The primary standards (measured by the indicator
TSP) were 260 [mu]g/m\3\, 24-hour average, not to be exceeded more than
once per year, and 75 [mu]g/m\3\, annual geometric mean. The secondary
standard was 150 [mu]g/m\3\, 24-hour average, not to be exceeded more
than once per year.
In October 1979, the EPA announced the first periodic review of the
criteria and NAAQS for PM, and significant revisions to the original
standards were promulgated in 1987 (52 FR 24634, July 1, 1987). In that
decision, the EPA changed the indicator for PM from TSP to
PM10, the latter including particles with an aerodynamic
diameter less than or equal to a nominal 10 [micro]m, which delineates
thoracic particles (i.e., that subset of inhalable particles small
enough to penetrate beyond the larynx to the thoracic region of the
respiratory tract). The EPA also revised the primary standards by (1)
replacing the 24-hour TSP standard with a 24-hour PM10
standard of 150 [mu]g/m\3\ with no more than one expected exceedance
per year and (2) replacing the annual TSP standard with a
PM10 standard of 50 [mu]g/m\3\, annual arithmetic mean. The
secondary standard was revised by replacing it with 24-hour and annual
PM10 standards identical in all respects to the primary
standards. The revisions also included a new reference method for the
measurement of PM10 in the ambient air and rules for
determining attainment of the new standards. On judicial review, the
revised standards were upheld in all respects. Natural Resources
Defense Council v. EPA, 902 F. 2d 962 (D.C. Cir. 1990).
In April 1994, the EPA announced its plans for the second periodic
review of the criteria and NAAQS for PM, and promulgated significant
revisions to the NAAQS in 1997 (62 FR 38652, July 18, 1997). Most
significantly, the EPA determined that although the PM NAAQS should
continue to focus on thoracic particles (PM10), the fine and
coarse fractions of PM10 should be considered separately.
New standards were added, using PM2.5 as the indicator for
fine particles. The PM10 standards were retained for the
purpose of regulating the coarse fraction of PM10 (referred
to as thoracic coarse particles or PM10-2.5).\5\ The EPA
established two new PM2.5 standards: an annual standard of
15.0 [mu]g/m\3\, based on the 3-year average of annual arithmetic mean
PM2.5 concentrations from single or multiple monitors sited
to represent community-wide air quality\6\ and a 24-hour standard of 65
[mu]g/m\3\, based on the 3-year average of the 98th percentile of 24-
hour PM2.5 concentrations at each population-oriented
monitor\7\ within an area. Also, the EPA established a new reference
method for the measurement of PM2.5 in the ambient air and
rules for determining attainment of the new standards. To continue to
address thoracic coarse particles, the annual PM10 standard
was retained, while the form, but not the level, of the 24-hour
PM10 standard was revised to be based on the 99th percentile
of 24-hour PM10 concentrations at each monitor in an area.
The EPA revised the secondary standards by making them identical in all
respects to the primary standards.
---------------------------------------------------------------------------

\5\ See 40 CFR parts 50, 53, and 58 for more information on
reference and equivalent methods for measuring PM in ambient air.
\6\ Monitoring stations sited to represent community-wide air
quality would typically be at the neighborhood or urban-scale;
however, where a population-oriented micro or middle-scale
PM2.5 monitoring station represents many such locations
throughout a metropolitan area, these smaller scales might also be
considered to represent community-wide air quality [40 CFR part 58,
appendix D, 4.7.1(b)].
\7\ Population-oriented monitoring (or sites) means residential
areas, commercial areas, recreational areas, industrial areas where
workers from more than one company are located, and other areas
where a substantial number of people may spend a significant
fraction of their day (40 CFR 58.1).
---------------------------------------------------------------------------

Following promulgation of the revised PM NAAQS in 1997, petitions
for review were filed by a large number of

[[Page 3092]]

parties, addressing a broad range of issues. In May 1998, a three-judge
panel of the U.S. Court of Appeals for the District of Columbia Circuit
issued an initial decision that upheld the EPA's decision to establish
fine particle standards, holding that ``the growing empirical evidence
demonstrating a relationship between fine particle pollution and
adverse health effects amply justifies establishment of new fine
particle standards.'' American Trucking Associations v. EPA, 175 F. 3d
1027, 1055-56 (D.C. Cir. 1999), rehearing granted in part and denied in
part, 195 F. 3d 4 (D.C. Cir. 1999), affirmed in part and reversed in
part, Whitman v. American Trucking Associations, 531 U.S. 457 (2001).
The panel also found ``ample support'' for the EPA's decision to
regulate coarse particle pollution, but vacated the 1997
P.M.10 standards, concluding, in part, that PM10
is a ``poorly matched indicator for coarse particulate pollution''
because it includes fine particles. Id. at 1053-55. Pursuant to the
court's decision, the EPA removed the vacated 1997 P.M.10
standards from the CFR (69 FR 45592, July 30, 2004) and deleted the
regulatory provision (at 40 CFR 50.6(d)) that controlled the transition
from the pre-existing 1987 P.M.10 standards to the 1997
P.M.10 standards. The pre-existing 1987 P.M.10
standards remained in place (65 FR 80776, December 22, 2000). The court
also upheld the EPA's determination not to establish more stringent
secondary standards for fine particles to address effects on visibility
(175 F. 3d at 1027).
More generally, the panel held (over a strong dissent) that the
EPA's approach to establishing the level of the standards in 1997, both
for the PM and for the ozone NAAQS promulgated on the same day,
effected ``an unconstitutional delegation of legislative authority.''
Id. at 1034-40. Although the panel stated that ``the factors EPA uses
in determining the degree of public health concern associated with
different levels of ozone and PM are reasonable,'' it remanded the rule
to the EPA, stating that when the EPA considers these factors for
potential non-threshold pollutants ``what EPA lacks is any determinate
criterion for drawing lines'' to determine where the standards should
be set. Consistent with the EPA's long-standing interpretation and D.C.
Circuit precedent, the panel also reaffirmed its prior holdings that in
setting NAAQS, the EPA is ``not permitted to consider the cost of
implementing those standards.'' Id. at 1040-41.
On EPA's petition for rehearing, the panel adhered to its position
on these points. American Trucking Associations v. EPA, 195 F. 3d 4
(D.C. Cir. 1999). The full Court of Appeals denied the EPA's request
for rehearing en banc, with five judges dissenting. Id. at 13. Both
sides filed cross appeals on these issues to the United States Supreme
Court, which granted certiorari. In February 2001, the Supreme Court
issued a unanimous decision upholding the EPA's position on both the
constitutional and cost issues. Whitman v. American Trucking
Associations, 531 U.S. 457, 464, 475-76. On the constitutional issue,
the Court held that the statutory requirement that NAAQS be
``requisite'' to protect public health with an adequate margin of
safety sufficiently cabined the EPA's discretion, affirming the EPA's
approach of setting standards that are neither more nor less stringent
than necessary. The Supreme Court remanded the case to the Court of
Appeals for resolution of any remaining issues that had not been
addressed in that court's earlier rulings. Id. at 475-76. In March
2002, the Court of Appeals rejected all remaining challenges to the
standards, holding under the statutory standard of review that the
EPA's PM2.5 standards were reasonably supported by the
administrative record and were not ``arbitrary and capricious.''
American Trucking Association v. EPA, 283 F. 3d 355, 369-72 (D.C. Cir.
2002).
In October 1997, the EPA published its plans for the next periodic
review of the air quality criteria and NAAQS for PM (62 FR 55201,
October 23, 1997). After CASAC and public review of several drafts, the
EPA's National Center for Environmental Assessment (NCEA) finalized the
Air Quality Criteria Document for Particulate Matter (henceforth, AQCD
or the ``Criteria Document'') in October 2004 (U.S. EPA, 2004) and
OAQPS finalized an assessment document, Particulate Matter Health Risk
Assessment for Selected Urban Areas (Abt Associates, 2005), and the
Review of the National Ambient Air Quality Standards for Particulate
Matter: Policy Assessment of Scientific and Technical Information, in
December 2005 (henceforth, ``Staff Paper,'' U.S. EPA, 2005). In
conjunction with its review of the Staff Paper, CASAC provided advice
to the Administrator on revisions to the PM NAAQS (Henderson, 2005a).
In particular, most CASAC PM Panel members favored revising the level
of the primary 24-hour PM2.5 standard within the range of 35
to 30 [mu]g/m\3\ with a 98th percentile form, in concert with revising
the level of the primary annual PM2.5 standard within the
range of 14 to 13 [mu]g/m\3\ (Henderson, 2005a, p.7). For thoracic
coarse particles, the Panel had reservations in recommending a primary
24-hour PM10-2.5 standard, and agreed that there was a need
for more research on the health effects of thoracic coarse particles
(Henderson, 2005b). With regard to secondary standards, most Panel
members strongly supported establishing a new, distinct secondary
PM2.5 standard to protect urban visibility (Henderson,
2005a, p. 9).
On January 17, 2006, the EPA proposed to revise the primary and
secondary NAAQS for PM (71 FR 2620) and solicited comment on a broad
range of options. Proposed revisions included: (1) Revising the level
of the primary 24-hour PM2.5 standard to 35 [mu]g/m\3\; (2)
revising the form, but not the level, of the primary annual
PM2.5 standard by tightening the constraints on the use of
spatial averaging; (3) replacing the primary 24-hour PM10
standard with a 24-hour standard defined in terms of a new indicator,
PM10-2.5, which was qualified so as to include any ambient
mix of PM10-2.5 dominated by particles generated by high-
density traffic on paved roads, industrial sources, and construction
sources, and to exclude any ambient mix of particles dominated by rural
windblown dust and soils and agricultural and mining sources (71 FR
2667 to 2668), set at a level of 70 [mu]g/m\3\ based on the 3-year
average of the 98th percentile of 24-hour PM10-2.5
concentrations; (4) revoking the primary annual PM10
standard; and (5) revising the secondary standards by making them
identical in all respects to the proposed suite of primary standards
for fine and coarse particles.\8\ Subsequent to the proposal, CASAC
provided additional advice to the EPA in a letter to the Administrator
requesting reconsideration of CASAC's recommendations for both the
primary and secondary PM2.5 standards as well as the
standards for thoracic coarse particles (Henderson, 2006a).
---------------------------------------------------------------------------

\8\ In recognition of an alternative view expressed by most
members of the CASAC PM Panel, the Agency also solicited comments on
a subdaily (4- to 8-hour averaging time) secondary PM2.5
standard to address visibility impairment, considering alternative
standard levels within a range of 20 to 30 [mu]g/m\3\ in conjunction
with a form within a range of the 92nd to 98th percentile (71 FR
2685, January 17, 2006).
---------------------------------------------------------------------------

On October 17, 2006, the EPA published revisions to the PM NAAQS to
provide increased protection of public health and welfare (71 FR
61144). With regard to the primary and secondary standards for fine
particles, the EPA revised the level of the primary 24-hour
PM2.5 standard to 35 [mu]g/m\3\, retained the level of the
primary annual PM2.5 standard at 15.0 [mu]g/m\3\, and

[[Page 3093]]

revised the form of the primary annual PM2.5 standard by
adding further constraints on the optional use of spatial averaging.
The EPA revised the secondary standards for fine particles by making
them identical in all respects to the primary standards. With regard to
the primary and secondary standards for thoracic coarse particles, the
EPA retained the level and form of the 24-hour PM10 standard
(such that the standard remained at a level of 150 [mu]g/m\3\ with a
one-expected exceedance form and retained the PM10
indicator) and revoked the annual PM10 standard. The EPA
also established a new Federal Reference Method (FRM) for the
measurement of PM10-2.5 in the ambient air (71 FR 61212 to
13). Although the standards for thoracic coarse particles were not
defined in terms of a PM10-2.5 indicator, the EPA adopted a
new FRM for PM10-2.5 to facilitate consistent research on
PM10-2.5 air quality and health effects and to promote
commercial development of Federal Equivalent Methods (FEMs) to support
future reviews of the PM NAAQS (71 FR 61212/2).
Following issuance of the final rule, CASAC articulated its concern
that the ``EPA's final rule on the NAAQS for PM does not reflect
several important aspects of the CASAC's advice'' (Henderson et al.,
2006b, p. 1). With regard to the primary PM2.5 annual
standard, CASAC expressed serious concerns regarding the decision to
retain the level of the standard at 15 [mu]g/m\3\. Specifically, CASAC
stated, ``It is the CASAC's consensus scientific opinion that the
decision to retain without change the annual PM2.5 standard
does not provide an `adequate margin of safety * * * requisite to
protect the public health' (as required by the Clean Air Act), leaving
parts of the population of this country at significant risk of adverse
health effects from exposure to fine PM'' (Henderson et al., 2006b, p.
2). Furthermore, CASAC pointed out that its recommendations ``were
consistent with the mainstream scientific advice that EPA received from
virtually every major medical association and public health
organization that provided their input to the Agency'' (Henderson et
al., 2006b, p. 2).\9\ With regard to EPA's final decision to retain the
24-hour PM10 standard for thoracic coarse particles, CASAC
had mixed views with regard to the decision to retain the 24-hour
standard and the continued use of PM10 as the indicator of
coarse particles, while also recognizing the need to have a standard in
place to protect against effects associated with short-term exposures
to thoracic coarse particles (Henderson et al., 2006b, p. 2). With
regard to the EPA's final decision to revise the secondary
PM2.5 standards to be identical in all respects to the
revised primary PM2.5 standards, CASAC expressed concerns
that its advice to establish a distinct secondary standard for fine
particles to address visibility impairment was not followed and
emphasized ``that continuing to rely on the primary standard to protect
against all PM-related adverse environmental and welfare effects
assures neglect, and will allow substantial continued degradation, of
visual air quality over large areas of the country'' (Henderson et al,
2006b, p. 2).
---------------------------------------------------------------------------

\9\ CASAC specifically identified input provided by the American
Medical Association, the American Thoracic Society, the American
Lung Association, the American Academy of Pediatrics, the American
College of Cardiology, the American Heart Association, the American
Cancer Society, the American Public Health Association, and the
National Association of Local Boards of Health (Henderson et al.,
2006b, p. 2).
---------------------------------------------------------------------------

2. Litigation Related to the 2006 PM Standards
Several parties filed petitions for review following promulgation
of the revised PM NAAQS in 2006. These petitions addressed the
following issues: (1) Selecting the level of the primary annual
PM2.5 standard; (2) retaining PM10 as the
indicator of a standard for thoracic coarse particles, retaining the
level and form of the 24-hour PM10 standard, and revoking
the PM10 annual standard; and (3) setting the secondary
PM2.5 standards identical to the primary standards. On
February 24, 2009, the U.S. Court of Appeals for the District of
Columbia Circuit issued its opinion in the case American Farm Bureau
Federation v. EPA, 559 F. 3d 512 (D.C. Cir. 2009). The court remanded
the primary annual PM2.5 NAAQS to the EPA because the EPA
failed to adequately explain why the standard provided the requisite
protection from both short- and long-term exposures to fine particles,
including protection for at-risk populations such as children. American
Farm Bureau Federation v. EPA, 559 F. 3d 512, 520-27 (D.C. Cir. 2009).
With regard to the standards for PM10, the court upheld the
EPA's decisions to retain the 24-hour PM10 standard to
provide protection from thoracic coarse particle exposures and to
revoke the annual PM10 standard. American Farm Bureau
Federation v. EPA, 559 F. 2d at 533-38. With regard to the secondary
PM2.5 standards, the court remanded the standards to the EPA
because the Agency's decision was ``unreasonable and contrary to the
requirements of section 109(b)(2)'' of the CAA. The court further
concluded that the EPA failed to adequately explain why setting the
secondary PM standards identical to the primary standards provided the
required protection for public welfare, including protection from
visibility impairment. American Farm Bureau Federation v. EPA, 559 F.
2d at 528-32.
The decisions of the court with regard to these three issues are
discussed further in sections III.A.2, IV.A.2, and VI.A.2 below. The
EPA is responding to the court's remands as part of the current review
of the PM NAAQS.
3. Current PM NAAQS Review
The EPA initiated the current review of the air quality criteria
for PM in June 2007 with a general call for information (72 FR 35462,
June 28, 2007). In July 2007, the EPA held two ``kick-off'' workshops
on the primary and secondary PM NAAQS, respectively (72 FR 34003 to
34004, June 20, 2007).\10\ These workshops provided an opportunity for
a public discussion of the key policy-relevant issues around which the
EPA would structure this PM NAAQS review and the most meaningful new
science that would be available to inform our understanding of these
issues.
---------------------------------------------------------------------------

\10\ See workshop materials available at: http://www.regulations.gov/search/Regs/home.html#home Docket ID numbers
EPA-HQ-OAR-2007-0492-008; EPA-HQ-OAR-2007-0492-009; EPA-HQ-OAR-2007-
0492-010; and EPA-HQ-OAR-2007-0492-012.
---------------------------------------------------------------------------

Based in part on the workshop discussions, the EPA developed a
draft Integrated Review Plan outlining the schedule, process, and key
policy-relevant questions that would guide the evaluation of the air
quality criteria for PM and the review of the primary and secondary PM
NAAQS (U.S. EPA, 2007a). On November 30, 2007, the EPA held a
consultation with CASAC on the draft Integrated Review Plan (72 FR
63177, November 8, 2007), which included the opportunity for public
comment. The final Integrated Review Plan (U.S. EPA, 2008a)
incorporated comments from CASAC (Henderson, 2008) and the public on
the draft plan as well as input from senior Agency
managers.^{11 12}
---------------------------------------------------------------------------

\11\ The process followed in this review varies from the NAAQS
review process described in section 1.1 of the Integrated Review
Plan (U.S. EPA, 2008a). On May 21, 2009, Administrator Jackson
called for key changes to the NAAQS review process including
reinstating a policy assessment document that contains staff
analyses of the scientific bases for alternative policy options for
consideration by senior Agency management prior to rulemaking. In
conjunction with this change, the EPA will no longer issue a policy
assessment in the form of an advance notice of proposed rulemaking
(ANPR) as discussed in the Integrated Review Plan (U.S. EPA, 2008a,
p. 3). For more information on the overall process followed in this
review including a description of the major elements of the process
for reviewing NAAQS see Jackson (2009).
\12\ All written comments submitted to the Agency are available
in the docket for this PM NAAQS review (EPA-HQ-OAR-2007-0429).
Transcripts of public meetings and teleconferences held in
conjunction with CASAC's reviews are also included in the docket.

---------------------------------------------------------------------------

[[Page 3094]]

A major element in the process for reviewing the NAAQS is the
development of an Integrated Science Assessment. This document provides
a concise evaluation and integration of the policy-relevant science,
including key science judgments upon which the risk and exposure
assessments build. As part of the process of preparing the PM
Integrated Science Assessment, NCEA hosted a peer review workshop in
June 2008 on preliminary drafts of key Integrated Science Assessment
chapters (73 FR 30391, May 27, 2008). CASAC and the public reviewed the
first external review draft Integrated Science Assessment (U.S. EPA,
2008b; 73 FR 77686, December 19, 2008) at a meeting held on April 1 to
2, 2009 (74 FR 2688, February 19, 2009). Based on CASAC (Samet, 2009e)
and public comments, NCEA prepared a second draft Integrated Science
Assessment (U.S. EPA, 2009b; 74 FR 38185, July 31, 2009), which was
reviewed by CASAC and the public at a meeting held on October 5 and 6,
2009 (74 FR 46586, September 10, 2009). Based on CASAC (Samet, 2009f)
and public comments, NCEA prepared the final Integrated Science
Assessment titled Integrated Science Assessment for Particulate Matter,
December 2009 (U.S. EPA, 2009a; 74 FR 66353, December 15, 2009).
Building upon the information presented in the PM Integrated
Science Assessment, the EPA prepared Risk and Exposure Assessments that
provide a concise presentation of the methods, key results,
observations, and related uncertainties. In developing the Risk and
Exposure Assessments for this PM NAAQS review, OAQPS released two
planning documents: Particulate Matter National Ambient Air Quality
Standards: Scope and Methods Plan for Health Risk and Exposure
Assessment and Particulate Matter National Ambient Air Quality
Standards: Scope and Methods Plan for Urban Visibility Impact
Assessment (henceforth, Scope and Methods Plans, U.S. EPA, 2009c,d; 74
FR 11580, March 18, 2009). These planning documents outlined the scope
and approaches that staff planned to use in conducting quantitative
assessments as well as key issues that would be addressed as part of
the assessments. In designing and conducting the initial health risk
and visibility impact assessments, the Agency considered CASAC comments
(Samet 2009a,b) on the Scope and Methods Plans made during an April
2009 consultation (74 FR 7688, February 19, 2009) as well as public
comments. CASAC and the public reviewed two draft assessment documents,
Risk Assessment to Support the Review of the PM2.5 Primary National
Ambient Air Quality Standards: External Review Draft, September 2009
(U.S. EPA, 2009e) and Particulate Matter Urban-Focused Visibility
Assessment--External Review Draft, September 2009 (U.S. EPA, 2009f) at
a meeting held on October 5 and 6, 2009 (74 FR 46586, September 10,
2009). Based on CASAC (Samet 2009c,d) and public comments, OAQPS staff
revised these draft documents and released second draft assessment
documents (U.S. EPA, 2010d,e) in January and February 2010 (75 FR 4067,
January 26, 2010) for CASAC and public review at a meeting held on
March 10 and 11, 2010 (75 FR 8062, February 23, 2010). Based on CASAC
(Samet, 2010a,b) and public comments on the second draft assessment
documents, the EPA revised these documents and released final
assessment documents titled Quantitative Health Risk Assessment for
Particulate Matter, June 2010 (henceforth, ``Risk Assessment,'' U.S.
EPA, 2010a) and Particulate Matter Urban-Focused Visibility
Assessment--Final Document, July 2010 (henceforth, ``Visibility
Assessment,'' U.S. EPA, 2010b) (75 FR 39252, July 8, 2010).
Based on the scientific and technical information available in this
review as assessed in the Integrated Science Assessment and Risk and
Exposure Assessments, the EPA staff prepared a Policy Assessment. The
Policy Assessment is intended to help ``bridge the gap'' between the
relevant scientific information and assessments and the judgments
required of the Administrator in reaching decisions on the NAAQS
(Jackson, 2009, attachment, p. 2). American Farm Bureau Federation v.
EPA, 559 F. 3d at 521. The Policy Assessment is not a decision
document; rather it presents the EPA staff conclusions related to the
broadest range of policy options that could be supported by the
currently available information. A preliminary draft Policy Assessment
(U.S. EPA, 2009g) was released in September 2009 for informational
purposes and to facilitate discussion with CASAC at the October 5 and
6, 2009 meeting on the overall structure, areas of focus, and level of
detail to be included in the Policy Assessment. The EPA considered
CASAC's comments on this preliminary draft in developing a first draft
Policy Assessment (U.S. EPA, 2010c; 75 FR 4067, January 26, 2010) that
built upon the information presented and assessed in the final
Integrated Science Assessment and second draft Risk and Exposure
Assessments. The EPA presented an overview of the first draft Policy
Assessment at a CASAC meeting on March 10, 2010 (75 FR 8062, February
23, 2010) and it was discussed during public CASAC teleconferences on
April 8 and 9, 2010 (75 FR 8062, February 23, 2010) and May 7, 2010 (75
FR 19971, April 16, 2010).
The EPA developed a second draft Policy Assessment (U.S. EPA,
2010f; 75 FR 39253, July 8, 2010) based on CASAC (Samet, 2010c) and
public comments on the first draft Policy Assessment. CASAC reviewed
the second draft document at a meeting on July 26 and 27, 2010 (75 FR
32763, June 9, 2010). The EPA staff considered CASAC (Samet, 2010d) and
public comments on the second draft Policy Assessment in preparing a
final Policy Assessment titled Policy Assessment for the Review of the
Particulate Matter National Ambient Air Quality Standards, April, 2011
(U.S. EPA, 2011a; 76, FR 22665, April 22, 2011). This document includes
final staff conclusions on the adequacy of the current PM standards and
alternative standards for consideration.
The schedule for the rulemaking in this review is subject to a
court order in a lawsuit filed in February 2012 by a group of
plaintiffs who alleged that the EPA had failed to perform its mandatory
duty, under section 109(d)(1), to complete a review of the PM NAAQS
within the period provided by statute. American Lung Association and
National Parks Conservation Association v. EPA, D.D.C. No. 12-cv-00243
(consol. with No. 12-cv-00531) Court orders in that case provide that
the EPA sign a notice of proposed rulemaking concerning its review of
the PM NAAQS no later than June 14, 2012 and a notice of final
rulemaking no later than December 14, 2012.
On June 14, 2012, the EPA issued its proposed decision to revise
the NAAQS for PM (77 FR 38890, June 29, 2012) (henceforth
``proposal''). In the proposal, the EPA identified revisions to the
standards, based on the air quality criteria for PM, and to related
data handling conventions and ambient air monitoring, reporting, and
network design requirements. The EPA proposed revisions to the PSD
permitting program with respect to the proposed NAAQS revisions. The
Agency also proposed

[[Page 3095]]

changes to the AQI for PM2.5, consistent with the proposed
primary PM2.5 standards. The proposal solicited public
comments on alternative primary and secondary standards and related
matters. The proposal is summarized in section II.D below.
The EPA held two public hearings to receive public comment on the
proposed revisions to the PM NAAQS (77 FR 39205, July 2, 2012). One
hearing took place in Philadelphia, PA on July 17, 2012 and a second
hearing took place in Sacramento, CA on July 19, 2012. At these public
hearings, the EPA heard testimony from 168 individuals representing
themselves or specific interested organizations.
The EPA received more than 230,000 comments from members of the
public and various interest groups on the proposed revisions to the PM
NAAQS by the close of the public comment period on August 31, 2012.
Major issues raised in the public comments are discussed throughout the
preamble of this final action. A more detailed summary of all
significant comments, along with the EPA's responses (henceforth
``Response to Comments'') can be found in the docket for this
rulemaking (Docket No. EPA-HQ-OAR-2007-0492) (U.S. EPA, 2012a).
In the proposal, the EPA recognized that there were a number of new
scientific studies on the health effects of PM that had been published
since the mid-2009 cutoff date for inclusion in the Integrated Science
Assessment.\13\ As in the last PM NAAQS review, the EPA committed to
conduct a provisional review and assessment of any significant ``new''
studies published since the close of the Integrated Science Assessment,
including studies submitted to the EPA during the public comment
period. The purpose of the provisional science assessment was to ensure
that the Administrator was fully aware of the ``new'' science that has
developed since 2009 before making final decisions on whether to retain
or revise the current PM NAAQS. The EPA screened and surveyed the
recent health literature, including studies submitted during the public
comment period, and conducted a provisional assessment (U.S. EPA,
2012b) that places the results of those studies of potentially greatest
policy relevance in the context of the findings of the Integrated
Science Assessment (U.S. EPA, 2009a). This provisional assessment,
including a summary of the key conclusions, can be found in the
rulemaking docket (EPA-HQ-OAR-2007-0492).
---------------------------------------------------------------------------

\13\ For ease of reference, these studies will be referred to as
``new'' studies or ``new'' science, using quotation marks around the
word new. Referring to studies that were published too recently to
have been included in the 2009 Integrated Science Assessment as
``new'' studies is intended to clearly differentiate such studies
from those that have been published since the last review and which
are included in the Integrated Science Assessment (these studies are
sometimes referred to as new (without quotation marks) or more
recent studies, to indicate that they were not included in the
Integrated Science Assessment and thus are newly available in this
review).
---------------------------------------------------------------------------

The provisional assessment found that the ``new'' studies expand
the scientific information considered in the Integrated Science
Assessment and provide important insights on the relationship between
PM exposure and health effects. The provisional assessment also found
that the ``new'' studies generally strengthen the evidence that long-
and short-term exposures to fine particles are associated with a wide
range of health effects. Some of the ``new'' epidemiological studies
report effects in areas with lower PM2.5-concentrations than
those in earlier studies considered in the Integrated Science
Assessment. ``New'' toxicological and epidemiological studies continue
to link various health effects with a range of fine particle sources
and components. With regard to thoracic coarse particles, the
provisional assessment recognized that a limited number of ``new''
studies provide evidence of an association with short-term
PM10-2.5 exposures and increased asthma-related emergency
department visits in children, but continue to provide no evidence of
an association between long-term PM10-2.5 exposure and
mortality. Further, the provisional assessment found that the results
reported in ``new'' studies do not materially change any of the broad
scientific conclusions regarding the health effects of PM exposure made
in the Integrated Science Assessment.
The EPA believes it was important to conduct a provisional
assessment in this proceeding, so that the Administrator would be aware
of the science that developed too recently for inclusion in the
Integrated Science Assessment. However, it is also important to note
that the EPA's review of that science to date has been limited to
screening, surveying, and preparing a provisional assessment of these
studies. Having performed this limited provisional assessment, the EPA
must decide whether to consider the ``new'' studies in this review and
to take such steps as may be necessary to include them in the basis for
the final decision, or to reserve such action for the next review of
the PM NAAQS.
As in prior NAAQS reviews, the EPA is basing its decision in this
review on studies and related information included in the Integrated
Science Assessment, Risk and Exposure Assessment, and Policy
Assessment, which have undergone CASAC and public review. The studies
assessed in the Integrated Science Assessment, and the integration of
the scientific evidence presented in that document, have undergone
extensive critical review by the EPA, CASAC, and the public during the
development of the Integrated Science Assessment. The rigor of that
review makes these studies, and their integrative assessment, the most
reliable source of scientific information on which to base decisions on
the NAAQS. NAAQS decisions can have profound impacts on public health
and welfare, and NAAQS decisions should be based on studies that have
been rigorously assessed in an integrative manner not only by the EPA
but also by the statutorily-mandated independent advisory committee,
CASAC, and have been subject as well to the public review that
accompanies this process. As described above, the provisional
assessment did not and could not provide that kind of in-depth critical
review.
This decision is consistent with the EPA's practice in prior NAAQS
reviews. Since the 1970 amendments, the EPA has taken the view that
NAAQS decisions are to be based on scientific studies and related
information that have been assessed as a part of the pertinent air
quality criteria. See e.g., 36 FR 8186 (April 30, 1971) (the EPA based
original NAAQS for six pollutants on scientific studies discussed in
air quality criteria documents and limited consideration of comments to
those concerning validity of scientific basis); 38 FR 25678, 25679-
25680 (September 14, 1973) (the EPA revised air quality criteria for
sulfur oxides to provide basis for reevaluation of secondary NAAQS).
This longstanding interpretation was strengthened by new legislative
requirements enacted in 1977, which added section 109(d)(2) of the CAA
concerning CASAC review of air quality criteria. The EPA has
consistently followed this approach. 52 FR 24634, 24637 (July 1, 1987)
(after review by CASAC, the EPA issued a post-proposal addendum to the
PM Air Quality Criteria Document, to address certain new scientific
studies not included in the 1982 Air Quality Criteria Document); 61 FR
25566, 25568 (May 22, 1996) (after review by CASAC, the EPA issued a
post-proposal supplement to the 1982 Air Quality Criteria Document to
address certain new health studies not included in the 1982 Air Quality
Criteria Document or 1986

[[Page 3096]]

Addendum). The EPA reaffirmed this approach in its decision not to
revise the ozone NAAQS in 1993, as well as in its final decision on the
PM NAAQS in the 1997 and 2006 reviews. 58 FR 13008, 13013 to 13014
(March 9, 1993) (ozone review); 62 FR 38652, 38662 (July 18, 1997) and
71 FR 61141, 61148 to 61149 (October 17, 2006) (PM reviews) (The EPA
conducted a provisional assessment but based the final PM decisions on
studies and related information included in the air quality criteria
that had been reviewed by CASAC).
As discussed in the EPA's 1993 decision not to revise the NAAQS for
ozone, `new' studies may sometimes be of such significance that it is
appropriate to delay a decision on revision of NAAQS and to supplement
the pertinent air quality criteria so the ``new'' studies can be taken
into account (58 FR, 13013 to 13014, March 9, 1993). In this
proceeding, the provisional assessment of recent studies concludes
that, taken in context, the ``new'' information and findings do not
materially change any of the broad scientific conclusions regarding the
health effects of PM exposure made in the Integrated Science Assessment
(U.S. EPA, 2012b). For this reason, reopening the air quality criteria
review would not be warranted even if there were time to do so under
the court order governing the schedule for this rulemaking.
Accordingly, the EPA is basing the final decisions in this review on
the studies and related information included in the PM air quality
criteria that have undergone CASAC and public review. The EPA will
consider the ``new'' published studies for purposes of decision making
in the next periodic review of the PM NAAQS, which will provide the
opportunity to fully assess them through a more rigorous review process
involving the EPA, CASAC, and the public.

C. Related Control Programs To Implement PM Standards

States are primarily responsible for ensuring attainment and
maintenance of NAAQS once the EPA has established them. Under section
110 of the CAA and related provisions, states are to submit, for the
EPA's approval, SIPs that provide for the attainment and maintenance of
such standards through control programs directed to sources of the
pollutants involved. The states, in conjunction with the EPA, also
administer the PSD permitting program (CAA sections 160 to 169). In
addition, federal programs provide for nationwide reductions in
emissions of PM and other air pollutants through the federal motor
vehicle and motor vehicle fuel control program under title II of the
Act (CAA sections 202 to 250) which involves controls for emissions
from mobile sources and controls for the fuels used by these sources,
and new source performance standards (NSPS) for stationary sources
under section 111 of the CAA.
Currently, there are 35 areas in the U.S. that are designated as
nonattainment for the current annual PM2.5 standard and 32
areas in the U.S. that are designated as nonattainment for the current
24-hour PM2.5 standards. With the revisions to the PM NAAQS
that are being finalized in this rule, the EPA will work with the
states to conduct a new area designation process. Those states with new
nonattainment areas will be required to develop SIPs to attain the
standards. In developing their attainment plans, states will have to
take into account projected emission reductions from federal and state
rules that have already been adopted at the time of plan submittal. A
number of significant emission reduction programs that will lead to
reductions of PM and its precursors are in place today or are expected
to be in place by the time any new SIPs will be due. Examples of such
rules include regulations for onroad and nonroad engines and fuels, the
utility and industrial boilers toxics rules, and various other programs
already adopted by states to reduce emissions from key emissions
sources. States will then evaluate the level of additional emission
reductions needed for each nonattainment area to attain the standards
``as expeditiously as practicable'' and adopt new state regulations, as
appropriate. Section IX includes additional discussion of designation
and implementation issues associated with the revised PM NAAQS.

D. Summary of Proposed Revisions to the PM NAAQS

For reasons discussed in the proposal, the Administrator proposed
to revise the current primary and secondary PM standards. With regard
to the primary PM2.5 standards, the Administrator proposed
to revise the level of the annual PM2.5 standard from 15.0
[mu]g/m\3\ to a level within a range of 12.0 to 13.0 [mu]g/m\3\ and to
retain the level of the 24-hour PM2.5 standard at 35
[micro]g/m\3\. The Administrator also proposed to eliminate spatial
averaging provisions as part of the form of the annual standard to
avoid potential disproportionate impacts on at-risk populations. The
EPA proposed to revise the AQI for PM2.5, consistent with
the proposed primary PM2.5 standards.
With regard to the primary coarse particle standard, the EPA
proposed to retain the current 24-hour PM10 standard to
continue to provide protection against effects associated with short-
term exposure to thoracic coarse particles (i.e., PM10-2.5).
With regard to the secondary PM standards, the EPA proposed to
revise the suite of secondary PM standards by adding a distinct
standard for PM2.5 to address PM-related visibility
impairment. The separate secondary standard was proposed to be defined
in terms of a PM2.5 visibility index, which would use
speciated PM2.5 mass concentrations and relative humidity
data to calculate PM2.5 light extinction, translated to the
deciview (dv) scale, similar to the Regional Haze Program; a 24-hour
averaging time; a 90th percentile form averaged over 3 years; and a
level set at one of two options--either 30 or 28 dv. The EPA also
proposed to retain the current secondary standards generally to address
non-visibility welfare effects.
The EPA also proposed to revise the data handling procedures
consistent with the revised primary and secondary standards for
PM2.5 including the computations necessary for determining
when these standards are met and the measurement data that are
appropriate for comparison to the standards. With regard to monitoring-
related activities, the EPA proposed to update several aspects of the
monitoring regulations and specifically to require that a small number
of PM2.5 monitors be relocated to be collocated with
measurements of other pollutants (e.g., nitrogen dioxide, carbon
monoxide) in the near-road environment.

E. Organization and Approach to Final PM NAAQS Decisions

This action presents the Administrator's final decisions on the
review of the current primary and secondary PM2.5 and
PM10 standards. Consistent with the decisions made by the
EPA in the last review and with the conclusions in the Integrated
Science Assessment and Policy Assessment, fine and thoracic coarse
particles continue to be considered as separate subclasses of PM
pollution. Primary standards for fine particles and for thoracic coarse
particles are addressed in sections III and IV, respectively. Changes
to the AQI for PM2.5, consistent with the revised primary
PM2.5 standards, are addressed in section V. Secondary
standards for fine and coarse particles are addressed in section VI.
Related data handling conventions and exceptional events are addressed
in section VII. Updates to the monitoring regulations are addressed in

[[Page 3097]]

section VIII. Implementation activities, including PSD-related actions,
are addressed in section IX. Section X addresses applicable statutory
and executive order reviews.
Today's final decisions addressing standards for fine and coarse
particles are based on a thorough review in the Integrated Science
Assessment of scientific information on known and potential human
health and welfare effects associated with exposure to these subclasses
of PM at levels typically found in the ambient air. These final
decisions also take into account: (1) Staff assessments in the Policy
Assessment of the most policy-relevant information in the Integrated
Science Assessment as well as a quantitative health risk assessment and
urban-focused visibility assessment based on that information; (2)
CASAC advice and recommendations, as reflected in its letters to the
Administrator, its discussions of drafts of the Integrated Science
Assessment, Risk and Exposure Assessments, and Policy Assessment at
public meetings, and separate written comments prepared by individual
members of the CASAC PM Review Panel; (3) public comments received
during the development of these documents, both in connection with
CASAC meetings and separately; and (4) extensive public comments
received on the proposed rulemaking.

III. Rationale for Final Decisions on the Primary PM2.5
Standards

This section presents the Administrator's final decision regarding
the need to revise the current primary PM2.5 standards and,
more specifically, regarding revisions to the level and form of the
existing primary annual PM2.5 standard in conjunction with
retaining the existing primary 24-hour PM2.5 standard. As
discussed more fully below, the rationale for the final decision is
based on a thorough review, in the Integrated Science Assessment, of
the latest scientific information, published through mid-2009, on human
health effects associated with long- and short-term exposures to fine
particles in the ambient air. The final decisions also take into
account: (1) Staff assessments of the most policy-relevant information
presented and assessed in the Integrated Science Assessment and staff
analyses of air quality and human risks presented in the Risk
Assessment and the Policy Assessment, upon which staff conclusions
regarding appropriate considerations in this review are based; (2)
CASAC advice and recommendations, as reflected in discussions of drafts
of the Integrated Science Assessment, Risk Assessment, and Policy
Assessment at public meetings, in separate written comments, and in
CASAC's letters to the Administrator; (3) the multiple rounds of public
comments received during the development of these documents, both in
connection with CASAC meetings and separately; and (4) extensive public
comments received on the proposal.
In developing this final rule, the Administrator recognizes that
the CAA requires her to reach a public health policy judgment as to
what standards would be requisite--neither more nor less stringent than
necessary--to protect public health with an adequate margin of safety,
based on scientific evidence and technical assessments that have
inherent uncertainties and limitations. This judgment requires making
reasoned decisions as to what weight to place on various types of
evidence and assessments, and on the related uncertainties and
limitations. Thus, in selecting the final standards, the Administrator
is seeking not only to prevent fine particle concentrations that have
been demonstrated to be harmful but also to prevent lower fine particle
concentrations that may pose an unacceptable risk of harm, even if the
risk is not precisely identified as to nature or degree.
As discussed below, as well as in more detail in the proposal, a
substantial amount of new research has been conducted since the close
of the science assessment in the last review of the PM2.5
NAAQS (U.S. EPA, 2004), with important new information coming from
epidemiological studies, in particular. This body of evidence includes
hundreds of new epidemiological studies conducted in many countries
around the world. In its assessment of the evidence judged to be most
relevant to making decisions on elements of the primary
PM2.5 standards, the EPA has placed greater weight on U.S.
and Canadian studies using PM2.5 measurements, since studies
conducted in other countries may reflect different demographic and air
pollution characteristics.\14\
---------------------------------------------------------------------------

\14\ Nonetheless, the Administrator recognizes the importance of
all studies, including international studies, in the Integrated
Science Assessment's considerations of the weight of the evidence
that informs causality determinations.
---------------------------------------------------------------------------

The newly available research studies as well as the earlier body of
scientific evidence presented and assessed in the Integrated Science
Assessment have undergone intensive scrutiny through multiple layers of
peer review and opportunities for public review and comment. In
developing this final rule, the EPA has drawn upon an integrative
synthesis of the entire body of evidence concerning exposure to ambient
fine particles and a broad range of health endpoints (U.S. EPA, 2009a,
Chapters 2, 4, 5, 6, 7, and 8) focusing on those health endpoints for
which the Integrated Science Assessment concludes that there is a
causal or likely causal relationship with long- or short-term
PM2.5 exposures. The EPA has also considered health
endpoints for which the Integrated Science Assessment concludes there
is evidence suggestive of a causal relationship with long-term
PM2.5 exposures.
The EPA has also drawn upon a quantitative risk assessment based
upon the scientific evidence described and assessed in the Integrated
Science Assessment. These analyses, discussed in the Risk Assessment
(U.S. EPA, 2010a) and Policy Assessment (U.S. EPA, 2011a, chapter 2),
have also undergone intensive scrutiny through multiple layers of peer
review and multiple opportunities for public review and comment.
Although important uncertainties remain in the qualitative and
quantitative characterizations of health effects attributable to
ambient fine particles, progress has been made in addressing these
uncertainties in this review. The EPA's review of this information has
been extensive and deliberate. This intensive evaluation of the
scientific evidence and quantitative assessments has provided a
comprehensive and adequate basis for regulatory decision making at this
time.
This section describes the integrative synthesis of the evidence
and technical information contained in the Integrated Science
Assessment, the Risk Assessment, and the Policy Assessment with regard
to the current and alternative standards. The EPA notes that the final
decision for retaining or revising the current primary PM2.5
standards is a public health policy judgment made by the Administrator.
The Administrator's final decision draws upon scientific information
and analyses related to health effects and risks; judgments about
uncertainties that are inherent in the scientific evidence and
analyses; CASAC advice; and comments received in response to the
proposal.
In presenting the rationale for the final decisions on the primary
PM2.5 standards, this section begins with a summary of the
approaches used in setting the initial primary PM2.5 NAAQS
in 1997 and in reviewing and revising those standards in 2006 (section
III.A.1). The DC Circuit Court of Appeals remand of the primary annual
PM2.5 standard in 2009 is discussed in section III.A.2.
Taking into consideration this

[[Page 3098]]

history, section III.A.3 describes the EPA's general approach used in
the current review for considering the need to retain or revise the
current suite of fine particle standards, taking into account public
comment on the proposed approach.
The scientific evidence and quantitative risk assessment were
presented in sections III.B and III.C of the proposal, respectively (77
FR 38906 to 38917, June 29, 2012) and are outlined in sections III.B
and III.C below. Subsequent sections of this preamble provide a more
complete discussion of the Administrator's rationale, in light of key
issues raised in public comments, for concluding that it is appropriate
to revise the suite of current primary PM2.5 standards
(section III.D), as well as a more complete discussion of the
Administrator's rationale for retaining or revising the specific
elements of the primary PM2.5 standards, namely the
indicator (section III.E.1); averaging time (section III.E.2); form
(section III.E.3); and level (section III.E.4). A summary of the final
decisions to revise the suite of primary PM2.5 standards is
presented in section III.F.

A. Background

There are currently two primary PM2.5 standards
providing public health protection from effects associated with fine
particle exposures. The annual standard is set at a level of 15.0
[mu]g/m\3\, based on the 3-year average of annual arithmetic mean
PM2.5 concentrations from single or multiple monitors sited
to represent community-wide air quality. The 24-hour standard is set at
a level of 35 [mu]g/m\3\, based on the 3-year average of the 98th
percentile of 24-hour PM2.5 concentrations at each
population-oriented monitor within an area.
The past and current approaches for reviewing the primary
PM2.5 standards described below are all based most
fundamentally on using information from epidemiological studies to
inform the selection of PM2.5 standards that, in the
Administrator's judgment, protect public health with an adequate margin
of safety. Such information can be in the form of air quality
distributions over which health effect associations have been observed
in scientific studies or in the form of concentration-response
functions that support quantitative risk assessment. However, evidence-
and risk-based approaches using information from epidemiological
studies to inform decisions on PM2.5 standards are
complicated by the recognition that no population threshold, below
which it can be concluded with confidence that PM2.5-related
effects do not occur, can be discerned from the available evidence.\15\
As a result, any general approach to reaching decisions on what
standards are appropriate necessarily requires judgments about how to
translate the information available from the epidemiological studies
into a basis for appropriate standards. This includes consideration of
how to weigh the uncertainties in the reported associations across the
distributions of PM2.5 concentrations in the studies and the
uncertainties in quantitative estimates of risk, in the context of the
entire body of evidence before the Agency. Such approaches are
consistent with setting standards that are neither more nor less
stringent than necessary, recognizing that a zero-risk standard is not
required by the CAA.
---------------------------------------------------------------------------

\15\ The term ``evidence-based'' approach or consideration
generally refers to using the information in the scientific evidence
to inform judgments on the need to retain or revise the NAAQS. The
term ``risk-based'' generally refers to using the quantitative
information in the Risk Assessment to inform such judgments.
---------------------------------------------------------------------------

1. General Approach Used in Previous Reviews
The general approach used to translate scientific information into
standards in the previous PM NAAQS reviews focused on consideration of
alternative standard levels that were somewhat below the long-term mean
PM2.5 concentrations reported in key epidemiological studies
(U.S. EPA, 2011a, section 2.1.1). This approach recognized that the
strongest evidence of PM2.5-related associations occurs
where the bulk of the data exists, which is over a range of
concentrations around the long-term (i.e., annual) mean.
In setting primary PM2.5 annual and 24-hour standards
for the first time in 1997, the Agency relied primarily on an evidence-
based approach that focused on epidemiological evidence, especially
from short-term exposure studies of fine particles judged to be the
strongest evidence at that time (U.S. EPA, 2011a, section 2.1.1.1). The
EPA selected a level for the annual standard that was at or below the
long-term mean PM2.5 concentrations in studies providing
evidence of associations with short-term PM2.5 exposures,
placing greatest weight on those short-term exposure studies that
reported clearly statistically significant associations with mortality
and morbidity effects. Further consideration of long-term mean
PM2.5 concentrations associated with mortality and
respiratory effects in children did not provide a basis for
establishing a lower annual standard level. The EPA did not place much
weight on quantitative risk estimates from the very limited risk
assessment conducted, but did conclude that the risk assessment results
confirmed the general conclusions drawn from the epidemiological
evidence that a serious public health problem was associated with
ambient PM levels allowed under the then current PM10
standards (62 FR 38665/1, July 18, 1997).
The EPA considered the epidemiological evidence and data on air
quality relationships to set an annual PM2.5 standard that
was intended to be the ``generally controlling'' standard; i.e., the
primary means of lowering both long- and short-term ambient
concentrations of PM2.5.\16\ In conjunction with the annual
standard, the EPA also established a 24-hour PM2.5 standard
to provide supplemental protection against days with high peak
concentrations, localized ``hotspots,'' and risks arising from seasonal
emissions that might not be well controlled by an annual standard (62
FR 38669/3).
---------------------------------------------------------------------------

\16\ In so doing, the EPA noted that because an annual standard
would focus control programs on annual average PM2.5
concentrations, it would not only control long-term exposure levels,
but would also generally control the overall distribution of 24-hour
exposure levels, resulting in fewer and lower 24-hour peak
concentrations. Alternatively, a 24-hour standard that focused
controls on peak concentrations could also result in lower annual
average concentrations. Thus, the EPA recognized that either
standard could provide some degree of protection from both short-
and long-term exposures, with the other standard serving to address
situations where the daily peaks and annual averages are not
consistently correlated (62 FR 38669, July 18, 1997). In the
circumstances presented in that review, the EPA determined that it
was appropriate to focus on the annual standard as the standard best
suited to control both annual and daily air quality distributions
(62 FR 38670).
---------------------------------------------------------------------------

In 2006, the EPA used a different evidence-based approach to assess
the appropriateness of the levels of the 24-hour and annual
PM2.5 standards (U.S. EPA, 2011a, section 2.1.1.2). Based on
an expanded body of epidemiological evidence that was stronger and more
robust than that available in the 1997 review, including additional
studies of both short- and long-term exposures, the EPA decided that
using evidence of effects associated with periods of exposure that were
most closely matched to the averaging time of each standard was the
most appropriate public health policy approach for evaluating the
scientific evidence to inform selecting the level of each standard.
Thus, the EPA relied upon evidence from the short-term exposure studies
as the principal basis for revising the level of the 24-hour
PM2.5 standard from 65 to 35 [micro]g/m\3\ to protect
against effects associated with short-term exposures. The EPA relied
upon evidence from long-term exposure

[[Page 3099]]

studies as the principal basis for retaining the level of the annual
PM2.5 standard at 15 [micro]g/m\3\ to protect against
effects associated with long-term exposures. This approach essentially
took the view that short-term studies were not appropriate to inform
decisions relating to the level of the annual standard, and long-term
studies were not appropriate to inform decisions relating to the level
of the 24-hour standard. With respect to quantitative risk-based
considerations, the EPA determined that the estimates of risks likely
to remain upon attainment of the 1997 suite of PM2.5
standards were indicative of risks that could be reasonably judged
important from a public health perspective and, thus, supported
revision of the standards. However, the EPA judged that the
quantitative risk assessment had important limitations and did not
provide an appropriate basis for selecting the levels of the revised
standards in 2006 (71 FR 61174/1-2, October 17, 2006).
2. Remand of Primary Annual PM2.5 Standard
As noted above in section II.B.2, several parties filed petitions
for review in the U.S. Court of Appeals for the District of Columbia
Circuit following promulgation of the revised PM NAAQS in 2006. These
petitions challenged several aspects of the final rule including the
level of the primary PM2.5 annual standard. The primary 24-
hour PM2.5 standard was not challenged by any of the
litigants and, thus, was not considered in the court's review and
decision.
On judicial review, the D.C. Circuit remanded the primary annual
PM2.5 NAAQS to the EPA on grounds that the Agency failed to
adequately explain why the annual standard provided the requisite
protection from both short- and long-term exposures to fine particles
including protection for at-risk populations. American Farm Bureau
Federation v. EPA, 559 F. 3d 512 (D.C. Cir. 2009). With respect to
human health protection from short-term PM2.5 exposures, the
court considered the different approaches used by the EPA in the 1997
and 2006 p.m. NAAQS decisions, as summarized in section III.A.1 above.
The court found that the EPA failed to adequately explain why a primary
24-hour PM2.5 standard by itself would provide the
protection needed from short-term exposures and remanded the primary
annual PM2.5 standard to the EPA ``for further consideration
of whether it is set at a level requisite to protect the public health
while providing an adequate margin of safety from the risk of short-
term exposures to PM2.5.'' American Farm Bureau Federation
v. EPA, 559 F. 3d at 520-24.
With respect to protection from long-term exposure to fine
particles, the court found that the EPA failed to adequately explain
how the primary annual PM2.5 standard provided an adequate
margin of safety for children and other at-risk populations. The court
found that the EPA did not provide a reasonable explanation of why
certain morbidity studies, including a study of children in Southern
California showing lung damage associated with long-term
PM2.5 exposure (Gauderman et al., 2000) and a multi-city
study (24-Cities Study) evaluating decreased lung function in children
associated with long-term PM2.5 exposures (Raizenne et al.,
1996), did not warrant a more stringent annual PM2.5
standard. Id. at 522-23. Specifically, the court found that:

EPA was unreasonably confident that, even though it relied
solely upon long-term mortality studies, the revised standard would
provide an adequate margin of safety with respect to morbidity among
children. Notably absent from the final rule, moreover, is any
indication of how the standard will adequately reduce risk to the
elderly or to those with certain heart or lung diseases despite (a)
the EPA's determination in its proposed rule that those
subpopulations are at greater risk from exposure to fine particles
and (b) the evidence in the record supporting that determination.
Id. at 525.

In addition, the court held that the EPA had not adequately
explained its decision to base the level of the annual standard
essentially exclusively on the results of long-term studies and the 24-
hour standard level essentially exclusively on the results of short-
term studies. See 559 F. 3d at 522 (``[e]ven if the long-term studies
available today are useful for setting an annual standard * * * it is
not clear why the EPA no longer believes it useful to look as well to
short-term studies in order to design the suite of standards that will
most effectively reduce the risks associated with short-term
exposure''); see also Id. at 523-24 (holding that the EPA had not
adequately explained why a standard based on levels in short-term
exposure studies alone provided appropriate protection from health
effects associated with short-term PM2.5 exposures given the
stated need to lower the entire air quality distribution, and not just
peak concentrations, in order to control against short-term effects).
In remanding the primary annual PM2.5 standard for
reconsideration, the court did not vacate the standard, Id. at 530, so
the standard remains in effect and is therefore the standard considered
by the EPA in this review.
3. General Approach Used in the Policy Assessment for the Current
Review
This review is based on an assessment of a much expanded body of
scientific evidence, more extensive air quality data and analyses, and
a more comprehensive quantitative risk assessment relative to the
information available in past reviews, as presented and assessed in the
Integrated Science Assessment and Risk Assessment and discussed in the
Policy Assessment. As a result, the EPA's general approach to reaching
conclusions about the adequacy of the current suite of PM2.5
standards and potential alternative standards that are appropriate to
consider was broader and more integrative than in past reviews. Our
general approach also reflected consideration of the issues raised by
the court in its remand of the primary annual PM2.5 standard
as discussed in section III.A.2 above, since decisions made in this
review, and the rationales for those decisions, will comprise the
Agency's response to the remand.
The EPA's general approach took into account both evidence-based
and risk-based considerations and the uncertainties related to both
types of information, as well as advice from CASAC (Samet, 2010c,d) and
public comments on the first and second draft Policy Assessments (U.S.
EPA, 2010c,f). In so doing, the EPA staff developed a final Policy
Assessment (U.S. EPA, 2011a) which provided as broad an array of policy
options as was supported by the available information, recognizing that
the selection of a specific approach to reaching final decisions on the
primary PM2.5 standards will reflect the judgments of the
Administrator as to what weight to place on the various approaches and
types of information available in the current review.
The Policy Assessment concluded it was most appropriate to consider
the protection against PM2.5-related mortality and morbidity
effects, associated with both long- and short-term exposures, afforded
by the annual and 24-hour standards taken together, as was done in the
1997 review, rather than to consider each standard separately, as was
done in the 2006 review (U.S. EPA, 2011a, section 2.1.3).\17\ As the
EPA recognized in 1997,

[[Page 3100]]

there are various ways to combine two standards to achieve an
appropriate degree of public health protection. The extent to which
these two standards are interrelated in any given area depends in large
part on the relative levels of the standards, the peak-to-mean ratios
that characterize air quality patterns in an area, and whether changes
in air quality designed to meet a given suite of standards are likely
to be of a more regional or more localized nature.
---------------------------------------------------------------------------

\17\ By utilizing this approach, the Agency also is responsive
to the remand of the 2006 standard. As noted in section III.A.2, the
D.C. Circuit, in remanding the 2006 primary annual PM2.5
standard, concluded that the Administrator had failed to adequately
explain why an annual standard was sufficiently protective in the
absence of consideration of the long-term mean PM2.5
concentrations in short-term exposure studies as well, and likewise
had failed to explain why a 24-hour standard was sufficiently
protective in the absence of consideration of the effect of an
annual standard on reducing the overall distribution of 24-hour
average PM2.5 concentrations. 559 F. 3d at 520-24.
---------------------------------------------------------------------------

In considering the combined effect of annual and 24-hour standards,
the Policy Assessment recognized that changes in PM2.5 air
quality designed to meet an annual standard would likely result not
only in lower annual average PM2.5 concentrations but also
in fewer and lower peak 24-hour PM2.5 concentrations. The
Policy Assessment also recognized that changes designed to meet a 24-
hour standard would result not only in fewer and lower peak 24-hour
PM2.5 concentrations but also in lower annual average
PM2.5 concentrations. Thus, either standard could be viewed
as providing protection from effects associated with both short- and
long-term exposures, with the other standard serving to address
situations where the daily peak and annual average concentrations are
not consistently correlated.
In considering the currently available evidence, the Policy
Assessment recognized that the short-term exposure studies were
primarily drawn from epidemiological studies that associated variations
in area-wide health effects with monitor(s) that measured the variation
in daily PM2.5 concentrations over the course of several
years. The strength of the associations in these data was demonstrably
in the numerous ``typical'' days within the air quality distribution,
not in the peak days. See also 71 FR 61168, October 17, 2006 and
American Farm Bureau Federation v. EPA, 559 F. 3d at 523, 524 (making
the same point). The quantitative risk assessments conducted for this
and previous reviews demonstrated the same point; that is, much, if not
most of the aggregate risk associated with short-term exposures results
from the large number of days during which the 24-hour average
concentrations are in the low-to mid-range, below the peak 24-hour
concentrations (U.S. EPA, 2011a, section 2.2.2; U.S. EPA, 2010a,
section 3.1.2.2). In addition, there was no evidence suggesting that
risks associated with long-term exposures were likely to be
disproportionately driven by peak 24-hour concentrations.\18\
---------------------------------------------------------------------------

\18\ In confirmation, a number of studies have presented
analyses excluding higher PM concentration days and reported a
limited effect on the magnitude of the effect estimates or
statistical significance of the association (e.g., Dominici, 2006b;
Schwartz et al., 1996; Pope and Dockery, 1992).
---------------------------------------------------------------------------

For these reasons, the Policy Assessment concluded that strategies
that focused primarily on reducing peak days were less likely to
achieve reductions in the PM2.5 concentrations that were
most strongly associated with the observed health effects. Furthermore,
the Policy Assessment concluded that a policy approach that focused on
reducing peak exposures would most likely result in more uneven public
health protection across the U.S. by either providing inadequate
protection in some areas or overprotecting in other areas (U.S. EPA,
2011a, p. 2-9; U.S. EPA, 2010a, section 5.2.3). This is because, as
discussed above, reductions based on control of peak days are less
likely to control the bulk of the air quality distribution.
The Policy Assessment concluded that a policy goal of setting a
``generally controlling'' annual standard that will lower a wide range
of ambient 24-hour PM2.5 concentrations, as opposed to
focusing on control of peak 24-hour PM2.5 concentrations,
was the most effective and efficient way to reduce total population
risk and so provide appropriate protection. This approach, in contrast
to one focusing on a generally controlling 24-hour standard, would
likely reduce aggregate risks associated with both long- and short-term
exposures with more consistency and would likely avoid setting national
standards that could result in relatively uneven protection across the
country, due to setting standards that are either more or less
stringent than necessary in different geographical areas (U.S. EPA,
2011a, p. 2-9).
The Policy Assessment also concluded that an annual standard
intended to serve as the primary means for providing protection from
effects associated with both long- and short-term PM2.5
exposures cannot be expected to offer sufficient protection against the
effects of all short-term PM2.5 exposures. As a result, in
conjunction with a generally controlling annual standard, the Policy
Assessment concluded it was appropriate to consider setting a 24-hour
standard to provide supplemental protection, particularly for areas
with high peak-to-mean ratios possibly associated with strong local or
seasonal sources, or PM2.5-related effects that may be
associated with shorter-than-daily exposure periods (U.S. EPA, 2011a,
p. 2-10).
The Policy Assessment's consideration of the protection afforded by
the current and alternative suites of standards focused on
PM2.5-related health effects associated with long-term
exposures for which the magnitude of quantitative estimates of risks to
public health generated in the risk assessment was appreciably larger
in terms of overall incidence and percent of total mortality or
morbidity effects than for short-term PM2.5-related effects.
Nonetheless, the EPA also considered health effects and estimated risks
associated with short-term exposures. In both cases, the Policy
Assessment placed greatest weight on health effects that had been
judged in the Integrated Science Assessment to have a causal or likely
causal relationship with PM2.5 exposures, while also
considering health effects judged to be suggestive of a causal
relationship or evidence that focused on specific at-risk populations.
The Policy Assessment placed relatively greater weight on statistically
significant associations that yielded relatively more precise effect
estimates and that were judged to be robust to confounding by other air
pollutants. In the case of short-term exposure studies, the Policy
Assessment placed greatest weight on evidence from large multi-city
studies, while also considering associations in single-city studies.
In translating information from epidemiological studies into the
basis for reaching staff conclusions on the adequacy of the current
suite of standards, the Policy Assessment considered a number of
factors. As an initial matter, the Policy Assessment considered the
extent to which the currently available evidence and related
uncertainties strengthens or calls into question conclusions from the
last review regarding associations between fine particle exposures and
health effects. The Policy Assessment also considered evidence of
health effects in at-risk populations and the potential impacts on such
populations. Further, the Policy Assessment explored the extent to
which PM2.5-related health effects had been observed in
areas where air quality distributions extend to lower concentrations
than previously reported or in areas that would likely have met the
current suite of standards.
In translating information from epidemiological studies into the
basis for reaching staff conclusions on

[[Page 3101]]

standard levels for consideration (U.S. EPA, 2011a, sections 2.1.3 and
2.3.4), the Policy Assessment first recognized the absence of
discernible thresholds in the concentration-response functions from
long- and short-term PM2.5 exposure studies (U.S. EPA,
2011a, section 2.4.3).\19\ In the absence of any discernible
thresholds, the Agency's general approach for identifying appropriate
standard levels for consideration involved characterizing the range of
PM2.5 concentrations over which we have the most confidence
in the associations reported in epidemiological studies. In so doing,
the Policy Assessment recognized that there is no single factor or
criterion that comprises the ``correct'' approach, but rather there are
various approaches that are reasonable to consider for characterizing
the confidence in the associations and the limitations and
uncertainties in the evidence. Identifying the implications of various
approaches for reaching conclusions on the range of alternative
standard levels that is appropriate to consider can help inform the
final decisions to either retain or revise the standards. Today's final
decisions also take into account public health policy judgments as to
the degree of health protection that is to be achieved.
---------------------------------------------------------------------------

\19\ The epidemiological studies evaluated in the Integrated
Science Assessment that examined the shape of concentration-response
relationships and the potential presence of a threshold focused on
cardiovascular-related hospital admissions and emergency department
visits associated with short-term PM10 exposures and
premature mortality associated with long-term PM2.5
exposure (U.S. EPA, 2009a, sections 6.5, 6.2.10.10 and 7.6).
Overall, the Integrated Science Assessment concluded that the
studies evaluated support the use of a no-threshold, log-linear
model but recognized that ``additional issues such as the influence
of heterogeneity in estimates between cities, and the effect of
seasonal and regional differences in PM on the concentration-
response relationship still require further investigation'' (U.S.
EPA, 2009a, section 2.4.3).
---------------------------------------------------------------------------

In reaching staff conclusions on the range of annual standard
levels that was appropriate to consider, the Policy Assessment focused
on identifying an annual standard that provided requisite protection
from effects associated with both long- and short-term exposures. In so
doing, the Policy Assessment explored different approaches for
characterizing the range of PM2.5 concentrations over which
our confidence in the nature of the associations for both long- and
short-term exposures is greatest, as well as the extent to which our
confidence is reduced at lower PM2.5 concentrations.
First, the Policy Assessment recognized that the approach that most
directly addressed this issue considered studies that analyzed
confidence intervals around concentration-response relationships and in
particular, analyses that averaged across multiple concentration-
response models rather than considering a single concentration-response
model.\20\ The Policy Assessment explored the extent to which such
analyses had been published for studies of health effects associated
with long- or short-term PM2.5 exposures. Such analyses
could potentially be used to characterize a concentration below which
uncertainty in a concentration-response relationship substantially
increases or is judged to be indicative of an unacceptable degree of
uncertainty about the existence of a continuing concentration-response
relationship. The Policy Assessment concluded that identifying this
area of uncertainty in the concentration-response relationship could be
used to inform identification of alternative standard levels that are
appropriate to consider.
---------------------------------------------------------------------------

\20\ This is distinct from confidence intervals around
concentration-response relationships that are related to the
magnitude of effect estimates generated at specific PM2.5
concentrations (i.e., point-wise confidence intervals) and that are
relevant to the precision of the effect estimate across the air
quality distribution, rather than to our confidence in the existence
of a continuing concentration-response relationship across the
entire air quality distribution on which a reported association was
based.
---------------------------------------------------------------------------

Further, the Policy Assessment explored other approaches that
considered different statistical metrics from epidemiological studies.
The Policy Assessment first took into account the general approach used
in previous PM reviews which focused on consideration of alternative
standard levels that were somewhat below the long-term mean
PM2.5 concentrations reported in epidemiological studies
using air quality distributions based on composite monitor
concentrations.\21\ This approach recognized that the strongest
evidence of PM2.5-related associations occurs at
concentrations around the long-term (i.e., annual) mean. In using this
approach, the Policy Assessment placed greatest weight on those long-
and short-term exposure studies that reported statistically significant
associations with mortality and morbidity effects.
---------------------------------------------------------------------------

\21\ Using the term ``composite monitor'' does not imply that
the EPA can identify one monitor that represents the air quality
evaluated in a specific study area. Rather, the composite monitor
concentration represents the average concentration across monitors
within each area with more than one monitor included in a given
study as typically reported in epidemiological studies. For multi-
city studies, this metric reflects concentrations averaged across
multiple monitors or from single monitors within each area and then
averaged across study areas for an overall study mean
PM2.5 concentration. This is consistent with the
epidemiological evidence considered in other NAAQS reviews.
---------------------------------------------------------------------------

In extending this approach, the Policy Assessment also considered
information beyond a single statistical metric of PM2.5
concentrations (i.e., the mean) to the extent such information was
available. Pursuant to an express comment from CASAC (Samet 2010d, p.
2), the Policy Assessment utilized distributional statistics (i.e.,
statistical characterization of an entire distribution of data) to
identify the broader range of PM2.5 concentrations that had
the most influence on the calculation of relative risk estimates in
both long- and short-term exposure epidemiological studies. Thus, the
Policy Assessment considered the part of the distribution of
PM2.5 concentrations in which the data analyzed in the study
(i.e., air quality and population-level data, as discussed below) were
most concentrated, specifically, the range of PM2.5
concentrations around the long-term mean over which our confidence in
the magnitude and significance of associations observed in the
epidemiological studies was greatest. The Policy Assessment then
focused on the lower part of the distribution to characterize where the
data became appreciably more sparse and, thus, where our understanding
of the magnitude and significance of the associations correspondingly
became more uncertain. The Policy Assessment recognized there was no
single percentile value within a given distribution that was most
appropriate or ``correct'' to use to characterize where our confidence
in the associations becomes appreciably lower. The Policy Assessment
concluded that the range from the 25th to 10th percentiles is a
reasonable range to consider as a region where we had appreciably less
confidence in the associations observed in epidemiological studies.\22\
---------------------------------------------------------------------------

\22\ In the PM NAAQS review completed in 2006, the Staff Paper
similarly recognized that the evidence of an association in any
epidemiological study is ``strongest at and around the long-term
average where the data in the study are most concentrated. For
example, the interquartile range of long-term average concentrations
within a study [with a lower bound of the 25th percentile] or a
range within one standard deviation around the study mean, may
reasonably be used to characterize the range over which the evidence
of association is strongest'' (U.S. EPA, 2005, p. 5-22). A range of
one standard deviation around the mean represents approximately 68
percent of normally distributed data, and below the mean falls
between the 25th and 10th percentiles.
---------------------------------------------------------------------------

In considering distributional statistics from epidemiological
studies, the final Policy Assessment focused on two types of
population-level metrics that CASAC advised were most useful to
consider in identifying the PM2.5 concentrations

[[Page 3102]]

most influential in generating the health effect estimates reported in
the epidemiological studies.\23\ Consistent with CASAC advice, the most
relevant information was the distribution of health events (e.g.,
deaths, hospitalizations) occurring within a study population in
relation to the distribution of PM2.5 concentrations.
However, in recognizing that access to health event data can be
restricted, the Policy Assessment also considered the number of study
participants within each study area as an appropriate surrogate for
health event data.
---------------------------------------------------------------------------

\23\ The second draft Policy Assessment focused on the
distributions of ambient PM2.5 concentrations and
associated population data across areas included in several multi-
city studies for which such data were available in seeking to
identify the most influential range of concentrations (U.S. EPA,
2010f, section 2.3.4.1). In its review of the second draft Policy
Assessment, CASAC advised that it ``would be preferable to have
information on the concentrations that were most influential in
generating the health effect estimates in individual studies''
(Samet, 2010d, p.2). Therefore, in the final Policy Assessment, the
EPA considered population-level data (i.e., area-specific health
event data and study area population data) along with corresponding
PM2.5 concentrations to generate a cumulative
distribution of the population-level data relative to long-term mean
PM2.5 concentrations to determine the most influential
part of the air quality distribution (U.S. EPA, 2011a, Figure 2-7
and associated text).
---------------------------------------------------------------------------

The Policy Assessment recognized that an approach considering
analyses of confidence intervals around concentration-response
functions was intrinsically related to an approach that considered
different distributional statistics. Both of these approaches could be
employed to understand the broader distribution of PM2.5
concentrations which correspond to the health events reported in
epidemiological studies. In applying these approaches, the Policy
Assessment, consistent with CASAC advice (Samet, 2010d, p. 3),
considered PM2.5 concentrations from long- and short-term
PM2.5 exposure studies using composite monitor
distributions.
In reaching staff conclusions on alternative standard levels that
were appropriate to consider, the Policy Assessment also included a
broader consideration of the uncertainties and limitations of the
current scientific evidence. Most notably, these uncertainties are
related to the heterogeneity observed in the epidemiological studies in
the eastern versus western parts of the U.S., the relative toxicity of
PM2.5 components, and the potential role of co-pollutants
(U.S. EPA, 2011a, pp. 2-25 to 2-26). The limitations and uncertainties
associated with the currently available scientific evidence, including
the availability of fewer studies toward the lower range of alternative
annual standard levels being considered in this proposal, are
summarized in section III.B below and further discussed in section
III.B.2 of the proposal.
The Policy Assessment recognized that the level of protection
afforded by the NAAQS relies both on the level and the form of the
standard. The Policy Assessment concluded that a policy approach that
used data based on composite monitor distributions to identify
alternative standard levels, and then compared those levels to
concentrations at maximum monitors to determine whether an area meets a
given standard, inherently has the potential to build in some margin of
safety (U.S. EPA, 2011a, p. 2-14).\24\ This conclusion was consistent
with CASAC's comments on the second draft Policy Assessment, in which
CASAC expressed its preference for focusing on an approach using
composite monitor distributions ``because of its stability, and for the
additional margin of safety it provides'' when ``compared to the
maximum monitor perspective'' (Samet, et al., 2010d, pp. 2 to 3).
---------------------------------------------------------------------------

\24\ Statistical metrics (e.g., means) based on composite
monitor distributions may be identical to or below the same
statistical metrics based on maximum monitor distributions. For
example, some areas may have only one monitor, in which case the
composite and maximum monitor distributions will be identical in
those areas. Other areas may have multiple monitors that may be very
close to the monitor measuring the highest concentrations, in which
case the composite and maximum monitor distributions could be
similar in those areas. As noted in Hassett-Sipple et al. (2010),
for studies involving a large number of areas, the composite and
maximum concentrations are generally within 5 percent of each other
(77 FR 38905, fn. 30). Still other areas may have multiple monitors
that may be separately impacted by local sources in which case the
composite and maximum monitor distributions could be quite different
(U.S. EPA, 2011a, p. 2-14). See further discussion of this issue in
section III.E.4.c.i below.
---------------------------------------------------------------------------

In reaching staff conclusions on alternative 24-hour standard
levels that are appropriate to consider for setting a 24-hour standard
intended to supplement the protection afforded by a generally
controlling annual standard, the Policy Assessment considered currently
available short-term PM2.5 exposure studies. The evidence
from these studies informed our understanding of the protection
afforded by the suite of standards against effects associated with
short-term exposures. In considering the short-term exposure studies,
the Policy Assessment evaluated both the distributions of 24-hour
PM2.5 concentrations, with a focus on the 98th percentile
concentrations (to the extent such data were available) to match the
form of the current 24-hour PM2.5 standard, as well as the
long-term mean PM2.5 concentrations reported in these
studies. In addition to considering the epidemiological evidence, the
Policy Assessment also considered air quality information based on
county-level 24-hour and annual design values \25\ to understand the
policy implications of the alternative standard levels supported by the
underlying science. In particular, the Policy Assessment considered the
extent to which different combinations of alternative annual and 24-
hour standards would support the policy goal of focusing on a generally
controlling annual standard in conjunction with a 24-hour standard that
would provide supplemental protection. In so doing, the Policy
Assessment discussed the roles that each standard might be expected to
play in the protection afforded by alternative suites of standards.
---------------------------------------------------------------------------

\25\ Design values are the metrics (i.e., statistics) that are
compared to the NAAQS levels to determine compliance.
---------------------------------------------------------------------------

Beyond these evidence-based considerations, the Policy Assessment
also considered the quantitative risk estimates and the key
observations presented in the Risk Assessment. This assessment included
an evaluation of 15 urban case study areas and estimated risk
associated with a number of health endpoints associated with long- and
short-term PM2.5 exposures (U.S. EPA, 2010a). As part of the
risk-based considerations, the Policy Assessment considered estimates
of the magnitude of PM2.5-related risks associated with
recent air quality levels and air quality simulated to just meet the
current and alternative suites of standards using alternative
simulation approaches. The Policy Assessment also characterized the
risk reductions, relative to the risks remaining upon just meeting the
current standards, associated with just meeting alternative suites of
standards. In so doing, the Policy Assessment recognized the
uncertainties inherent in such risk estimates, and took such
uncertainties into account by considering the sensitivity of the
``core'' risk estimates to alternative assumptions and methods likely
to have substantial impact on the estimates. In addition, the Policy
Assessment considered additional analyses characterizing the
representativeness of the urban study areas within a broader national
context to understand the roles that the annual and 24-hour standards
may play in affording protection against effects related to both long-
and short-term PM2.5 exposures.
Based on the approach discussed above, the Policy Assessment
reached conclusions related to the primary PM2.5 standards
that reflected an

[[Page 3103]]

understanding of both evidence-based and risk-based considerations to
inform two overarching questions related to: (1) The adequacy of the
current suite of PM2.5 standards and (2) revisions to the
standards that were appropriate to consider in this review to protect
against health effects associated with both long- and short-term
exposures to fine particles. When evaluating the health protection
afforded by the current or any alternative suites of standards
considered, the Policy Assessment took into account the four basic
elements of the NAAQS: The indicator, averaging time, form, and level.
The general approach for reviewing the primary PM2.5
standards described above provided a comprehensive basis that helped to
inform the Administrator's judgments in reaching her proposed and final
decisions to revise the current suite of primary fine particle NAAQS
and in responding to the remand of the 2006 primary annual
PM2.5 standard.

B. Overview of Health Effects Evidence

This section outlines the key information presented in section
III.B of the proposal (77 FR 38906 to 38911, June 29, 2012) and
discussed more fully in the Integrated Science Assessment (Chapters 2,
4, 5, 6, 7, and 8) and the Policy Assessment (Chapter 2) related to
health effects associated with fine particle exposures. Section III.B.
of the proposal discusses available information on the health effects
associated with exposures to PM2.5, including the nature of
such health effects (section III.B.1) and associated limitations and
uncertainties (section III.B.2), at-risk populations (section III.B.3),
and potential PM2.5-related impacts on public health
(section III.B.4). As was true in the last two reviews, evidence from
epidemiological, controlled human exposure and animal toxicological
studies played a key role in the Integrated Science Assessment's
evaluation of the scientific evidence.
The 2006 PM NAAQS review concluded that there was ``strong
epidemiological evidence'' for linking long-term PM2.5
exposures with cardiovascular-related and lung cancer mortality and
respiratory-related morbidity and for linking short-term
PM2.5 exposures with cardiovascular-related and respiratory-
related mortality and morbidity (U.S. EPA, 2004, p. 9-46; U.S. EPA,
2005, p. 5-4). Overall, the evidence from epidemiological,
toxicological, and controlled human exposure studies supported ``likely
causal associations'' between PM2.5 and both mortality and
morbidity from cardiovascular and respiratory diseases, based on ``an
assessment of strength, robustness, and consistency in results'' (U.S.
EPA, 2004, p. 9-48).\26\
---------------------------------------------------------------------------

\26\ The term ``likely causal association'' was used in the 2004
Criteria Document to summarize the strength of the available
evidence available in the last review for PM2.5. However,
this terminology was not based on a formal framework for evaluating
evidence for inferring causation. Since the last review, the EPA has
developed a more formal framework for reaching causal determinations
with standardized language to express evaluation of the evidence
(U.S. EPA, 2009a, section 1.5).
---------------------------------------------------------------------------

In this review, based on the expanded body of evidence, the EPA
finds that:

(1) In looking across the extensive new scientific evidence
available in this review, our overall understanding of health
effects associated with fine particle exposures has been greatly
expanded. The currently available evidence is largely consistent
with evidence available in the last review and substantially
strengthens what is known about the effects associated with fine
particle exposures.
(2) A number of large multi-city epidemiological studies have
been conducted throughout the U.S., including extended analyses of
long-term exposure studies that were important to inform decision-
making in the last review. The body of currently available
scientific evidence has also been expanded greatly by the
publication of a number of new multi-city, time-series studies that
have used uniform methodologies to investigate the effects of short-
term PM2.5 exposures on public health. This body of
evidence provides a more expansive data base and considers multiple
locations representing varying regions and seasons that provide
evidence of the influence of different air pollution mixes on
PM2.5-associated health effects. These studies provide
more precise estimates of the magnitude of effects associated with
short-term PM2.5 exposure than most smaller-scale single-
city studies that were more commonly available in the last review.
These studies have reported consistent increases in morbidity and/or
premature mortality related to ambient PM2.5
concentrations, with the strongest evidence reported for
cardiovascular-related effects.
(3) In addition, the findings of new toxicological and
controlled human exposure studies greatly expand and provide
stronger support for a number of potential biological mechanisms or
pathways for cardiovascular and respiratory effects associated with
long- and short-term PM exposures. These studies provide coherence
and biological plausibility for the effects observed in
epidemiological studies.
(4) Using a more formal framework for reaching causal
determinations than used in prior reviews,\27\ the EPA concludes
that a causal relationship exists between both long- and short-term
exposures to PM2.5 and premature mortality and
cardiovascular effects and a likely causal relationship exists
between long- and short-term PM2.5 exposures and
respiratory effects. Further, there is evidence suggestive of a
causal relationship between long-term PM2.5 exposures and
other health effects, including developmental and reproductive
effects (e.g., low birth weight, infant mortality) and carcinogenic,
mutagenic, and genotoxic effects (e.g., lung cancer mortality).\28\
---------------------------------------------------------------------------

\27\ The causal framework draws upon the assessment and
integration of evidence from across epidemiological, controlled
human exposure, and toxicological studies, and the related
uncertainties that ultimately influence our understanding of the
evidence. This framework employs a five-level hierarchy that
classifies the overall weight of evidence and causality using the
following categorizations: causal relationship, likely to be causal
relationship, suggestive of a causal relationship, inadequate to
infer a causal relationship, and not likely to be a causal
relationship (U.S. EPA, 2009a, Table 1-3). The development of the
causal framework reflects considerable input from CASAC and the
public, with CASAC concluding that, ``The five-level classification
of strength of evidence for causal inference has been systemically
applied [for PM]; this approach has provided transparency and a
clear statement of the level of confidence with regard to causation,
and we recommend its continued use in future ISAs'' (Samet, 2009f,
p. 1).
\28\ These causal inferences are based not only on the more
expansive epidemiological evidence available in this review but also
reflect consideration of important progress that has been made to
advance our understanding of a number of potential biologic modes of
action or pathways for PM-related cardiovascular and respiratory
effects (U.S. EPA, 2009a, chapter 5).
---------------------------------------------------------------------------

(5) The newly available evidence significantly strengthens the
link between long- and short-term exposure to PM2.5 and
premature mortality, while providing indications that the magnitude
of the PM2.5-mortality association with long-term
exposures may be larger than previously estimated. The strongest
evidence comes from recent studies investigating long-term exposure
to PM2.5 and cardiovascular-related mortality. The
evidence supporting a causal relationship between long-term
PM2.5 exposure and mortality also includes consideration
of new studies that demonstrated an improvement in community health
following reductions in ambient fine particles.
(6) Several new studies have examined the association between
cardiovascular effects and long-term PM2.5 exposures in
multi-city studies conducted in the U.S. and Europe. While studies
were not available in the last review with regard to long-term
exposure and cardiovascular-related morbidity, recent studies have
provided new evidence linking long-term exposure to PM2.5
with an array of cardiovascular effects such as heart attacks,
congestive heart failure, stroke, and mortality. This evidence is
coherent with studies of short-term exposure to PM2.5
that have observed associations with a continuum of effects ranging
from subtle changes in indicators of cardiovascular health to
serious clinical events, such as increased hospitalizations and
emergency department visits due to cardiovascular disease and
cardiovascular mortality.
(7) Extended analyses of studies available in the last review as
well as new epidemiological studies conducted in the U.S. and abroad
provide stronger evidence of respiratory-related morbidity effects
associated with long-term PM2.5 exposure. The strongest
evidence for respiratory-related

[[Page 3104]]

effects is from studies that evaluated decrements in lung function
growth, increased respiratory symptoms, and asthma development. The
strongest evidence from short-term PM2.5 exposure studies
has been observed for increased respiratory-related emergency
department visits and hospital admissions for chronic obstructive
pulmonary disease (COPD) and respiratory infections.
(8) The body of scientific evidence is somewhat expanded from
the 2006 review but is still limited with respect to associations
between long-term PM2.5 exposures and developmental and
reproductive effects as well as cancer, mutagenic, and genotoxic
effects. The strongest evidence for an association between
PM2.5 and developmental and reproductive effects comes
from epidemiological studies of low birth weight and infant
mortality, especially due to respiratory causes during the post-
neonatal period (i.e., 1 month-12 months of age). With regard to
cancer effects, ``[m]ultiple epidemiologic studies have shown a
consistent positive association between PM2.5 and lung
cancer mortality, but studies have generally not reported
associations between PM2.5 and lung cancer incidence''
(U.S. EPA 2009a p. 2-13).
(9) Efforts to evaluate the relationships between PM composition
and health effects continue to evolve. While many constituents of
PM2.5 can be linked with differing health effects, the
evidence is not yet sufficient to allow differentiation of those
constituents or sources that may be more closely related to specific
health outcomes nor to exclude any individual component or group of
components associated with any source categories from the fine
particle mixture of concern.
(10) Specific groups within the general population are at
increased risk for experiencing adverse health effects related to PM
exposures. The currently available evidence expands our
understanding of previously identified at-risk populations (i.e.,
children, older adults, and individuals with pre-existing heart and
lung disease) and supports the identification of additional at-risk
populations (e.g., persons with lower socioeconomic status, genetic
differences). Evidence for PM-related effects in these at-risk
populations has expanded and is stronger than previously observed.
There is emerging, though still limited, evidence for additional
potentially at-risk populations, such as those with diabetes, people
who are obese, pregnant women, and the developing fetus.
(11) The population potentially affected by PM2.5 is
large. In addition, large subgroups of the U.S. population have been
identified as at-risk populations. While individual effect estimates
from epidemiological studies may be small in size, the public health
impact of the mortality and morbidity associations can be quite
large given the extent of exposure. Taken together, this suggests
that exposure to ambient PM2.5 concentrations can have
substantial public health impacts.
(12) While the currently available scientific evidence is
stronger and more consistent than in previous reviews, providing a
strong basis for decision making in this review, the EPA recognizes
that important uncertainties and limitations in the health effects
evidence remain. Epidemiological studies evaluating health effects
associated with long- and short-term PM2.5 exposures have
reported heterogeneity in responses between cities and geographic
regions within the U.S. This heterogeneity may be attributed, in
part, to differences in the fine particle composition or related to
exposure measurement error, which can introduce bias and increased
uncertainty in associated health effect estimates. Variability in
the associations observed across PM2.5 epidemiological
studies may be due in part to exposure error related to measurement-
related issues, the use of central fixed-site monitors to represent
population exposure to PM2.5, models used in lieu of or
to supplement ambient measurements, and our limited understanding of
factors that may influence exposures (e.g., topography, the built
environment, weather, source characteristics, ventilation usage,
personal activity patterns, photochemistry). In addition, where
PM2.5 and other pollutants (e.g., ozone, nitrogen
dioxide, and carbon monoxide) are correlated, it can be difficult to
distinguish the effects of the various pollutants in the ambient
mixture (i.e., co-pollutant confounding).\29\
---------------------------------------------------------------------------

\29\ A copollutant meets the criteria for potential confounding
in PM-health associations if: (1) It is a potential risk factor for
the health effect under study; (2) it is correlated with PM; and (3)
it does not act as an intermediate step in the pathway between PM
exposure and the health effect under study (U.S. EPA, 2004, p. 8-
10).
---------------------------------------------------------------------------

While uncertainties and limitations still remain in the available
health effects evidence, the Administrator judges the currently
available scientific data base to be stronger and more consistent than
in previous reviews providing a strong basis for decision making in
this review.

C. Overview of Quantitative Characterization of Health Risks

In addition to a comprehensive evaluation of the health effects
evidence available in this review, the EPA conducted an expanded
quantitative risk assessment for selected health endpoints to provide
additional information and insights to inform decisions on the primary
PM2.5 NAAQS.\30\ As discussed in section III.C of the
proposal, the approach used to develop quantitative risk estimates
associated with PM2.5 exposures was built on the approach
used and lessons learned in the last review and focused on improving
the characterization of the overall confidence in the risk estimates,
including related uncertainties, by incorporating a number of
enhancements, in terms of both the methods and data used in the
analyses.
---------------------------------------------------------------------------

\30\ The quantitative risk assessment conducted for this review
is more fully described and presented in the Risk Assessment (U.S.
EPA, 2010a) and summarized in detail in the Policy Assessment (U.S.
EPA, 2011a, sections 2.2.2. and 2.3.4.2). The scope and methodology
for this risk assessment were developed over the last few years with
considerable input from CASAC and the public as described in section
II.B.3 above.
---------------------------------------------------------------------------

The goals of this quantitative risk assessment were largely the
same as those articulated in the risk assessment conducted for the last
review. These goals included: (1) To provide estimates of the potential
magnitude of premature mortality and/or selected morbidity effects in
the population associated with recent ambient levels of
PM2.5 and with simulating just meeting the current and
alternative suites of PM2.5 standards in 15 selected urban
study areas,\31\ including, where data were available, consideration of
impacts on at-risk populations; (2) to develop a better understanding
of the influence of various inputs and assumptions on the risk
estimates to more clearly differentiate among alternative suites of
standards; and (3) to gain insights into the distribution of risks and
patterns of risk reductions and the variability and uncertainties in
those risk estimates. In addition, the quantitative risk assessment
included nationwide estimates of the potential magnitude of premature
mortality associated with long-term exposure to recent ambient
PM2.5 concentrations to more broadly characterize this risk
on a national scale and to support the interpretation of the more
detailed risk estimates generated for selected urban study areas.
---------------------------------------------------------------------------

\31\ The Risk Assessment concluded that these 15 urban study
areas were generally representative of urban areas in the U.S.
likely to experience relatively elevated levels of risk related to
ambient PM2.5 exposure with the potential for better
characterization at the higher end of that distribution (U.S. EPA,
2011a, p. 2-42; U.S. EPA, 2010a, section 4.4, Figure 4-17). The
representativeness analysis also showed that the 15 urban study
areas do not capture areas with the highest baseline morality risks
or the oldest populations (both of which can result in higher
PM2.5-related mortality estimates). However, some of the
areas with the highest values for these attributes had relatively
low PM2.5 concentrations (e.g., urban areas in Florida)
and, consequently, the Risk Assessment concluded failure to include
these areas in the set of urban study areas was unlikely to exclude
high PM2.5-risk locations (U.S. EPA, 2010a, section
4.4.1).
---------------------------------------------------------------------------

The expanded and updated risk assessment conducted in this review
included estimates of risk for: (1) All-cause, ischemic heart disease-
related, cardiopulmonary-related, and lung cancer-related mortality
associated with long-term PM2.5 exposure; (2) non-
accidental, cardiovascular-related, and respiratory-related mortality
associated with short-term PM2.5 exposure; and (3)
cardiovascular-related and respiratory-related hospital admissions and
asthma-related emergency department visits

[[Page 3105]]

associated with short-term PM2.5 exposure.\32\
---------------------------------------------------------------------------

\32\ The evidence available for these selected health effect
endpoints generally focused on the entire population, although some
information was available to support analyses that considered
differences in estimated risk for at-risk populations including
older adults and persons with pre-existing cardiovascular and
respiratory diseases.
---------------------------------------------------------------------------

The Risk Assessment included a core set of risk estimates
supplemented by an alternative set of risk results generated using
single-factor and multi-factor sensitivity analyses. The core set of
risk estimates was developed using the combination of modeling elements
and input data sets identified in the Risk Assessment as having higher
confidence relative to inputs used in the sensitivity analyses. The
results of the sensitivity analyses provided information to evaluate
and rank the potential impacts of key sources of uncertainty on the
core risk estimates. In addition, the sensitivity analyses represented
a set of reasonable alternatives to the core set of risk estimates that
fell within an overall set of plausible risk estimates surrounding the
core estimates.
The EPA recognized that there were many sources of variability and
uncertainty inherent in the inputs to its quantitative risk
assessment.\33\ The design of the risk assessment included a number of
elements to address these issues in order to increase the overall
confidence in the risk estimates generated for the 15 urban study
areas, including using guidance from the World Health Organization
(WHO, 2008) as a framework for characterizing uncertainty in the
analyses.\34\
---------------------------------------------------------------------------

\33\ Variability refers to the heterogeneity of a variable of
interest within a population or across different populations.
Uncertainty refers to the lack of knowledge regarding the actual
values of inputs to an analysis (U.S. EPA, 2010a, p. 3-63).
\34\ The extent to which key sources of potential variability
were (or were not) fully captured in the design of the risk
assessment are discussed in section 3.5.2 of the Risk Assessment
(U.S. EPA, 2010a, pp. 3-67 to 3-69).
---------------------------------------------------------------------------

With respect to the sources of variability, the Risk Assessment
considered those that contributed to differences in risk across urban
study areas, but did not directly affect the degree of risk reduction
associated with the simulation of just meeting current or alternative
standard levels (e.g., differences in baseline incidence rates,
demographics and population behavior). The Risk Assessment also focused
on factors that not only introduced variability into risk estimates
across study areas, but also played an important role in determining
the magnitude of risk reductions upon simulation of just meeting
current or alternative standard levels (e.g., peak-to-mean ratios of
ambient PM2.5 concentrations within individual urban study
areas and the nature of the rollback approach used to simulate just
meeting the current or alternative standards). Key sources of potential
variability that were likely to affect population risks included the
following: (1) Intra-urban variability in ambient PM2.5
concentrations, including PM2.5 composition; (2) variability
in the patterns of reductions in PM2.5 concentrations
associated with different rollback approaches when simulating just
meeting the current or alternative standards; (3) co-pollutant
exposures; (4) factors related to demographic and socioeconomic status;
(5) behavioral differences across urban study areas (e.g., time spent
outdoors); (6) baseline incidence rates; and (7) longer-term temporal
variability in ambient PM2.5 concentrations reflecting
meteorological trends as well as future changes in the mix of
PM2.5 sources, including changes in air quality related to
future regulatory actions.
With regard to uncertainties, single and multi-factor sensitivity
analyses were combined with a qualitative analysis to assess the impact
of potential sources of uncertainty on the core risk estimates. Key
sources of uncertainty included: (1) Characterizing intra-urban
population exposure in the context of epidemiological studies linking
PM2.5 to specific health effects; (2) statistical fit of the
concentration-response functions for short-term exposure-related health
endpoints; (3) shape of the concentration-response functions; (4)
specifying the appropriate lag structure for short-term exposure
studies; (5) transferability of concentration-response functions from
study locations to urban study area locations for long-term exposure-
related health endpoints; (6) use of single-city versus multi-city
studies in the derivation of concentration-response functions; (7)
impact of historical air quality on estimates of health risk associated
with long-term PM2.5 exposures; and (8) potential variation
in effect estimates reflecting compositional differences in
PM2.5.
Beyond characterizing uncertainty and variability, a number of
design elements were included in the risk assessment to increase the
overall confidence in the risk estimates generated for the 15 urban
study areas (U.S. EPA, 2011a, pp. 2-38 to 2-41). These elements
included: (1) Use of a deliberative process for specifying components
of the risk model that reflects consideration of the latest research on
PM2.5 exposure and risk (U.S. EPA, 2010a, section 5.1.1);
(2) integration of key sources of variability into the design as well
as the interpretation of risk estimates (U.S. EPA, 2010a, section
5.1.2); (3) assessment of the degree to which the urban study areas are
representative of areas in the U.S. experiencing higher
PM2.5-related risk (U.S. EPA, 2010a, section 5.1.3); and (4)
identification and assessment of important sources of uncertainty and
the impact of these uncertainties on the core risk estimates (U.S. EPA,
2010a, section 5.1.4). Further, additional analyses examined potential
bias and overall confidence in the risk estimates. Greater confidence
is associated with risk estimates based on simulated annual mean
PM2.5 concentrations that are within the region of the air
quality distribution used in deriving the concentration-response
functions where the bulk of the data reside (e.g., within one standard
deviation around the long-term mean PM2.5 concentration)
(U.S. EPA, 2011a, p. 2-38).
Key observations and insights from the PM2.5 risk
assessment, together with important caveats and limitations, were
discussed in section III.C.3 of the proposal. In general, in
considering the set of quantitative risk estimates and related
uncertainties and limitations related to long- and short-term
PM2.5 exposure together with consideration of the health
endpoints which could not be quantified, the Policy Assessment
concluded this information provided strong evidence that risks
estimated to remain upon simulating just meeting the current suite of
PM2.5 standards are important from a public health
perspective, both in terms of severity and magnitude of effects.
Patterns of increasing estimated risk reductions were generally
observed as either the annual or 24-hour standard level, or both, were
reduced over the ranges considered in the Risk Assessment.
The magnitude of both long- and short-term exposure-related risk
estimated to remain upon just meeting the current suite of standards as
well as alternative standard levels was strongly associated with the
simulated change in annual mean PM2.5 concentrations.
Although long- and short-term exposure-related mortality rates have
similar patterns in terms of the subset of urban study areas
experiencing risk reductions for the current suite of standard levels,
the magnitude of risk remaining is higher for long-term exposure-
related mortality and substantially lower for short-term exposure-
related mortality. Short-term exposure-related morbidity risk estimates
were greater for cardiovascular-related than respiratory-related events
and emergency

[[Page 3106]]

department visits for asthma-related events were significant:
Furthermore, most of the aggregate risk associated with short-term
exposures was not primarily driven by the small number of days with
PM2.5 concentrations in the upper tail of the air quality
distribution, but rather by the large number of days with
PM2.5 concentrations at and around the mean of the
distribution, that is, the 24-hour average concentrations that are in
the low- to mid-range, well below the peak 24-hour concentrations (U.S.
EPA, 2011a, p. 2-3).
With regard to characterizing estimates of PM2.5-related
risk associated with simulation of alternative standards, the Policy
Assessment recognized that greater overall confidence was associated
with estimates of risk reduction than for estimates of absolute risk
remaining (U.S. EPA, 2011a, p. 2-94). Furthermore, the Policy
Assessment recognized that estimates of absolute risk remaining for
each of the alternative standard levels considered, particularly in the
context of long-term exposure-related mortality, may be
underestimated.\35\ In addition, the Policy Assessment observed that in
considering the overall confidence associated with the quantitative
analyses, the Risk Assessment recognized that: (1) Substantial
variability existed in the magnitude of risk remaining across urban
study areas and (2) in general, higher confidence was associated with
risk estimates based on PM2.5 concentrations near the mean
PM2.5 concentrations in the underlying epidemiological
studies providing the concentration-response functions (e.g., within
one standard deviation of the mean PM2.5 concentration
reported). Furthermore, although the Risk Assessment estimated that the
alternative 24-hour standard levels considered (when controlling) would
result in additional estimated risk reductions beyond those estimated
for alternative annual standard levels alone, these additional
estimated reductions were highly variable. Conversely, the Risk
Assessment recognized that alternative annual standard levels, when
controlling, resulted in more consistent risk reductions across urban
study areas, thereby potentially providing a more consistent degree of
public health protection (U.S. EPA, 2010a, p. 5-17).
---------------------------------------------------------------------------

\35\ Based on the consideration of both the qualitative and
quantitative assessments of uncertainty, the Risk Assessment
concluded that it is unlikely that the estimated risks are over-
stated, particularly for premature mortality related to long-term
PM2.5 exposures. In fact, the Policy Assessment and the
Risk Assessment concluded that the core risk estimates for this
category of health effects may well be biased low based on
consideration of alternative model specifications evaluated in the
sensitivity analyses (U.S. EPA, 2011a, p. 2-41; U.S. EPA, 2010a, p.
5-16; Figures 4-7 and 4-8). In addition, the Policy Assessment
recognized that the currently available scientific information
included evidence for a broader range of health endpoints and at-
risk populations beyond those included in the quantitative risk
assessment, including decrements in lung function growth and
respiratory symptoms in children as well as reproductive and
developmental effects (U.S. EPA, 2011a, section 2.2.1).
---------------------------------------------------------------------------

D. Conclusions on the Adequacy of the Current Primary PM2.5
Standards

1. Introduction
The initial issue to be addressed in the current review of the
primary PM2.5 standards is whether, in view of the advances
in scientific knowledge and other information reflected in the
Integrated Science Assessment, the Risk Assessment, and the Policy
Assessment, the existing standards should be retained or revised. In
considering the adequacy of the current suite of PM2.5
standards, the Administrator has considered the large body of evidence
presented and assessed in the Integrated Science Assessment (U.S. EPA,
2009a), the quantitative assessment of risks, staff conclusions and
associated rationales presented in the Policy Assessment, views
expressed by CASAC, and public comments. The Administrator has taken
into account both evidence- and risk-based considerations \36\ in
developing final conclusions on the adequacy of the current primary
PM2.5 standards.
---------------------------------------------------------------------------

\36\ Evidence-based considerations include the assessment of
epidemiological, toxicological, and controlled human exposure
studies evaluating long- or short-term exposures to
PM2.5, with supporting evidence related to dosimetry and
potential pathways/modes of action, as well as the integration of
evidence across each of these disciplines, as assessed in the
Integrated Science Assessment (U.S. EPA, 2009a) and focus on the
policy-relevant considerations as discussed in section III.B above
and in the Policy Assessment (U.S. EPA, 2011a, section 2.2.1). Risk-
based considerations draw from the results of the quantitative
analyses presented in the Risk Assessment (U.S. EPA, 2010a) and
focus on the policy-relevant considerations as discussed in section
III.C above and in the Policy Assessment (U.S. EPA, 2011a, section
2.2.2).
---------------------------------------------------------------------------

a. Evidence- and Risk-based Considerations in the Policy Assessment
In considering the available epidemiological evidence in this
review, the Policy Assessment took a broader approach than was used in
the last review. This approach reflected the more extensive and
stronger body of evidence available since the last review on health
effects related to both long- and short-term exposure to
PM2.5. As discussed in section III.A.3 above, this broader
approach focused on setting the annual standard as the ``generally
controlling'' standard for lowering both short- and long-term
PM2.5 concentrations and so providing requisite protection
to public health. In conjunction with such an annual standard, this
approach focused on setting the 24-hour standard to provide
supplemental protection against days with high peak PM2.5
concentrations.
In addressing the question whether the evidence now available in
this review supports consideration of standards that are more
protective than the current PM2.5 standards, the Policy
Assessment considered whether: (1) Statistically significant health
effects associations with long- or short-term exposures to fine
particles occur in areas that would likely have met the current
PM2.5 standards [see American Trucking Associations, 283 F.
3d at 369, 376 (revision of level of PM NAAQS justified when health
effects are observed in areas meeting the existing standard)], and (2)
associations with long-term exposures to fine particles extend down to
lower air quality concentrations than had previously been observed.
With regard to associations observed in long-term PM2.5
exposure studies, the Policy Assessment recognized that extended
follow-up analyses of the ACS and Harvard Six Cities studies provided
consistent and stronger evidence of an association with mortality at
lower air quality distributions than had previously been observed (U.S.
EPA, 2011a, pp. 2-31 to 2-32). The original and reanalysis of the ACS
study reported positive and statistically significant effects
associated with a long-term mean PM2.5 concentration of 18.2
[mu]g/m\3\ across 50 metropolitan areas for 1979 to 1983 (Pope et al.,
1995; Krewski et al., 2000).\37\ In extended analyses, positive and
statistically significant effects of approximately similar magnitude
were associated with declining PM2.5 concentrations, from an
aggregate long-term mean in 58 metropolitan areas of 21.2 [micro]g/m\3\
in the original monitoring period (1979 to 1983) to 14.0 [micro]g/m\3\
for 116 metropolitan areas in the most recent years evaluated (1999-
2000), with an overall average across the two study periods in 51
metropolitan areas of 17.7 [micro]g/m\3\ (Pope et al., 2002; Krewski et
al., 2009). With regard to the Harvard Six Cities Study, the original
and reanalysis reported positive and statistically significant effects
associated

[[Page 3107]]

with a long-term mean PM2.5 concentration of 18.0 [mu]g/m\3\
for 1980 to 1985 (Dockery et al., 1993; Krewski et al., 2000). In an
extended follow-up of this study, the aggregate long-term mean
concentration across all years evaluated was 16.4 [mu]g/m\3\ for 1980
to 1988 \38\ (Laden et al., 2006). In an additional analysis of the
extended follow-up of the Harvard Six Cities study, investigators
reported that the concentration-response relationship was linear and
``clearly continuing below the level'' of the current annual standard
(U.S. EPA, 2009a, p. 7-92; Schwartz et al., 2008).
---------------------------------------------------------------------------

\37\ The study periods referred to in the Policy Assessment
(U.S. EPA, 2011a) and in this final rule reflect the years of air
quality data that were included in the analyses, whereas the study
periods identified in the Integrated Science Assessment (U.S. EPA,
2009a) reflect the years of health event data that were included.
\38\ Aggregate mean concentration provided by study author
(personal communication from Dr. Francine Laden, 2009).
---------------------------------------------------------------------------

Cohort studies conducted since the last review provided additional
evidence of mortality associated with air quality distributions that
are generally lower than those reported in the ACS and Harvard Six
Cities studies, with effect estimates that were similar or, in some
studies, significantly greater in magnitude than in the ACS and Harvard
Six Cities studies (see also, section III.D.1.a of the proposal, 77 FR
38918 to 28919; U.S. EPA, 2011a, pp. 2-32 to 2-33). The Women's Health
Initiative (WHI) study reported positive and most often statistically
significant associations between long-term PM2.5 exposure
and cardiovascular-related mortality as well as morbidity effects, with
much larger relative risk estimates for mortality than in the ACS and
Harvard Six Cities studies, at an aggregate long-term mean
PM2.5 concentration of 12.9 [mu]g/m\3\ for 2000 (Miller et
al., 2007).\39\
---------------------------------------------------------------------------

\39\ The Policy Assessment noted that in comparison to other
long-term exposure studies, the Miller et al. (2007) study was more
limited in that it was based on only one year of air quality data
(U.S. EPA, 2011a, p. 2-82). The proposal further noted that the air
quality data considered were extrapolated from that one single year
of air quality data (2000) to the whole study, and that the air
quality data post-dated the years of health events considered (i.e.,
1994 to 1998) (77 FR 38918, fn 62).
---------------------------------------------------------------------------

Using the Medicare cohort, Eftim et al. (2008) reported somewhat
higher effect estimates than in the ACS and Harvard Six Cities studies
with aggregate long-term mean concentrations of 13.6 [mu]g/m\3\ and
14.1 [mu]g/m\3\, respectively, for 2000 to 2002. Zeger et al. (2008)
reported associations between long-term PM2.5 exposure and
mortality for the eastern region of the U.S. at an aggregated long-term
PM2.5 median concentration of 14.0 [mu]g/m\3\, although no
association was reported for the western region with an aggregate long-
term PM2.5 median concentration of 13.1 [mu]g/m\3\ (U.S.
EPA, 2009a, p. 7-88).\40\
---------------------------------------------------------------------------

\40\ Zeger et al. (2008) also reported positive and
statistically significant effects for the central region, with an
aggregate long-term mean PM2.5 concentration of 10.7
[mu]g/m\3\. However, in contrast to the eastern and western risk
estimates, the central risk estimate increased with adjustment for
COPD (used as a proxy for smoking status). Due to the potential for
confounding bias influencing the risk estimate for the central
region, the Policy Assessment did not focus on the results reported
in the central region to inform the adequacy of the current suite of
standards or alternative annual standard levels (U.S. EPA, 2011a, p.
2-32).
---------------------------------------------------------------------------

Premature mortality in children reported in a national infant
mortality study as well as mortality in a cystic fibrosis cohort
including both children and adults reported positive but statistically
nonsignificant effects associated with long-term aggregate mean
concentrations of 14.8 [mu]g/m\3\ and 13.7 [mu]g/m\3\, respectively
(Woodruff et al., 2008; Goss et al., 2004).
With respect to respiratory morbidity effects associated with long-
term PM2.5 exposure, the across-city mean of 2-week average
PM2.5 concentrations reported in the initial Southern
California Children's Health Study was approximately 15.1 [micro]g/m\3\
(Peters et al., 1999). These results were found to be consistent with
results of cross-sectional analyses of the 24-Cities Study (Dockery et
al., 1996; Raizenne et al., 1996), which reported a long-term cross-
city mean PM2.5 concentration of 14.5 [mu]g/m\3\.\41\ In
this review, extended analyses of the Southern California Children's
Health Study provide stronger evidence of PM2.5-related
respiratory effects, at lower air quality concentrations than had
previously been reported, with a four-year aggregate mean concentration
of 13.8 [mu]g/m\3\ across the 12 study communities (McConnell et al.,
2003; Gauderman et al., 2004, U.S. EPA, 2009a, Figure 7-4).
---------------------------------------------------------------------------

\41\ See American Farm Bureau Federation v. EPA, 559 F. 3d at
525 (noting the importance of these studies, as well as EPA's
failure to properly take them into account).
---------------------------------------------------------------------------

In also considering health effects for which the Integrated Science
Assessment concluded evidence was suggestive of a causal relationship,
the Policy Assessment noted a limited number of birth outcome studies
that reported positive and statistically significant effects related to
aggregate long-term mean PM2.5 concentrations down to
approximately 12 [mu]g/m\3\ (U.S. EPA, 2011a, p. 2-33).
Collectively, the Policy Assessment concluded that currently
available evidence provided support for associations between long-term
PM2.5 exposure and mortality and morbidity effects that
extend to distributions of PM2.5 concentrations that are
lower than those that had previously been associated with such effects,
with aggregate long-term mean PM2.5 concentrations extending
to well below the level of the current annual standard.
The Policy Assessment also considered the long-term mean
PM2.5 concentrations in short-term exposure studies in
assessing the appropriateness of the level of the current annual
standard. See American Farm Bureau Federation v. EPA, 559 F. 3d at 522,
523-24 (remanding 2006 standard because the EPA had not adequately
explained its choice not to consider long-term means of short-term
exposure studies in assessing adequacy of primary annual
PM2.5 standard). In light of the mixed findings reported in
single-city, short-term exposure studies, the Policy Assessment placed
comparatively greater weight on the results from multi-city studies in
considering the adequacy of the current suite of standards (U.S. EPA,
2011a, pp. 2-34 to 2-35).
With regard to associations reported in short-term PM2.5
exposure studies, the Policy Assessment recognized that long-term mean
concentrations reported in new multi-city U.S. and Canadian studies
provided evidence of associations between short-term PM2.5
exposure and mortality at similar air quality distributions to those
that had previously been observed in an 8-cities Canadian study
(Burnett and Goldberg, 2003; aggregate long-term mean PM2.5
concentration of 13.3 [mu]g/m\3\). In a multi-city time-series analysis
of 112 U.S. cities, Zanobetti and Schwartz (2009) reported a positive
and statistically significant association with all-cause,
cardiovascular-related (e.g., heart attacks, stroke), and respiratory-
related mortality and short-term PM2.5 exposure, in which
the aggregate long-term mean PM2.5 concentration was 13.2
[mu]g/m\3\ (U.S. EPA, 2009a, Figure 6-24). Furthermore, city-specific
effect estimates indicated the association between short-term exposure
to PM2.5 and total mortality and cardiovascular- and
respiratory-related mortality was consistently positive for an
overwhelming majority (99 percent) of the 112 cities across a wide
range of air quality concentrations (long-term mean concentrations
ranging from 6.6 [mu]g/m\3\ to 24.7 [mu]g/m\3\; U.S. EPA, 2009a, Figure
6-24, p. 6-178 to 179). The EPA staff noted that for all-cause
mortality, city-specific effect estimates were statistically
significant for 55 percent of the 112 cities, with long-term city-mean
PM2.5 concentrations ranging from 7.8 [mu]g/m\3\ to 18.7
[mu]g/m\3\ and 24-hour PM2.5 city-mean 98th percentile
concentrations ranging from 18.4 to 64.9

[[Page 3108]]

[mu]g/m\3\ (personal communication with Dr. Antonella Zanobetti,
2009).\42\
---------------------------------------------------------------------------

\42\ Single-city Bayes-adjusted effect estimates for the 112
cities analyzed in Zanobetti and Schwartz (2009) were provided by
the study authors (personal communication with Dr. Antonella
Zanobetti, 2009; see also U.S. EPA, 2009a, Figure 6-24).
---------------------------------------------------------------------------

With regard to cardiovascular and respiratory morbidity effects, in
the first analysis of the Medicare cohort conducted by Dominici et al.
(2006a) across 204 U.S. counties, investigators reported a
statistically significant association with hospitalizations for
cardiovascular and respiratory diseases and short-term PM2.5
exposure, in which the aggregate long-term mean PM2.5
concentration was 13.4 [mu]g/m\3\. Furthermore, a sub-analysis
restricted to days with 24-hour average concentrations of
PM2.5 at or below 35 [mu]g/m\3\ indicated that, in spite of
a reduced statistical power from a smaller number of study days,
statistically significant associations were still observed between
short-term exposure to PM2.5 and hospital admissions for
cardiovascular and respiratory diseases (Dominici, 2006b).\43\ In an
extended analysis of this cohort, Bell et al. (2008) reported a
positive and statistically significant increase in cardiovascular
hospitalizations associated with short-term PM2.5 exposure,
in which the aggregate long-term mean PM2.5 concentration
was 12.9 [mu]g/m\3\. These results, along with the observation that
approximately 50 percent of the 204 county-specific mean 98th
percentile PM2.5 concentrations in the study aggregated
across all years were below the 24-hour standard of 35 [mu]g/m\3\, not
only indicated that effects are occurring in areas that would meet the
current standards but also suggested that the overall health effects
observed across the U.S. are not primarily driven by the higher end of
the PM2.5 air quality distribution (Bell, 2009a, personal
communication from Dr. Michelle Bell regarding air quality data for
Bell et al., 2008 and Dominici et al., 2006a).
---------------------------------------------------------------------------

\43\ This sub-analysis was not included in the original
publication (Dominici et al., 2006a). The study authors provided
sub-analysis results for the Administrator's consideration as a
letter to the docket following publication of the proposed rule in
January 2006 (personal communication with Dr. Francesca Dominici,
2006b). As noted in section III.A.3, this study is part of the basis
for the conclusion that there is no evidence suggesting that risks
associated with long-term exposures are likely to be
disproportionately driven by peak 24-hour concentrations.
---------------------------------------------------------------------------

Collectively, the Policy Assessment concluded that the findings
from short-term PM2.5 exposure studies provided evidence of
PM2.5-associated health effects occurring in areas that
would likely have met the current suite of PM2.5 standards
(U.S. EPA, 2011a, p. 2-35). These findings were further bolstered by
evidence of statistically significant PM2.5-related health
effects occurring in analyses restricted to days in which 24-hour
average PM2.5 concentrations were below 35 [mu]g/m\3\
(Dominici, 2006b).
In evaluating the currently available scientific evidence, as
summarized in section III.B of the proposal, the Policy Assessment
first concluded that there was stronger and more consistent and
coherent support for associations between long- and short-term
PM2.5 exposures and a broad range of health outcomes than
was available in the last review, providing the basis for fine particle
standards at least as protective as the current PM2.5
standards (U.S. EPA, 2011a, p. 2-26). Having reached this initial
conclusion, the Policy Assessment addressed the question of whether the
available evidence supported consideration of standards that were more
protective than the current standards. In so doing, the Policy
Assessment considered whether there was now evidence that health effect
associations have been observed in areas that likely met the current
suite of PM2.5 standards. As discussed above, long- and
short-term PM2.5 exposure studies provided evidence of
associations with mortality and cardiovascular and respiratory effects
both at lower ambient PM2.5 concentrations than had been
observed in the previous review and at concentrations allowed by the
current standards (U.S. EPA, 2011a, p. 2-35).
In reviewing this information, the Policy Assessment recognized
that important limitations and uncertainties associated with this
expanded body of scientific evidence, as discussed in section III.B.2
of the proposal, needed to be carefully considered in determining the
weight to be placed on the body of studies available in this review.
Taking these limitations and uncertainties into consideration, the
Policy Assessment concluded that the currently available evidence
clearly calls into question whether the current suite of primary
PM2.5 standards protects public health with an adequate
margin of safety from effects associated with long- and short-term
exposures. Furthermore, the Policy Assessment concluded this evidence
provides strong support for considering fine particle standards that
would afford increased protection beyond that afforded by the current
standards (U.S. EPA, 2011a, p. 2-35).
In addition to evidence-based consideration, the Policy Assessment
also considered the extent to which health risks estimated to occur
upon simulating just meeting the current PM2.5 standards may
be judged to be important from a public health perspective, taking into
account key uncertainties associated with the quantitative health risk
estimates. In so doing, the Policy Assessment first noted that the
quantitative risk assessment addresses: (1) The core PM2.5-
related risk estimates; (2) the related uncertainty and sensitivity
analyses, including additional sets of reasonable risk estimates
generated to supplement the core analysis; (3) an assessment of the
representativeness of the urban study areas within a national context;
\44\ and (4) consideration of patterns in design values and air quality
monitoring data to inform interpretation of the risk estimates, as
discussed in section III.C above.
---------------------------------------------------------------------------

\44\ Based on analyses of the representativeness of the 15 urban
study areas in the broader national context, the Policy Assessment
concludes that these study areas are generally representative of
urban areas in the U.S. likely to experience relatively elevated
levels of risk related to ambient PM2.5 exposures (U.S.
EPA, 2011a, p. 2-42).
---------------------------------------------------------------------------

In considering the health risks estimated to remain upon simulation
of just meeting the current suite of standards and considering both the
qualitative and quantitative assessment of uncertainty completed as
part of the assessment, the Policy Assessment concluded these risks are
important from a public health standpoint and provided strong support
for consideration of alternative standards that would provide increased
protection beyond that afforded by the current PM2.5 (U.S.
EPA, 2011a, pp. 2-47 to 2-48). This conclusion reflected consideration
of both the severity and the magnitude of the effects. For example, the
Risk Assessment indicated the possibility that premature deaths related
to ischemic heart disease associated with long-term PM2.5
exposure alone would likely be on the order of thousands of deaths per
year in the 15 urban study areas upon simulating just meeting the
current standards \45\ (U.S. EPA, 2011a, pp. 2-46 to 2-47). Moreover,
additional risks were anticipated for premature mortality related to
cardiopulmonary effects and lung cancer associated with long-term
PM2.5 exposure as well as mortality and cardiovascular- and
respiratory-related morbidity effects (e.g., hospital admissions,
emergency department visits) associated with short-term
PM2.5 exposures. Based on the consideration of both
qualitative and

[[Page 3109]]

quantitative assessments of uncertainty completed as part of the
quantitative risk assessment, the Risk Assessment concluded that it was
unlikely that the estimated risks are over-stated, particularly for
mortality related to long-term PM2.5 exposure, and may well
be biased low based on consideration of alternative model
specifications evaluated in the sensitivity analyses (U.S. EPA, 2010a,
p. 5-16; U.S. EPA, 2011a, p. 2-41). Furthermore, the currently
available scientific information summarized in section III.B of the
proposal provided evidence for a broader range of health endpoints and
at-risk populations beyond those included in the quantitative risk
assessment (U.S. EPA, 2011a, p. 2-47).
---------------------------------------------------------------------------

\45\ Premature mortality for all causes attributed to
PM2.5 exposure was estimated to be on the order of tens
of thousands of deaths per year on a national scale based on 2005
air quality data (U.S. EPA, 2010a, Appendix G, Table G-1).
---------------------------------------------------------------------------

b. CASAC Advice
The CASAC, based on its review of drafts of the Integrated Science
Assessment, the Risk Assessment, and the Policy Assessment, provided an
array of advice both with regard to interpreting the scientific
evidence and quantitative risk assessment, as well as with regard to
consideration of the adequacy of the current PM2.5 standards
(Samet, 2009a,b,c,d,e,f; Samet 2010a,b,c,d). With regard to the
adequacy of the current standards, CASAC concluded that the ``currently
available information clearly calls into question the adequacy of the
current standards'' (Samet, 2010d, p. i) and that the current standards
are ``not protective'' (Samet, 2010d, p. 1). Further, in commenting on
the first draft Policy Assessment, CASAC noted:

With regard to the integration of evidence-based and risk-based
considerations, CASAC concurs with EPA's conclusion that the new
data strengthens the evidence available on associations previously
considered in the last round of the assessment of the
PM2.5 standard. CASAC also agrees that there are
significant public health consequences at the current levels of the
standard that justify consideration of lowering the PM2.5
NAAQS further (Samet, 2010c, p. 12).

c. Administrator's Proposed Conclusions Concerning the Adequacy of the
Current Primary PM2.5 Standards

At the time of the proposal, in considering the body of scientific
evidence, the Administrator concluded there was stronger and more
consistent and coherent support for associations between long- and
short-term PM2.5 exposure and a broader range of health
outcomes than was available in the last review, providing the basis for
fine particle standards at least as protective as the current
PM2.5 standards. In particular, the Administrator recognized
in section III.D.4 of the proposal that the Integrated Science
Assessment concluded that the results of epidemiological and
experimental studies form a plausible and coherent data set that
supports a causal relationship between long- and short-term
PM2.5 exposures and mortality and cardiovascular effects and
a likely causal relationship between long- and short-term
PM2.5 exposures and respiratory effects. Furthermore, the
Administrator reflected that effects had been observed at lower ambient
PM2.5 concentrations than what had been observed in the last
review, including at ambient PM2.5 concentrations in areas
that likely met the current PM2.5 NAAQS. With regard to the
results of the quantitative risk assessment, the Administrator noted
that the Risk Assessment concluded that the risks estimated to remain
upon simulation of just meeting the current standards were important
from a public health standpoint in terms of both the severity and
magnitude of the effects.
At the time of the proposal, in considering whether the current
suite of PM2.5 standards should be revised to provide
requisite public health protection, the Administrator carefully
considered the staff conclusions and rationales presented in the Policy
Assessment, the advice and recommendations from CASAC, and public
comments to date on this issue. In so doing, the Administrator placed
primary consideration on the evidence obtained from the epidemiological
studies and provisionally found the evidence of serious health effects
reported in long- and short-term exposure studies conducted in areas
that would have met the current standards to be compelling, especially
in light of the extent to which such studies are part of an overall
pattern of positive and frequently statistically significant
associations across a broad range of studies that collectively
represent a strong and robust body of evidence.
As discussed in the Integrated Science Assessment and Policy
Assessment, the Administrator recognized that much progress has been
made since the last review in addressing some of the key uncertainties
that were important considerations in establishing the current suite of
PM2.5 standards. For example, progress made since the last
review provides increased confidence in the long- and short-term
exposure studies as a basis for considering whether any revision of the
annual standard is appropriate and increased confidence in the short-
term exposure studies as a basis for considering whether any revision
of the 24-hour standard is appropriate.\46\
---------------------------------------------------------------------------

\46\ The EPA notes that this increased confidence in the long-
and short-term associations generally reflects less uncertainty as
to the likely causal nature of such associations, but does not
address directly the question of the extent to which such
associations remain toward the lower end of the range of ambient
PM2.5 concentrations. This question is central to the
Agency's evaluation of the relevant evidence to determine
appropriate standards levels, as discussed below in section III.E.4.
---------------------------------------------------------------------------

Based on her consideration of these conclusions, as well as
consideration of CASAC's conclusion that the evidence and risk
assessment clearly called into question the adequacy of the public
health protection provided by the current PM2.5 NAAQS and
public comments on the proposal, the Administrator provisionally
concluded that the current primary PM2.5 standards, taken
together, were not requisite to protect public health with an adequate
margin of safety and that revision was needed to provide increased
public health protection. The Administrator provisionally concluded
that the scientific evidence and information on risk provided strong
support for consideration of alternative standards that would provide
increased public health protection beyond that afforded by the current
PM2.5 standards.
2. Comments on the Need for Revision
This section addresses general comments based on relevant facts
that either support or oppose any change to the current suite of
primary PM2.5 standards. Comments on specific long- and
short-term exposure studies that relate to consideration of the
appropriate levels of the annual and 24-hour standards are addressed in
section III.E.4 below. Many public comments asserted that the current
PM2.5 standards are insufficient to protect public health
with an adequate margin of safety and that revisions to the standards
are therefore appropriate, indeed necessitated.
Among those calling for revisions to the current standards were the
Children's Health Protection Advisory Committee (CHPAC); major medical
and public health groups including the American Heart Association
(AHA), American Lung Association (ALA), American Public Health
Association (APHA), American Thoracic Society (ATS); the Physicians for
Social Responsibility (PSR); major environmental groups such as the
Clean Air Council, Clean Air Task Force, Earthjustice, Environmental
Defense Fund (EDF), National Resources Defense Council (NRDC), and
Sierra Club; many environmental justice organizations as

[[Page 3110]]

well as medical doctors, academic researchers, health professionals,
and many private citizens. For example, the American Heart Association
and other major national public health and medical organizations stated
that, ``[o]ur organizations are keenly aware of the public health and
medical threats from particulate matter'' and called on the EPA to
``significantly strengthen'' both the annual and 24-hour
PM2.5 standards ``to help us protect the health of our
patients and our nation'' (AHA et al., 2012, pp. 1 and 13). AHA et al.
and ALA et al., as well as a group of more than 350 physicians,
environmental health researchers, and public health and medical
professionals articulated similar comments on the available evidence:

Ample scientific evidence supports adopting tighter standards to
protect the health of people who are most susceptible to the serious
health effects of these pollutants. More than 10,000 peer-reviewed
scientific studies have been published since 1997 when EPA adopted
the current annual standard. These studies validate and extend
earlier epidemiologic research linking both acute and chronic fine
particle pollution with serious morbidity and mortality. The newer
research has also expanded our understanding of the range of health
outcomes associated with PM and has identified adverse respiratory
and cardiovascular health effects at lower exposure levels than
previously reported. As discussed and interpreted in the EPA's 2009
Integrated Science Assessment for Particulate Matter, the new
evidence reinforces already strong existing studies and supports the
conclusion that PM2.5 is causally associated with
numerous adverse health effects in humans at exposure levels far
below the current standard. Such a conclusion demands prompt action
to protect human health. (AHA et al., 2012, pp. 1 to 2; ALA et al.,
pp. 4 to 5; similar comment submitted by Rom et al., 2012, p. 1).

All of these medical and public health commenters stated that the
current PM2.5 standards need to be revised, and that even
more protective standards than those proposed by the EPA are needed to
adequately protect public health, particularly for at-risk populations.
Many environmental justice organizations and individual commenters also
expressed such views.
The National Association of Clean Air Agencies (NACAA), the
Northeast States for Coordinated Air Use Management (NESCAUM), and many
State and local air agencies and health departments who commented on
the PM2.5 standards supported revision of the suite of
current PM2.5 standards, as did five state attorneys general
(Schneiderman et al., 2012) and the National Tribal Air Association
(NTAA).
These commenters based their views chiefly on the body of evidence
and technical analyses presented and discussed in the Integrated
Science Assessment, the Risk Assessment, and the Policy Assessment
finding the available scientific information to be stronger and more
compelling than in the last review. These commenters generally placed
much weight on CASAC's recommendation to revise the PM2.5
standards to provide increased public health protection and on the EPA
staff conclusions presented in the final Policy Assessment.
Some of these commenters specifically mentioned extended analyses
of seminal long-term exposure studies--the ACS (Krewski et al., 2009),
Harvard Six Cities (Laden et al., 2006), and Southern California
Children's Health (Gauderman et al., 2004) studies. These commenters
also highlighted the availability of additional long-term exposure
studies in this review, specifically a study of premature mortality in
older adults (Eftim et al., 2008) and the WHI study of cardiovascular
morbidity and mortality effects in women (Miller et al., 2007)
providing stronger evidence of mortality and morbidity effects
associated with long-term PM2.5 exposures at lower
concentrations than had previously been observed, including studies of
effects in at-risk populations. For example, some commenters asserted:

Evidence during the last review showed clearly that the annual
average standard needed to be much lower than the standard of 15
[micro]g/m\3\ that was first set in 1997. The evidence has only
grown since then. Multiple, multi-city studies over long periods of
time have shown clear evidence of premature death, cardiovascular
and respiratory harm as well as reproductive and developmental harm
at contemporary concentrations far below the level of the current
(annual) standard (ALA et al., 2012, p. 39; AHA et al., 2012, p.
10).

These commenters also highlighted the availability of a number of
short-term PM2.5 exposure studies as providing evidence of
mortality and morbidity effects at concentrations below the level of
the current 24-hour PM2.5 standard. Specifically, these
commenters made note of multi-city studies of premature mortality
(Zanobetti and Schwartz, 2009) and increased hospitalizations for
cardiovascular and respiratory-related effects in older adults (Bell et
al., 2008). These commenters also asserted the importance of many of
the single-city studies, arguing that these studies ``provide valuable
information regarding impacts on susceptible populations and on health
risk in areas with high peak to mean concentration ratios'' (ALA, et
al., 2012, p. 65). Collectively, considering the multi- and single-city
short-term exposure studies, these commenters asserted ``the record
clearly supports a more stringent 24-hour standard of 25 [micro]g/m\3\
to provide uniform protection in all regions of the country
particularly from short-term spikes in pollution and from the sub-daily
exposures that trigger heart attacks and strokes'' (ALA et al., 2012,
p. 62). A group of more than 350 physicians, environmental health
researchers, and public health and medical professionals argued,
``[s]tudies of short-term exposure demonstrate that PM2.5
air pollution increases the risk of hospital admissions for heart and
lung problems even when you exclude days with pollution concentrations
at or above the current daily standard of 35 [micro]g/m\3\. Daily
concentrations must be capped at lower levels to protect against peak
exposure days that occur due to local and seasonal sources of
emissions'' (Rom et al., 2012, p. 2).
In addition, many of these commenters generally concluded that
progress had been made in reducing many of the uncertainties identified
in the last review, in better understanding mechanisms by which
PM2.5 may be causing the observed health effects, and in
improving our understanding of at-risk populations. Further, a number
of commenters argued that by making the standards more protective, the
PM2.5 NAAQS would be more consistent with other existing
standards (e.g., California's annual average standard of 12 [micro]g/
m\3\) (CARB, 2012; CA OEHHA, 2012). Other commenters argued that
revising the primary PM2.5 standards would be more
consistent with the recommendations of the World Health Organization
(WHO) and/or Canada (e.g., ALA et al., 2012, p. 62; ISEE, 2012, p. 2;
MOE-Ontario, 2012, p. 1).
With regard to the scope of the literature reviewed for
PM2.5-related health effects, some commenters asserted that
the EPA inappropriately narrowed the scope of the review by excluding a
number of categories of relevant studies, specifically related to
studies of diesel pollution and traffic-related pollution (ALA, et al.,
2012, p. 17). These commenters argued that, based upon the exclusion of
these types of studies, the Integrated Science Assessment ``came to the
erroneous conclusion that the causal relationship between PM and cancer
is merely suggestive. This conclusion does not square with the
International Agency Research on Cancer (IARC) finding that diesel
emissions are a known human carcinogen nor with the conclusions of

[[Page 3111]]

the extended analyses of the [Harvard] Six Cities and ACS cohort
studies that report positive and statistically significant associations
between PM2.5 and lung cancer.'' Id.
Some of these commenters also noted the results of the EPA's
quantitative risk assessment, concluding that it showed that the risks
estimated to remain when the current standards are met are large and
important from a public health perspective and warrant increased
protection. For example, ALA et al., noted that the Risk Assessment
indicated the quantitative risk analyses likely underestimated
PM2.5-related mortality (U.S. EPA, 2010a, p. 5-16) and
argued that ``the measurements of risk should be treated
conservatively'' (ALA, et al., 2012, p. 73). These commenters also
summarized an expanded analysis of alternative PM2.5
standard levels that they argued documented the need for more
protective standards (McCubbin, 2011).
In general, all of these commenters agreed on the importance of
results from the large body of scientific studies reviewed in the
Integrated Science Assessment and on the need to revise the suite of
PM2.5 standards as articulated in the EPA's proposal, while
generally differing with the EPA's proposed judgments about the extent
to which the standards should be revised based on this evidence,
specifically for providing protection for at-risk populations.
The EPA generally agrees with these commenters' conclusion
regarding the need to revise the current suite of PM2.5
standards. The scientific evidence noted by these commenters was
generally the same as that assessed in the Integrated Science
Assessment and the Policy Assessment, and the EPA agrees that this
evidence provides a strong basis for concluding that the current
PM2.5 standards, taken together, are not requisite to
protect public health with an adequate margin of safety, and they need
to be revised to provide increased protection. For reasons discussed in
section III.E.4.c below, however, the EPA disagrees with aspects of
these commenters' views on the level of protection that is appropriate.
The EPA disagrees with these commenters' views that diesel exhaust
studies were excluded from the Integrated Science Assessment and were
not considered when making the causality determination for cancer,
mutagenicity, and genotoxicity. As discussed in section 7.5 of the
Integrated Science Assessment, diesel exhaust studies were integrated
within the broader body of scientific evidence that was considered in
reaching the causality determination for these health endpoints.
Additionally, as discussed in section 1.5.3 of the Integrated Science
Assessment, the evidence from diesel exhaust studies was also
considered as part of the collective evidence evaluated when making
determinations for other, noncancer health outcomes (e.g.,
cardiovascular and respiratory effects).\47\ Specifically, when
evaluating this evidence, the focus was on understanding the effects of
diesel exhaust particles.
---------------------------------------------------------------------------

\47\ In developing the second draft Integrated Science
Assessment, the EPA reexamined the controlled human exposure and
toxicological studies of fresh diesel and gasoline exhaust. This
information, in addition to other considerations, supported a change
in the causal determinations for ultrafine particles. Specifically,
in reevaluating the causal determinations for short-term ultrafine
particle exposures and cardiovascular and respiratory effects, the
EPA changed the classification from ``inadequate'' to ``suggestive''
for both categories of health outcomes (Vandenberg, 2009, p. 3).
CASAC agreed with the EPA's rationale for revising these causal
determinations (Samet, 2009f, p. 10).
---------------------------------------------------------------------------

It is important to recognize that the Integrated Science Assessment
focused on experimental studies of diesel exhaust that evaluated
exposures that were relevant to ambient concentrations, i.e., ``within
one or two orders of magnitude of ambient PM concentrations'' (U.S.
EPA, 2009a, section 1.3). The causal determination for cancer,
mutagenicity, and genotoxicity presented in the Integrated Science
Assessment represents an integration of experimental and observational
evidence of exposures to ambient PM concentrations. The EPA fully
considered the findings of studies that assessed these and other health
effects associated with exposure to diesel particles in reaching
causality determinations regarding health outcomes associated with
PM2.5 exposures. Furthermore, CASAC supported the EPA's
change to the causal determination for cancer and long-term
PM2.5 concentrations from ``inadequate'' to ``suggestive''
(Samet, 2009f, p. 2).
With regard to traffic studies, the EPA disagrees with the
commenters' views that traffic studies that focused on exposure
indicators such as distance to roadways should have been included in
the Integrated Science Assessment. These studies were excluded from
consideration because they did not measure ambient concentrations of
specific air pollutants, including PM2.5, but instead were
studies evaluating exposure to the undifferentiated ``traffic related
air pollution'' mixture (ALA et al., 2012, p. 17) (U.S. EPA, 2009a,
section 1.3). As a result, these studies do not add to the collective
body of evidence on the relationship between long- or short-term
exposure to ambient concentrations of PM2.5 and health
effects.
Some of these commenters also identified ``new'' studies that were
not included in the Integrated Science Assessment as providing further
support for the need to revise the primary PM2.5 standards.
As discussed in section II.B.3 above, the EPA notes that, as in past
NAAQS reviews, the Agency is basing the final decisions in this review
on the studies and related information included in the PM air quality
criteria that have undergone CASAC and public review and will consider
the ``new'' studies for purposes of decision making in the next PM
NAAQS review. Nonetheless, in provisionally evaluating commenters'
arguments (see Response to Comments document), the EPA notes that its
provisional assessment of ``new'' science found that such studies did
not materially change the conclusions in the Integrated Science
Assessment (U.S. EPA, 2012b).
Another group of commenters opposed revising the current
PM2.5 standards. These views were most extensively presented
in comments from the Utility Air Regulatory Group (UARG), representing
a group of electric generating companies and organizations and several
national trade associations; the American Petroleum Institute (API)
representing more than 500 oil and natural gas companies; the National
Association of Manufacturers (NAM), the American Chemistry Council
(ACC), the American Fuel & Petroleum Manufacturers (AFPM), the Alliance
of Automobile Manufacturers (AAM), and other manufacturing
associations; the Electric Power Research Institute (EPRI); and the
Texas Commission on Environmental Quality (Texas CEQ). These commenters
generally mentioned many of the same studies that were cited by the
commenters who supported revising the standards, as well as other
studies, but highlighted different aspects of these studies in reaching
substantially different conclusions about their strength and the extent
to which progress has been made in reducing uncertainties in the
evidence since the last review. Furthermore, they asserted that the
evidence that has become available since the last review does not
establish a more certain risk or a risk of effects that are
significantly different in character to those that provided a basis for
the current standards, nor does the evidence demonstrate that the risk
to public health upon attainment of the current standards would be
greater than was

[[Page 3112]]

understood when the EPA established the current standards in 2006.
These commenters generally expressed the view that the current
standards provide the requisite degree of public health protection. In
supporting their view, these commenters generally argued that the EPA's
conclusions are inconsistent with the current state of the science and
questioned the underlying scientific evidence including the causal
determinations reached in the Integrated Science Assessment. More
specifically, this group of commenters argued that: (1) The EPA did not
apply its framework for causal determination consistently across
studies or health outcomes and, in the process, the EPA relied on a
selective group of long- and short-term exposure studies to reach
conclusions regarding causality; (2) toxicological and controlled human
exposure studies do not provide supportive evidence that the health
effects observed in epidemiological studies are biologically plausible;
(3) uncertainties in the underlying health science are as great or
greater than in 2006; (4) there is no evidence of greater risk since
the last review to justify tightening the current annual
PM2.5 standard; and (5) ``new'' studies not included in the
Integrated Science Assessment continue to increase uncertainty about
possible health risks associated with exposure to PM2.5.
These comments are discussed in turn below.
(l) Some of these commenters asserted that the EPA did not apply
its framework for causal determinations consistently across studies or
health outcomes (e.g., ACC, 2012, Attachment A, pp. 1 to 2; API, 2012,
Attachment 1, p. 30; NAM et al., 2012, pp. 22 to 25; Texas CEQ, 2012,
pp 2 to 3).\48\ These commenters argued that the EPA downplayed
epidemiological studies with null or inconsistent results,
inappropriately used the Hill criteria when evaluating the
epidemiological evidence, and used the same study and the same
underlying database to conclude that there was a causal association
between mortality and multiple criteria pollutants.
---------------------------------------------------------------------------

\48\ The EPA notes that the same concerns about the causal
determinations presented in the Integrated Science Assessment were
raised in comments to CASAC on the draft Integrated Science
Assessments (e.g., UARG, 2009; API, 2009; ACC, 2012, Appendix B).
CASAC, therefore, had the opportunity to consider these comments in
reaching consensus conclusions on this issue.
---------------------------------------------------------------------------

The EPA disagrees with these commenters' views. First, the EPA
recognizes that the evaluation of the scientific evidence and its
application of the causal framework used in the current PM NAAQS review
was the subject of exhaustive and detailed review by CASAC and the
public. As summarized in section II.B.3 above, prior to finalizing the
Integrated Science Assessment, two drafts were released for CASAC and
public review to evaluate the scientific integrity of the documents.
Evidence related to the substantive issues raised by CASAC and public
commenters with regard to the content of the first and second draft
Integrated Science Assessments were discussed at length during these
public CASAC meetings and considered in developing the final Integrated
Science Assessment. CASAC supported the development of the EPA's
causality framework and its use in the current PM NAAQS review and
concluded:

The five-level classification of strength of evidence for causal
inference has been systematically applied; this approach has
provided transparency and a clear statement of the level of
confidence with regard to causation, and we recommend its continued
use in future Integrated Science Assessments (Samet 2009f, p. 1).

These commenters asserted that during the application of the causal
framework the EPA inappropriately relied on a selective group of long-
and short-term exposure studies in reaching causal inferences (API,
2012, pp 12 to 17; ACC, 2012, Attachment A, pp 1 to 2; NAM et al.,
2012, pp. 22 to 25; Texas CEQ, 2012, pp 2 to 3). Additionally, these
commenters expressed the view that the EPA focused on a subset of
epidemiological studies that reported positive and statistically
significant results while ignoring other studies, especially those that
reported no statistically significant associations, those that reported
potential thresholds, or those that highlighted uncertainties and
limitations in study design or results. Furthermore, some of these
commenters argued that epidemiological studies are observational in
nature and cannot provide evidence of a causal association.
The EPA disagrees with these commenters' views on assessing the
health effects evidence and on the conclusions regarding the causality
determinations reached in the Integrated Science Assessment. In
conducting a comprehensive evaluation of the evidence in the Integrated
Science Assessment, the EPA recognized the distinction between the
evaluation of the relative scientific quality of individual study
results and the evaluation of the pattern of results within the broader
body of scientific evidence and considered both in reaching causality
determinations. The more detailed characterizations of individual
studies included an assessment of the quality of the study based on
specific criteria as described in the Integrated Science Assessment
(U.S. EPA, 2009a, section 1.5.3).
In developing an integrated assessment of the health effects
evidence for PM, the EPA emphasized the importance of examining the
pattern of results across various studies and did not focus solely on
statistical significance \49\ as a criterion of study strength. This
approach is consistent with views clearly articulated throughout the
epidemiological and causal inference literature, specifically, that it
is important not to focus on results of statistical tests to the
exclusion of other information.\50\ The concepts underlying the EPA's
approach to evaluating statistical associations have been discussed in
numerous publications, including a report by the U.S. Surgeon General
on the health consequences of smoking (Centers for Disease Control and
Prevention, 2004). This report cautions against over-reliance on
statistical significance in evaluating the overall evidence for an
exposure-response relationship. Criteria characterized by Hill (1965)
also addressed the value, or lack thereof, of statistical tests in the
determination of cause:
---------------------------------------------------------------------------

\49\ Statistical significance is an indicator of the precision
of a study's results, which is influenced by a variety of factors
including, but not limited to, the size of the study, exposure and
measurement error, and statistical model specifications. Studies
typically calculate ``p-values'' to determine whether the study
results are statistically significant or whether the study results
are likely to occur simply by chance. In general practice, effects
are considered statistically significant if p values are less than
0.05.
\50\ For example, Rothman (1998) stated, ``Many data analysts
appear to remain oblivious to the qualitative nature of significance
testing [and that] * * * statistical significance is itself only a
dichotomous indicator. As it has only two values, significant or not
significant * * *. Nevertheless, p-values still confound effect size
with study size, the two components of estimation that we believe
need to be reported separately.'' As a result, Rothman recommended
that p-values be omitted as long as point and interval estimates are
available.

No formal tests of significance can answer those [causal]
questions. Such tests can, and should, remind us of the effects the
play of chance can create, and they will instruct us in the likely
magnitude of those effects. Beyond that, they contribute nothing to
---------------------------------------------------------------------------
the `proof' of our hypothesis (Hill, 1965, p. 299).

The statistical significance of individual study findings has
played an important role in the EPA's evaluation of the study's results
and the EPA has placed greater emphasis on studies reporting
statistically significant results. However, in the broader evaluation
of the evidence from many

[[Page 3113]]

epidemiological studies, and subsequently during the process of forming
causality determinations in integrating evidence across
epidemiological, controlled human exposure, and toxicological studies,
the EPA has emphasized the pattern of results across epidemiological
studies, and whether the effects observed were coherent across the
scientific disciplines for drawing conclusions on the relationship
between PM2.5 and different health outcomes. Thus, the EPA
did not limit its focus or consideration to just studies that reported
positive associations or where the results were statistically
significant.
In addition, some commenters asserted that the EPA inappropriately
used the Hill criteria by failing to consider the limitations of
studies with weak associations, thereby overstating the consistency of
the observed associations (API, 2012, Attachment 1, pp. 30 to 35).
These commenters argued that risk estimates greater than 3 to 4 reflect
strong associations supportive of a causal link, while smaller risk
estimates (i.e., 1.5 to 3) are considered to be weak and require other
lines of evidence to demonstrate causality.
As discussed in section 1.5.3 of the Integrated Science Assessment,
the EPA thoroughly considered the limitations of all studies during its
evaluation of the scientific literature (U.S. EPA,, 2009a, p. 1-14).
This collective body of evidence, including known uncertainties and
limitations of the studies evaluated, were considered during the
process of forming causality determinations as discussed in chapters 6
and 7 of the Integrated Science Assessment. For example, the EPA
concluded that ``a causal relationship exists between short-term
PM2.5 exposure and cardiovascular effects,'' however, in
reaching this conclusion, the Agency recognized and considered
limitations of the current evidence that still requires further
examination (U.S. EPA, 2009a., in section 6.2.12.1). Therefore, the EPA
disagrees with these commenters' views that the Hill criteria were
inappropriately used in that the limitations of studies were not
considered.
The EPA also disagrees with the commenters' assertion that the
magnitude of the association must be large to support a determination
of causality. As discussed in the Integrated Science Assessment, the
strength of the observed association is an important aspect to aid in
judging causality and ``while large effects support causality, modest
effects therefore do not preclude it'' (U.S. EPA, 2009a, Table 1-2,
section 1.5.4). The weight of evidence approach used by the EPA
encompasses a multitude of factors of which the magnitude of the
association is only one component (U.S. EPA, 2009a, Table 1-3). An
evaluation of the association across multiple investigators and
locations supports the ``reproducibility of findings [which]
constitutes one of the strongest arguments for causality'' (U.S. EPA,
2009a, Table 1-2). Even though the risk estimates for air pollution
studies may be modest, the associations are consistent across hundreds
of studies as demonstrated in the Integrated Science Assessment.
Furthermore, the causality determinations rely on different lines of
evidence, by integrating evidence across disciplines, including animal
toxicological studies and controlled human exposure studies.
Furthermore, as summarized in section III.B above and discussed
more fully in section III.B.3 of the proposal, the EPA recognizes that
the population potentially affected by PM2.5 is
considerable, including large subgroups of the U.S. population that
have been identified as at-risk populations (e.g., children, older
adults, persons with underlying cardiovascular or respiratory disease).
While individual effect estimates from epidemiological studies may be
modest in size, the public health impact of the mortality and morbidity
associations can be quite large given that air pollution is ubiquitous.
Indeed, with the large population exposed, exposure to a pollutant
causally associated at a population level with mortality and serious
illness has significant public health consequences, virtually
regardless of the relative risk. Taken together, this information
indicates that exposure to ambient PM2.5 concentrations has
substantial public health impacts.
In addition, these commenters believed that the EPA downplayed null
or inconsistent findings in numerous long-term mortality studies with
reported PM2.5 concentrations above and below the level of
the current annual standard. The EPA disagrees that studies with null
or inconsistent findings were not accurately presented and considered
in the Integrated Science Assessment. For example, as discussed
throughout section 7.6 and depicted in Figures 7-6 and 7-7 of the
Integrated Science Assessment, the EPA presented the collective
evidence from all studies that examined the association between long-
term PM2.5 exposure and mortality. Overall, across these
studies there was evidence of consistent positive associations in
different cohorts. That evidence, in combination with the biological
plausibility provided by experimental and toxicological studies
evaluated in sections 7.1 and 7.2 of the Integrated Science Assessment,
supported a causal relationship exists between long-term
PM2.5 exposure and mortality.
Lastly, some of these commenters argued that in some cases, the EPA
used the same study and the same underlying database to conclude that
there is a causal association between mortality and multiple criteria
pollutants. These commenters argued, ``[i]n doing so, EPA attributes
the cause of the mortality effects observed to whichever criteria
pollutant it is reviewing at the time'' (API, 2012, pp. 14 to 16).
The EPA strongly disagrees that the Agency ``attributes the cause
of mortality effects observed to whichever criteria pollutant it is
reviewing at the time.'' The EPA consistently recognizes that other
pollutants are also associated with health outcomes, as is reflected in
the fact that the EPA has established regulations to limit emissions of
particulate criteria pollutants as well as other gaseous criteria
pollutants. Epidemiological studies often examine the association
between short- and long-term exposures to multiple air pollutants and
mortality within a common dataset in an attempt to identify the air
pollutant(s) of the complex mixture most strongly associated with
mortality. In evaluating these studies, the EPA employs specific study
selection criteria to identify those studies most relevant to the
review of the NAAQS. In its assessment of the health evidence regarding
PM2.5, the EPA has carefully evaluated the potential for
confounding, effect measure modification, and the role of
PM2.5 as a component of a complex mixture of air pollutants
(U.S. EPA, 2009a, p. 1-9). The EPA used a rigorous weight of evidence
approach to inform causality that evaluated consistency across studies
within a discipline, evidence for coherence across disciplines, and
biological plausibility. Additionally, during this process, the EPA
assessed the limitations of each study in the context of the collective
body of evidence. It was the collective evidence, not one individual
study that ultimately determined whether a causal relationship exists
between a pollutant and health outcome. In the Integrated Science
Assessment, the combination of epidemiological and experimental
evidence formed the basis for the Agency concluding for the first time
that a causal relationship exists between short- or long-term exposure
to a criteria pollutant and mortality (U.S. EPA, 2009, sections 2.3.1.1
and 2.3.1.2).

[[Page 3114]]

Additionally, while the EPA has evaluated some of the studies used
to inform the causality determination for PM in the Integrated Science
Assessments for other criteria air pollutants, the Agency has done so
in the context of examining the collective body of evidence for each of
the respective criteria air pollutants. As such, the body of evidence
to inform causality has varied from pollutant to pollutant resulting in
the association between each criteria air pollutant and mortality being
classified at a different level of the five-level hierarchy used to
inform causation (e.g., U.S. EPA, 2008e, U.S. EPA, 2008f, U.S. EPA,
2010k).
The EPA notes that the final causality determinations presented in
the Integrated Science Assessment reflected CASAC's recommendations on
the second draft Integrated Science Assessment (Samet, 2009f, pp. 2 to
3). Specifically, CASAC supported the EPA's changes (in the second
versus first draft Integrated Science Assessment) from ``likely
causal'' to ``causal'' for long-term exposure to PM2.5 and
cardiovascular effects and for cancer and PM2.5 (from
``inadequate'' to ``suggestive''). Id. Furthermore, CASAC recommended
``upgrading'' the causal classification for PM2.5 and total
mortality to ``causal'' for both the short- and long-term timeframes.
Id. With regard to mortality, the ``EPA carefully reevaluated the body
of evidence, including the collective evidence for biological
plausibility for mortality effects, and determined that a causal
relationship exists for short- and long-term exposure to
PM2.5 and mortality, consistent with the CASAC comments''
(Jackson, 2010).
(2) With regard to toxicological and controlled human exposure
studies, these commenters argued that the available evidence does not
provide coherence or biological plausibility for health effects
observed in epidemiological studies (API, 2012, pp. 21 to 22,
Attachment 1, pp. 25 to 29; AAM, 2012, pp. 15 to 16; Texas CEQ, 2012,
p. 3). With regard to the issue of mechanisms, these commenters noted
that although the EPA recognizes that new evidence is now available on
potential mechanisms and plausible biological pathways, the evidence
provided by toxicological and controlled human exposure studies still
does not resolve all questions about how PM2.5 at ambient
concentrations could produce the mortality and morbidity effects
observed in epidemiological studies. More specifically, for example,
some of these commenters argued that:

A review of the Integrated Science Assessment, however, suggests
that the experimental evidence is inconsistent and not coherent with
findings in epidemiology studies. Specifically, the findings of mild
and reversible effects in most experimental studies conducted at
elevated exposures are not consistent with the more serious
associations described in epidemiology studies (e.g., hospital
admissions and mortality). Also, both animal studies and controlled
human exposure studies have identified no effect levels for acute
and chronic exposure to PM and PM constituents at concentrations
considerably above ambient levels. EPA should consider the
experimental findings in light of these higher exposure levels and
what the relevance may be for ambient exposures (API, 2012,
Attachment 1, p. 25).

The EPA notes that in the review completed in 1997, the Agency
considered the lack of demonstrated biological mechanisms for the
varying effects observed in epidemiological studies to be an important
caution in its integrated assessment of the health evidence upon which
the standards were based (71 FR 61157, October 17, 2006). In the review
completed in 2006, the EPA recognized the findings from additional
research that indicated that different health responses were linked
with different particle characteristics and that both individual
components and complex particle mixtures appeared to be responsible for
many biologic responses relevant to fine particle exposures. Id. Since
that review, there has been a great deal of research directed toward
advancing our understanding of biologic mechanisms. While this research
has not resolved all questions, and further research is warranted (U.S.
EPA, 2011a, section 2.5), it has provided important insights as
discussed in section III.B.1 of the proposal (77 FR at 38906 to 38909)
and discussed more fully in the Integrated Science Assessment (U.S.
EPA, 2009a, Chapter 5).
As noted in the proposal, toxicological studies provide evidence to
support the biological plausibility of cardiovascular and respiratory
effects associated with long- and short-term PM exposures observed in
epidemiological studies (77 FR 38906) and provide supportive
mechanistic evidence that the cardiovascular morbidity effects observed
in long-term exposure epidemiological studies are coherent with studies
of cardiovascular-related mortality (77 FR 38907). The Integrated
Science Assessment concluded that the new evidence available in this
review ``greatly expands'' upon the evidence available in the last
review ``particularly in providing greater understanding of the
underlying mechanisms for PM2.5 induced cardiovascular and
respiratory effects for both short- and long-term exposures'' (U.S.
EPA, 2009a, p. 2-17). The mechanistic evidence now available, taken
together with newly available epidemiological evidence, increases the
Agency's confidence that a causal relationship exists between long- and
short-term exposure to PM2.5 and cardiovascular effects and
mortality.\51\ In addition, CASAC supported the Integrated Science
Assessment approach and characterization of potential mechanisms or
modes of action (Samet, 2009e, pp. 7 to 8; Samet, 2009f, p. 11), as
well as the findings of a causal relationship at the population level
between exposure to PM2.5 and mortality and cardiovascular
effects (Samet, 2009f, pp. 2 to 3).\52\
---------------------------------------------------------------------------

\51\ See American Trucking Associations v. EPA, 175 F. 3d 1027,
1055-56 (DC Cir. 1999) reversed in part and affirmed in part sub
nom, Whitman v. American Trucking Associations, 531 U.S. 457 (2001)
holding that the EPA could establish NAAQS without identifying a
biological mechanism (``To begin with, the statute itself requires
no such proof. The Administrator may regulate air pollutants
``emissions of which, in his judgment, cause or contribute to air
pollution which may reasonably be anticipated to endanger public
health or welfare.'' (emphasis added by the court). Moreover, this
court has never required the type of explanation petitioners seek
from EPA. In fact, we have expressly held that EPA's decision to
adopt and set air quality standards need only be based on
`reasonable extrapolations from some reliable evidence'* * *.
Indeed, were we to accept petitioners' view, EPA (or any agency for
that matter) would be powerless to act whenever it first recognizes
clear trends of mortality or morbidity in areas dominated by a
particular pathogen.'').
---------------------------------------------------------------------------

Additionally, the EPA disagrees with commenters that the mild and
reversible effects observed in controlled human exposure studies are
inconsistent with the more serious effects observed in epidemiological
studies. Ethical considerations regarding the types of studies that can
be performed with human subjects generally limit the effects that can
be evaluated to those that are transient, reversible, and of limited
short-term consequence. The relatively small number of subjects
recruited for controlled exposure studies should also be expected to
have less variability in health status and risk factors than occurring
in the general population.\53\ Consequently, the severity

[[Page 3115]]

of health effects observed in controlled human exposure studies
evaluating the effects of PM should be expected to be less than
observed in epidemiologic studies. Nonetheless, that effects are
observed in relatively healthy individuals participating in controlled
exposure studies serves as an indicator that PM is initiating health
responses and that more severe responses may reasonably be expected in
a more diverse population.
---------------------------------------------------------------------------

\53\ For example, the EPA excludes from its controlled human
exposure studies involving exposure to PM2.5 any
individual with a significant risk factor for experiencing adverse
effects from such exposure. Thus, the EPA excludes a priori the
following categories of persons: those with a history of angina,
cardiac arrhythmias, and ischemic myocardial infarction or coronary
bypass surgery; those with a cardiac pacemaker; those with
uncontrolled hypertension (greater than 150 systolic and 90
diastolic); those with neurogenetive diseases; those with a history
of bleeding diathesis; those taking beta-blockers; those using oral
anticoagulants; those who are pregnant, attempting to become
pregnant, or breastfeeding; those who have experienced a respiratory
infection within four weeks of exposure; those experiencing eye or
abdominal surgery within six weeks of exposure; those with active
allergies; those with a history of chronic illnesses such as
diabetes, cancer, rheumatologic diseases, immunodeficiency state,
known cardiovascular disease, or chronic respiratory diseases;
smokers. The EPA ``Application for Independent Review Board Approval
of Human Subjects Research: Cardiopulmonary Effects of healthy Older
GSTM1 Null and Sufficient individuals to Concentrated Ambient Air
Particles (CAPTAIN)'', Nov. 9, 2011, p. 9.
---------------------------------------------------------------------------

It should also be noted that there is a small body of toxicological
evidence demonstrating mortality in rodents exposed to PM (e.g.,
Killingsworth et al. 1997). Overall it is not surprising that lethality
is not induced in more toxicological research, as these types of
studies do not readily lend themselves to this endpoint.
Epidemiological studies have observed associations between PM and
mortality in communities with populations in the range of many
thousands to millions of people. Clearly, it is not feasible to expose
hundreds (if not thousands) of animals to ambient PM (potentially over
many years) in a laboratory setting to induce enough lethalities to
distinguish between natural deaths and those attributable to PM.
Furthermore, the heterogeneous human populations sampled in
epidemiological studies are comprised of individuals with different
physical, genetic, health, and socioeconomic backgrounds which may
impact the outcome. However, in toxicological studies, the rodent
groups are typically inbred, such that inter-individual variability is
minimized. Thus, if the rodent strain used is quite robust, PM-induced
effects may not be observed at low exposure concentrations.
(3) In asserting that the uncertainties in the underlying health
science are as great or greater than in the last review and therefore
do not support revision to the standards at this time, commenters in
this group variously discussed a number of issues related to: (a)
Confounding, (b) heterogeneity in risk estimates, (c) exposure
measurement error, (d) model specification, (e) the shape of the
concentration-response relationship, and (f) understanding the relative
toxicity of components within the mixture of fine particles. Each of
these issues is addressed below and some are discussed in more detail
in the Response to Comments document.
In summary, these commenters concluded that the substantial
uncertainties present in the last review have not been resolved and/or
that the uncertainty about the possible health risks associated with
PM2.5 exposure has not diminished. As discussed below, the
EPA believes that the overall uncertainty about possible health risks
associated with both long- and short-term PM2.5 exposure has
diminished to an important degree since the last review. While the EPA
agrees that important uncertainties remain, and that future research
directed toward addressing these uncertainties is warranted, the EPA
disagrees with commenters' views that the remaining uncertainties in
the scientific evidence are too great to warrant revising the current
PM2.5 NAAQS.
(a) Confounding
Some commenters have criticized the EPA for not adequately
addressing the issue of confounding in both long- and short-term
exposure studies of mortality and morbidity. This includes confounding
due to copollutants, as well as unmeasured confounding.\54\
---------------------------------------------------------------------------

\54\ The Integrated Science Assessment defines confounding as
``a confusion of effects. Specifically, the apparent effect of the
exposure of interest is distorted because the effect of an
extraneous factor is mistaken for or mixed with the actual exposure
effect (which may be null) (Rothman and Greenland, 1998)'' (U.S.
EPA, 2009a, p. 1-16). Epidemiological analyses attempt to adjust or
control for these characteristics (i.e., potential confounders) that
differ between exposed and non-exposed individuals (U.S. EPA, 2009a,
section 1.5.3). Not all risk factors can be controlled for within a
study design/model and are termed ``unmeasured confounders.'' An
unmeasured confounder is a confounder that has not previously been
measured and therefore is not included in the study design/model.
---------------------------------------------------------------------------

With regard to copollutant confounding, these commenters asserted
that the EPA has not adequately interpreted the results from studies
that examined the effect of copollutants on the relationship between
long- and short-term PM2.5 exposures and mortality and
morbidity outcomes. These commenters contend that the EPA has
inappropriately concluded that PM2.5-related mortality and
morbidity associations are generally robust to confounding. The
commenters stated that statistically significant PM2.5
associations in single-pollutant models in epidemiological studies do
not remain statistically significant in copollutant models.
The loss of statistical significance or the reduction in the
magnitude of the effect estimate when a co-pollutant model is used may
be the result of factors other than confounding. These changes do not
prove either the existence or absence of confounding. These impacts
must be evaluated in a broader context that considers the entire body
of evidence. The broader examination of this issue in the Integrated
Science Assessment included a focus on evaluating the stability of the
size of the effect estimates in epidemiological studies conducted by a
number of research groups using single- and copollutant models (U.S.
EPA, 2009a, sections 6.2.10.9, 6.3.8.5, and 6.5, Figures 6-5, 6-9, and
6-15). This examination found that, for most epidemiological studies,
there was little change in effect estimates based on single- and
copollutant models, although the Integrated Science Assessment
recognized that in some cases, the PM2.5 effect estimates
were markedly reduced in size and lost statistical significance.
Additionally, the EPA notes that these comments do not adequately
reflect the complexities inherent in assessing the issue of copollutant
confounding. As discussed in the proposal (77 FR 38907, 38909, and
38910) and more fully in the Integrated Science Assessment (U.S.EPA,
2009a, sections 6.2, 6.3, and 6.5), although copollutant models may be
useful tools for assessing whether gaseous copollutants may be
potential confounders, such models alone cannot determine whether
copollutants are in fact confounders. Interpretation of the results of
copollutant models is complicated by correlations that often exist
among air pollutants, by the fact that some pollutants play a role in
the atmospheric reactions that form other pollutants such as secondary
fine particles, and by the statistical power of the studies in question
inherent in the study methodology. For example, the every-third or
sixth-day sampling schedule often employed for PM2.5
measurements compared to daily measurements of gaseous copollutants
drastically reduces the overall sample size to assess the effect of
copollutants on the PM2.5-morbidity or mortality
relationship, such that the reduced sample size can lead to less
precise effect estimates (e.g., wider confidence intervals).
The EPA recognizes that when PM2.5 is correlated with
gaseous pollutants it can be difficult to identify the effect of
individual pollutants in the ambient mixture (77 FR 38910). However,
based on the available evidence, the EPA

[[Page 3116]]

concludes epidemiological studies continue to support the conclusion
that PM2.5 associations with mortality and morbidity
outcomes are robust to the inclusion of gaseous copollutants in
statistical models. The EPA evaluated the potential confounding effects
of gaseous copollutants and, although it is recognized that
uncertainties and limitations still remain, the Agency concluded the
collective body of scientific evidence is ``stronger and more
consistent than in previous reviews providing a strong basis for
decision making in this review'' (77 FR 38910/1).
Several commenters offered detailed comments on the long-term
PM2.5 exposure studies arguing that associations from
mortality studies are subjected to unmeasured confounding and as a
result are not appropriately characterized as providing evidence of a
causal relationship between long-term PM2.5 exposure and
mortality (e.g., UARG, 2012, pp. 10 to 11, Attachment A, pp. 17 to 23;
API, 2012, pp. 13 to 14, Attachment 1, pp. 11 to 14, Attachment 7, pp.
2-10; ACC, 2012, p. 18 to 21; AFPM, 2012, p. 8; Texas CEQ, 2012, p. 4).
Specifically, commenters cited two studies (i.e., Janes et al., 2007
and Greven et al., 2011) that used a new type of statistical analysis
to examine associations between annual (long-term) and monthly (sub-
chronic) PM2.5 exposure and mortality. The commenters
interpreted the results of these analyses as evidence of unmeasured
confounding in the long-term PM2.5 exposure-mortality
relationship. These commenters interpreted these studies as raising
fundamental questions regarding the EPA's determination that a causal
relationship exists between long-term PM2.5 exposure and
mortality. In addition to the commenters mentioned above, all of the
authors of the publications by Janes et al. (2007) and Greven et al.
(2011) (i.e., Francesca Dominici, Scott Zeger, Holly Janes, and Sonja
Greven) submitted a joint comment to the public docket in order to
clarify specific points regarding these two studies (Dominici et al.,
2012).
The first study, Janes et al. (2007), was evaluated in the
Integrated Science Assessment (U.S. EPA, 2009a, p. 7-88). The second
study, Greven et al. (2011), an extension of the Janes et al. (2007)
study adding three more years of data, is a ``new'' study discussed in
the Provisional Science Assessment (U.S. EPA, 2012). Both studies used
nationwide Medicare mortality data to examine the association between
monthly average PM2.5 concentrations over the preceding 12
months and monthly mortality rates in 113 U.S. counties and examined
whether community-specific trends in monthly PM2.5
concentrations and mortality declined at the same rate as the national
rate. The investigators examined this by decomposing the association
between PM2.5 and mortality into two components: (1)
National trends, defined as the association between the national
average trend in monthly PM2.5 concentrations averaged over
the previous 12 months and the national average trend in monthly
mortality rates, and (2) local trends, defined as county-specific
deviations in monthly PM2.5 concentrations and monthly
mortality rates from national trends.
The EPA does not question the results of the national trends
analyses conducted by Janes et al. (2007) and Greven et al. (2011).\55\
Both Janes et al. (2007) and Greven et al. (2011) observed positive and
statistically significant associations between long-term exposure to
PM2.5 and mortality in their national analyses. However,
Janes et al. (2007) and Greven et al. (2011) eliminated all of the
spatial variation in air pollution and mortality in their data set when
estimating the national effect, focusing instead on both chronic
(yearly) and sub-chronic (monthly) temporal differences in the data
(Dominici et al. 2012). Janes et al. (2007) (Table 1) highlighted that
over 90 percent of the variance in the data set used for the analyses
conducted by both Janes et al. (2007) and Greven et al. (2011) was
attributable to spatial variability, which the authors chose to
discard. As noted above, the focus of the analyses by Janes et al.
(2007) and Greven et al. (2011) was on two components: (1) A temporal
or time component, i.e., the ``national'' trends analysis, which
examined the association between the national average trend in monthly
PM2.5 concentrations averaged over the previous 12 months
and the national average trend in monthly mortality rates and (2) a
space-by-time component, i.e., the ``local'' trends analysis, which
examined county-specific deviations in monthly PM2.5
concentrations and monthly mortality rates from national trends. These
two components combined comprised less than 10 percent of the variance
in the data set. The authors included a focus on the space-by-time
component, which represented approximately 5 percent of the variance in
the data set, in an attempt to identify, absent confounding, if
PM2.5 was associated with mortality at this unique exposure
window. Thus, these studies are not directly comparable to other cohort
studies investigating the relationship between long-term exposure to
PM2.5 and mortality, which make use of spatial variability
in air pollution and mortality data.\56\ This point was highlighted by
the study authors who stated that ``when one considers that this wealth
of information is not accounted for in [Janes 2007], it is not as
surprising that * * * vastly different estimates of the
PM2.5/mortality relationship [were observed] than in other
studies that do exploit that variability'' (Dominici et al., 2012, p.
2).
---------------------------------------------------------------------------

\55\ In its evaluation of Janes et al. (2007) in the Integrated
Science Assessment, the EPA did not identify limitations in the
statistical methods used per se (U.S. EPA, 2009a, p. 7-88) and
included the results of the national-scale analyses in that study in
the body of evidence that supported the determination that there is
a causal relationship between long-term PM2.5 exposure
and mortality.
\56\ Though not directly comparable, the national effect
estimates for mortality reported by Janes et al. (2007) and Greven
et al. (2011) are coincidentally similar in magnitude to those
previously reported. It is important to note that previous cohort
studies have focused on identifying spatial differences in
PM2.5 concentrations between cities, while Janes et al.
(2007) and Greven et al. (2011) focus primarily on temporal
differences in PM2.5 concentrations. In fact, Greven et
al. (2011) state ``We do not focus here on a third type [of
statistical approach] used in cohort studies, measuring the
association between average PM2.5 levels and average age-
adjusted mortality rates across cities (purely spatial or cross-
sectional association).''
---------------------------------------------------------------------------

The EPA notes that the results of the local trends analyses
conducted by Janes et al. (2007) and Greven et al. (2011) are limited
by the monthly timescale used in these analyses. This view is
consistent with comments on the Janes et al. (2007) study articulated
in Pope and Burnett (2007),\57\ which noted that an important
limitation of the local scale analysis conducted by Janes et al. (2007)
and subsequently by Greven et al. (2011) was the subchronic exposure
window considered in these analyses. Both studies used annual average
PM2.5 concentrations to characterize long-term national
trends which was consistent with exposure windows considered in other
studies of long-term exposure to PM2.5 and mortality.\58\
However, the local scale analyses used monthly average PM2.5
concentrations to characterize county-specific deviations from national
trends (the local scale). The use of monthly average data likely does
not provide

[[Page 3117]]

enough exposure contrast to observe temporal changes in mortality at
the local scale. It also represents a different exposure window than
considered in the large body of evidence of health effects related to
short-term (from less than one day to up to several days) and chronic
(one or more years) measures of PM2.5.
---------------------------------------------------------------------------

\57\ Some commenters argued that there were flaws in the
criticisms offered by Pope and Burnett (2007) on the paper by Janes
et al. (2007) (UARG, 2012, Attachment A, pp. 19 to 23). The EPA
responds to each of these specific comments in the Response to
Comments document.
\58\ As noted above, however, Janes et al. (2007) and Greven et
al. (2011) focused on temporal variability and other studies of
long-term exposure to PM2.5 and mortality focus on
spatial variability.
---------------------------------------------------------------------------

Furthermore, the EPA disagrees with commenters that studies by
Janes et al. (2007) and Greven et al. (2011) provide evidence that
other studies of long-term exposure to PM2.5 and mortality
are affected by unmeasured confounding. As noted above, the design of
the studies conducted by Janes et al. (2007) and Greven et al. (2011)
are fundamentally different than those used in other studies of long-
term exposure to PM2.5 and mortality, including the ACS
cohort and the Harvard Six Cities study. Studies, such as the ACS and
Harvard Six Cities studies, used the spatial variation between cities
to measure the effect of long-term (annual) exposures to
PM2.5 on mortality risk, and did not conduct any analyses
relying on the temporal variation in PM2.5. The opposite is
true of the Janes et al. (2007) and Greven et al. (2011) studies which
first removed the spatial variability in PM2.5 and then
examined the temporal variation at both the national and local scale to
measure the effects of temporal differences in PM2.5 on
mortality risk. Janes et al. (2007) and Greven et al. (2011) focus on
changes in PM2.5 concentrations over time and, therefore,
control for confounders would be based on including variables that vary
over time rather than over space. As a result, any evidence of
potential confounding of the PM2.5-mortality risk
relationship derived from Janes et al. (2007) and Greven et al. (2011)
cannot be extrapolated to draw conclusions related to potential spatial
confounding in studies based on the spatial variation in
PM2.5 concentrations.
As detailed in the Integrated Science Assessment (U.S. EPA, 2009a,
section 7.6), and recognized by the authors of Janes et al. (2007) and
Greven et al. (2011), the cohort studies that informed the causality
determination for long-term PM2.5 exposure and mortality
``have developed approaches to adjust for measured and unmeasured
confounders'' (Dominici et al., 2012, p. 2). These approaches were
specifically designed to adjust for spatial confounding. The hypothesis
that the authors of Janes et al. (2007) and Greven et al. (2011) chose
to examine was that differences in the local and national effects
indicated unmeasured temporal confounding in either the local or
national effect estimate. This hypothesis was specific to these two
studies that examined temporal variability in exposure to air pollution
and did not include known potential confounders at either the national
or local scale as time-varying covariates in the statistical model. The
authors acknowledged that the interpretation of either the national or
local estimates needs to occur with an appreciation of the potential
confounding effects of national and local scale covariates that were
omitted from the model (Dominici et al., 2012).
It is important to recognize that because Janes et al. (2007) and
Greven et al. (2011) focused on variations in PM2.5 over
time and not space, the results from these two studies do not provide
any indication that other studies of long-term exposure to
PM2.5 and mortality exhibit spatial confounding, or that
PM2.5 does not cause mortality.\59\ The authors of Janes et
al. (2007) and Greven et al. (2011) recognized that ``it is entirely
possible that these papers are looking for an association at a
timescale for which no association truly exists'' (Dominici et al.,
2012, p. 3). Furthermore, as highlighted in the Integrated Science
Assessment and discussed by Pope and Burnett (2007), the conclusions of
Janes et al. (2007) ``are overstated * * * [T]heir analysis tells us
little or nothing about unmeasured confounding in those and related
studies because the methodology of Janes et al. largely excludes the
sources of variability that are exploited in those other studies. By
using monthly mortality counts and lagged 12-month average pollution
concentrations, the authors eliminate the opportunity to exploit short-
term or day-to-day variability.''
---------------------------------------------------------------------------

\59\ Further, the EPA notes that Janes et al. (2007) and Greven
et al. (2011) provide no information relevant to examining
confounding in studies of short-term exposure to PM2.5.
---------------------------------------------------------------------------

In conclusion, the EPA interprets the results of the analyses
conducted by Janes et al. (2007) and Greven et al. (2011) as being
consistent with prior knowledge of examining associations with long-
term exposure to PM2.5 at the national scale using long-term
average PM2.5 concentrations. For the reasons presented
above and discussed in more detail in the Response to Comments
document, the Agency disagrees with the commenters' assumption that the
results of Janes et al. (2007) and Greven et al. (2011) indicate
unmeasured confounding in the results of other cohort studies of long-
term exposure to PM2.5 and mortality. Therefore, the EPA
concludes that these studies do not invalidate the large body of
epidemiological evidence that supports the EPA's determination that a
causal relationship exists between long-term PM2.5 exposure
and mortality.\60\
---------------------------------------------------------------------------

\60\ The EPA notes that the EPA's conclusion with regard to
interpretation of the results from Janes et al. (2007) and Greven et
al. (2012) is supported by the study authors' conclusion that
``[o]ur results do not invalidate previous epidemiological studies''
(Dominici, 2012, p. 1 (emphasis original)).
---------------------------------------------------------------------------

(b) Heterogeneity in Risk Estimates
Some commenters argued that the heterogeneity in risk estimates
observed in multi-city epidemiological studies and the lack of
statistical significance in many regional or seasonal estimates
highlights a potential bias associated with combined multi-city
epidemiological study results (e.g., API, 2012, Attachment 1, pp. 15 to
19). These commenters further argued that more refined intra-urban
exposure estimates conducted for two of the largest cities included in
the ACS study, Los Angeles and New York City, based on land-use
regression models and/or kriging methods (Krewski et al., 2009)
``underscore the importance of considering city-specific health
estimates, which may account for heterogeneity in PM2.5
concentrations or other differences among cities, rather than relying
on pooled nationwide results from multi-city studies'' (API, 2012,
Attachment 1, p. 17).
With respect to understanding the nature and magnitude of
PM2.5-related risks, the EPA agrees that epidemiological
studies evaluating health effects associated with long- and short-term
PM2.5 exposures have reported heterogeneity in responses
between cities and effect estimates across geographic regions of the
U.S. (U.S. EPA, 2009a, sections 6.2.12.1, 6.3.8.1, 6.5.2, and 7.6.1;
U.S. EPA, 2011a, p. 2-25). For example, when focusing on short-term
PM2.5 exposure, the Integrated Science Assessment found that
multi-city studies that examined associations with mortality and
cardiovascular and respiratory hospital admissions and emergency
department visits demonstrated greater cardiovascular effects in the
eastern versus the western U.S. (Dominici, et al., 2006a; Bell et al.,
2008; Franklin et al. (2007, 2008)).
In addition, the Integrated Science Assessment evaluated studies
that provided some evidence for seasonal differences in
PM2.5 risk estimates, specifically in the northeast. The
Integrated Science Assessment found evidence indicating that
individuals may be at greater risk of dying from higher exposures to
PM2.5 in the warmer months, and at greater risk of
PM2.5 associated hospitalization for

[[Page 3118]]

cardiovascular and respiratory diseases during colder months of the
year. The limited influence of seasonality on PM risk estimates in
other regions of the U.S. may be due to a number of factors including
varying PM composition by season, exposure misclassification due to
regional tendencies to spend more or less time outdoors and air
conditioning usage, and the prevalence of infectious diseases during
the winter months (U.S. EPA, 2009a, p. 3-182).
Overall, the EPA took note in the proposal that uncertainties still
remain regarding various factors that contribute to heterogeneity
observed in epidemiological studies (77 FR 38909/3). Nonetheless, the
EPA recognizes that this heterogeneity could be attributed, at least in
part, to differences in PM2.5 composition across the U.S.,
as well as to exposure differences that vary regionally such as
personal activity patterns, microenvironmental characteristics, and the
spatial variability of PM2.5 concentrations in urban areas
(U.S. EPA, 2009a, section 2.3.2; 77 FR 38910).
As recognized in the Policy Assessment, the current epidemiological
evidence and the limited amount of city-specific speciated
PM2.5 data do not allow conclusions to be drawn that
specifically differentiate effects of PM2.5 in different
locations (U.S. EPA, 2011a, p. 2-25). Furthermore, the Integrated
Science Assessment concluded ``that many constituents of
PM2.5 can be linked with multiple health effects, and the
evidence is not yet sufficient to allow differentiation of those
constituents or sources that are more closely related to specific
health outcomes'' (U.S. EPA, 2009a, p. 2-17). CASAC thoroughly reviewed
the EPA's presentation of the scientific evidence indicating
heterogeneity in PM2.5 effect estimates in epidemiological
studies and concurred with the overall conclusions presented in the
Integrated Science Assessment.
(c) Exposure Measurement Error
Some commenters argued that the EPA did not adequately consider
exposure measurement error, which they asserted is an important source
of bias in epidemiological studies that can bias effect estimates in
either direction (e.g., API, 2012, Attachment 1, pp. 19 to 20).
The EPA agrees that exposure measurement error is an important
source of uncertainty and that the variability in risk estimates
observed in multi-city studies could be attributed, in part, to
exposure error due to measurement-related issues (77 FR 38910).
However, the Agency disagrees with the commenters' assertion that
exposure measurement error was not adequately considered in this
review. The Integrated Science Assessment included an extensive
discussion that addresses issues of exposure measurement error (U.S.
EPA, 2009a, sections 2.3.2 and 3.8.6). Exposure measurement error may
lead to bias in effect estimates in epidemiological studies. A number
of studies evaluated in the last review (U.S. EPA, 2004, section 8.4.5)
and in the current review (U.S. EPA, 2009a, section 3.8.6) have
discussed the direction and magnitude of bias resulting from specified
patterns of exposure measurement error (Armstrong 1998; Thomas et al.
1993; Carroll et al. 1995) and have generally concluded ``classical''
(i.e., random, within-person) exposure measurement error can bias
effect estimates towards the null. Therefore, consistent with
conclusions reached in the last review, the Integrated Science
Assessment concluded ``in most circumstances, exposure error tends to
bias a health effect estimate downward'' (U.S. EPA, 2009a, sections
2.3.2 and 3.8.6) (emphasis added). Thus, the EPA has both considered
and accounted for the possibility of exposure measurement error, and
the possible bias would make it more difficult to detect true
associations, not less difficult.
(d) Model Specification
Commenters contended that the EPA did not account for the fact that
``selecting an appropriate statistical model for epidemiologic studies
of air pollution involves several choices that involve much ambiguity,
scant biological evidence, and a profound impact on analytic results,
given that many estimated associations are weak'' (ACC, 2012, p. 5).
For short-term exposure studies, the EPA recognizes, as summarized in
the HEI review panel commentary that selecting a level of control to
adjust for time-varying factors, such as temperature, in time-series
epidemiological studies involves a trade-off (HEI, 2003). For example,
if the model does not sufficiently adjust for the relationship between
the health outcome and temperature, some effects of temperature could
be falsely ascribed to the pollution variable. Conversely, if an overly
aggressive approach is used to control for temperature, the result
would possibly underestimate the pollution-related effect and
compromise the ability to detect a small but true pollution effect
(U.S. EPA, 2004, p. 8-236; HEI, 2003, p. 266). The selection of
approaches to address such variables depends in part on prior knowledge
and judgments made by the investigators, for example, about weather
patterns in the study area and expected relationships between weather
and other time-varying factors and health outcomes considered in the
study. As demonstrated in section 6.5 of the Integrated Science
Assessment, the EPA thoroughly considered each of these issues and the
overall effect of different model specifications on the association
between short-term PM2.5 exposure and mortality. Regardless
of the model employed, consistent positive associations were observed
across studies that controlled for the potential confounding effects of
time and weather using different approaches (U.S. EPA 2009a, Figure 6-
27). The EPA also considered the influence of model specification in
the examination of long-term PM2.5 exposure studies. For
example, in section 7.6 of the Integrated Science Assessment, Figures
7-6 and 7-7 summarize the collective evidence that evaluated the
association between long-term PM2.5 exposure and mortality.
Regardless of the model used, these studies collectively found evidence
of consistent positive associations between long-term PM2.5
exposure and mortality.
The EPA, therefore, disagrees with commenters that model
specification was not considered when evaluating the epidemiological
evidence used to form causality determinations. The EPA specifically
points out that the process of assessing the scientific quality and
relevance of epidemiological studies includes examining ``important
methodological issues (e.g., lag or time period between exposure and
effects, model specifications, thresholds, mortality displacement)
related to interpretation of the health evidence (U.S. EPA, 2009, p. 1-
9).'' Consistent with the conclusions of the 2004 PM Air Quality
Criteria Document, the EPA recognizes that there is still no clear
consensus at this time as to what constitutes appropriate control of
weather and temporal trends in short-term exposure studies, and that no
single statistical modeling approach is likely to be most appropriate
in all cases (U.S. EPA, 2004, p. 8-238). However, the EPA believes that
the available evidence interpreted in light of these remaining
uncertainties does provide increased confidence relative to the last
review in the reported associations between short- and long-term
PM2.5 exposures and mortality and morbidity effects, alone
and in combination with other pollutants.
(e) Concentration-Response Relationship
Additionally, commenters questioned the interpretation of the shape
of the

[[Page 3119]]

concentration-response relationship, specifically stating that multiple
studies have demonstrated that there is a threshold in the PM-health
effect relationship and that the log-linear model is not biologically
plausible (API, 2012, Attachment 9; ACC, 2012, Appendix A, pp. 7 to 8).
The EPA disagrees with this assertion due to the number of studies
evaluated in the Integrated Science Assessment that continue to support
the use of a no-threshold, log-linear model to most appropriately
represent the PM concentration-response relationship (U.S. EPA, 2009a,
section 2.4.3). While recognizing that uncertainties remain, the EPA
believes that our understanding of this issue for both long- and short-
term exposure studies has advanced since the last review. As discussed
in the Integrated Science Assessment, both long- and short-term
exposure studies have employed a variety of statistical approaches to
examine the shape of the concentration-response function and whether a
threshold exists. While the EPA recognizes that there likely are
individual biological thresholds for specific health responses, the
Integrated Science Assessment concluded the overall evidence from
existing epidemiological studies does not support the existence of
thresholds at the population level, for effects associated with either
long-term or short-term PM exposures within the ranges of air quality
observed in these studies (U.S. EPA, 2009a, section 2.4.3).\61\ The
Integrated Science Assessment concluded that this evidence collectively
supported the conclusion that a no-threshold, log-linear model is most
appropriate (U.S. EPA, 2009a, sections 6.2.10.10, 6.5.2.7, and 7.6.4).
CASAC likewise advised that ``[a]lthough there is increasing
uncertainty at lower levels, there is no evidence of a threshold''
(Samet, 2010d, p. ii).
---------------------------------------------------------------------------

\61\ While epidemiological analyses have not identified a
population threshold in the range of air quality concentrations
evaluated in these studies, the EPA recognizes that it is possible
that such thresholds exist towards the lower end of these ranges (or
below these ranges).
---------------------------------------------------------------------------

The EPA recognizes that some short-term exposure studies have
examined the PM2.5 concentration-response relationship in
individual cities or on a city-to-city basis and observed heterogeneity
in the shape of the concentration-response curve across cities. As
discussed in (b) above, these findings are a source of uncertainty that
the EPA agrees requires further investigation. Nonetheless, the
Integrated Science Assessment concluded that ``the studies evaluated
further support the use of a no-threshold, log-linear model, but
additional issues such as the influence of heterogeneity in estimates
between cities and the effects of seasonal and regional differences in
PM on the concentration-response-relationship still require further
investigation'' (U.S. EPA, 2009a, p. 2-25).
(f) Relative Toxicity of PM2.5 Components
Some commenters highlighted uncertainties in understanding the role
of individual constituents within the mix of fine particles. These
commenters asserted that a mass-based standard may not be appropriate
due to the growing body of evidence indicating that certain
PM2.5 components may be more closely related to specific
health outcomes (e.g., EC and OC) (EPRI, 2012, p. 2).
With regard to questions about the role of individual constituents
within the mix of fine particles, as a general matter, the EPA
recognizes that although new research directed toward this question has
been conducted since the last review, important questions remain and
the issue remains an important element in the Agency's ongoing research
program. At the time of the last review, the Agency determined that it
was appropriate to continue to control fine particles as a group, as
opposed to singling out any particular component or class of fine
particles (71 FR 61162 to 61164, October 17, 2006). This distinction
was based largely on epidemiological evidence of health effects using
various indicators of fine particles in a large number of areas that
had significant contributions of differing components or sources of
fine particles, together with some limited experimental studies that
provided some evidence suggestive of health effects associated with
high concentrations of numerous fine particle components.
In this review, as discussed in the proposal (77 FR 38922 to 38923)
and in section III.E.1 below, while most epidemiological studies
continue to be indexed by PM2.5 mass, several recent
epidemiological studies included in the Integrated Science Assessment
have used PM2.5 speciation data to evaluate health effects
associated with fine particle exposures. In the Integrated Science
Assessment, the EPA thoroughly evaluated the scientific evidence that
examined the effect of different PM2.5 components and
sources on a variety of health outcomes (U.S. EPA, 2009a, section 6.6)
and observed that the available information continues to suggest that
many different chemical components of fine particles and a variety of
different types of source categories are all associated with, and
probably contribute to, effects associated with PM2.5. The
Integrated Science Assessment concluded that the current body of
scientific evidence indicated that ``many constituents of PM can be
linked with differing health effects and the evidence is not yet
sufficient to allow differentiation of those constituents or sources
that are more closely related to specific health outcomes'' (U.S. EPA,
2009a, p. 2-26 and 6-212). Furthermore, the Policy Assessment concluded
that the evidence is not sufficient to support eliminating any
component or group of components associated with any specific source
categories from the mix of fine particles included in the
PM2.5 indicator (U.S. EPA, 2009a, p. 2-56). CASAC agreed
that it was reasonable to retain PM2.5 as an indicator for
fine particles in this review as ``[t]here was insufficient peer-
reviewed literature to support any other indicator at this time''
(Samet, 2010c, p. 12).
This information is relevant to the Agency's decision to retain
PM2.5 as the indicator for fine particles as discussed in
section III.E.1 below. The EPA also believes that it is relevant to the
Agency's conclusion as to whether revision of the suite of primary
PM2.5 standards is appropriate. While there remain
uncertainties about the role and relative toxicity of various
components of fine PM, the current evidence continues to support the
view that fine particles should be addressed as a group for purposes of
public health protection.
In summary, in considering the above issues related to
uncertainties in the underlying health science, on balance, the EPA
believes that the available evidence interpreted in light of these
remaining uncertainties does provide increased confidence relative to
the last review in the reported associations between long- and short-
term PM2.5 exposures and mortality and morbidity effects,
alone and in combination with other pollutants, and supports stronger
inferences as to the causal nature of the associations. The EPA also
believes that this increased confidence, when taken in context of the
entire body of available health effects evidence and in light of the
evidence from epidemiological studies of associations observed in areas
meeting the current primary PM2.5 standards, specifically in
areas meeting the current primary annual PM2.5 standard,
adds support to its conclusion that the current suite of
PM2.5 standards needs to be revised to provide increased
public health protection.
(4) In asserting that there is no evidence of greater risk since
the 2006

[[Page 3120]]

review to justify lowering the current annual PM2.5
standard, some commenters argued that, ``if the current primary
PM2.5 annual standard of 15 [mu]g/m\3\ was considered to be
adequately protective of public health in 2006, given relative risk
estimates that EPA was using at that time, then that standard would
surely still be adequately protective of the public health if relative
risk estimates remain at the same level (or lower)'' (UARG, 2012,
Attachment 1, p. 24). These commenters compared risk coefficients used
for mortality in the EPA's risk assessment done in the last review with
those from the Agency's core risk assessment done as part of this
review, and they concluded that ``the entire range of the core relative
risk for long-term mortality is lower now than it was in the prior
review'' (UARG, 2012, Attachment 1, p. 24). These commenters used this
conclusion as the basis for a claim that there is no reason to revise
the current annual PM2.5 standard.
The EPA believes that this claim is fundamentally flawed. In
comparing the scientific understanding of the risk presented by
exposure to PM2.5 between the last and current reviews, one
must examine not only the quantitative estimate of risk from those
exposures (e.g., the numbers of premature deaths or increased hospital
admissions at various concentrations), but also the degree of
confidence that the Agency has that the observed health effects are
causally linked to PM2.5 exposure at those concentrations.
As documented in the Integrated Science Assessment and in the
recommendations and conclusions of CASAC, the EPA recognizes
significant advances in our understanding of the health effects of
PM2.5, based on evidence that is stronger than in the last
review. As a result of these advances, the EPA is now more certain that
fine particles, alone or in combination with other pollutants, present
a significant risk to public health at concentrations allowed by the
current primary PM2.5 standards. From this more
comprehensive perspective, since the risks presented by
PM2.5 are more certain, similar or even somewhat lower
relative risk estimates would not be a basis to conclude that no
revision to the suite of PM2.5 standards is ``requisite'' to
protect public health with an adequate margin of safety. This also
ignores that the relative risk estimate is only one factor considered
by the Administrator, e.g. it ignores that epidemiological studies
since the last review indicate associations between PM2.5
and mortality and morbidity in areas meeting the current annual
standard.
In any case, the commenters' reliance on the flawed 2006 review is
misplaced. As discussed in section III.A.2 above, the D.C. Circuit
remanded Administrator Johnson's 2006 decision to retain the primary
annual PM2.5 standard because the Agency failed to
adequately explain why the annual standard provided the requisite
protection from both short- and long-term exposure to fine particles
including protection for at-risk populations. The 2006 standard was
also at sharp odds with CASAC advice and recommendations as to the
requisite level of protection (Henderson, 2006a,b). In other words, the
2006 primary annual PM2.5 standard is not an appropriate
benchmark for comparison.
(5) Some of these commenters also identified ``new'' as well as
older studies that had been included in prior reviews as providing
additional evidence that the causality determinations presented in the
Integrated Science Assessment did not consider the totality of the
scientific literature, further supporting their view that a revision of
the PM2.5 is unwarranted. As discussed in section II.B.3
above, the EPA notes that, as in past NAAQS reviews, the Agency is
basing the final decisions in this review on the studies and related
information included in the Integrated Science Assessment that have
undergone CASAC and public review, and will consider newly published
studies for purposes of decisionmaking in the next PM NAAQS review. In
provisionally evaluating commenters' arguments (see Response to
Comments document), the EPA notes that its provisional assessment of
``new'' science found that such studies did not materially change the
conclusions reached in the Integrated Science Assessment (U.S. EPA,
2012b).
3. Administrator's Final Conclusions Concerning the Adequacy of the
Current Primary PM2.5 Standards
Having carefully considered the public comments, as discussed
above, the Administrator believes the fundamental scientific
conclusions on the effects of PM2.5 reached in the
Integrated Science Assessment, and discussed in the Policy Assessment,
are valid. In considering whether the suite of primary PM2.5
standards should be revised, the Administrator places primary
consideration on the evidence obtained from the epidemiological
studies. The Administrator believes that this literature, combined with
the other scientific evidence discussed in the Integrated Science
Assessment, collectively represents a strong and generally robust body
of evidence of serious health effects associated with both long- and
short-term exposures to PM2.5. As discussed in the
Integrated Science Assessment and Policy Assessment, the EPA believes
that much progress has been made since the last review in reducing some
of the major uncertainties that were important considerations in
establishing the current suite of PM2.5 standards. In that
context, the Administrator finds the evidence of serious health effects
reported in exposure studies conducted in areas with long-term mean
concentrations ranging from approximately at or above the level of the
annual standard to long-term mean concentrations significantly below
the level of the annual standard to be compelling, especially in light
of the extent to which such studies are part of an overall pattern of
positive and frequently statistically significant associations across a
broad range of studies. The information in the quantitative risk
assessment lends support to this conclusion.
There has been extensive critical review of this body of evidence,
the quantitative risk assessment, and related uncertainties, including
review by CASAC and the public. The public comments on the basis for
the EPA's proposed decision to revise the suite of primary
PM2.5 standards have identified a number of issues about
which different parties disagree including issues for which additional
research is warranted. Having weighed all comments and the advice of
CASAC, the Administrator believes that since the last review the
overall uncertainty about the public health risks associated with both
long- and short-term exposure to PM2.5 has been diminished
to an important degree. The remaining uncertainties in the available
evidence do not diminish confidence in the associations between
exposure to fine particles and mortality and serious morbidity effects.
Based on her increased confidence in the association between exposure
to PM2.5 and serious public health effects, combined with
evidence of such an association in areas that would meet the current
standards, the Administrator agrees with CASAC that revision of the
current suite of PM2.5 standards to provide increased public
health protection is necessary. Based on these considerations, the
Administrator concludes that the current suite of primary
PM2.5 standards is not sufficient, and thus not requisite,
to protect public health with an

[[Page 3121]]

adequate margin of safety, and that revision is needed to increase
public health protection.
It is important to note that this conclusion, and the reasoning on
which it is based, do not resolve the question of what specific
revisions are appropriate. That requires looking specifically at the
current 24-hour and annual PM2.5 standards, including their
indicator, averaging times, forms, and levels, and evaluating the
scientific evidence and other information relevant to determining the
appropriate revision of the suite of standards.

E. Conclusions on the Elements of the Primary Fine Particle Standards

1. Indicator
In initially setting standards for fine particles in 1997, the EPA
concluded it was appropriate to control fine particles as a group,
rather than singling out any particular component or class of fine
particles. The EPA noted that community health studies had found
significant associations between various indicators of fine particles,
and that health effects in a large number of areas had significant mass
contributions of differing components or sources of fine particles. In
addition, a number of toxicological and controlled human exposure
studies had reported health effects associations with high
concentrations of numerous fine particle components. It was also not
possible to rule out any component within the mix of fine particles as
not contributing to the fine particle effects found in the
epidemiologic studies (62 FR 38667, July 18, 1977). In establishing a
size-based indicator in 1977 to distinguish fine particles from
particles in the coarse mode, the EPA noted that the available
epidemiological studies of fine particles were based largely on
PM2.5 and also considered monitoring technology that was
generally available. The selection of a 2.5 [micro]m size cut reflected
the regulatory importance of defining an indicator that would more
completely capture fine particles under all conditions likely to be
encountered across the U.S., especially when fine particle
concentrations and humidity are likely to be high, while recognizing
that some small coarse particles would also be captured by current
methods to monitor PM2.5 (62 FR 38666 to 38668, July 18,
1997). In the last review, based on the same considerations, the EPA
again recognized that the available information supported retaining the
PM2.5 indicator and remained too limited to support a
distinct standard for any specific PM2.5 component or group
of components associated with any source categories of fine particles
(71 FR 61162 to 61164, October 17, 2006).
In this current review, the same considerations continue to apply
for selection of an appropriate indicator for fine particles. As an
initial matter, the Policy Assessment recognizes that the available
epidemiological studies linking mortality and morbidity effects with
long- and short-term exposures to fine particles continue to be largely
indexed by PM2.5. For the same reasons discussed in the last
two reviews, the Policy Assessment concluded that it was appropriate to
consider retaining a PM2.5 indicator to provide protection
from effects associated with long- and short-term fine particle
exposures (U.S. EPA, 2011a, p. 2-50).
The Policy Assessment also considered the expanded body of evidence
available in this review to consider whether there was sufficient
evidence to support a separate standard for ultrafine particles \62\ or
whether there was sufficient evidence to establish distinct standards
focused on regulating specific PM2.5 components or a group
of components associated with any source categories of fine particles
(U.S. EPA, 2011a, section 2.3.1).
---------------------------------------------------------------------------

\62\ Ultrafine particles, generally including particles with a
mobility diameter less than or equal to 0.1 [micro]m, are emitted
directly to the atmosphere or are formed by nucleation of gaseous
constituents in the atmosphere (U.S. EPA, 2009a, p. 3-3).
---------------------------------------------------------------------------

A number of studies available in this review have evaluated
potential health effects associated with short-term exposures to
ultrafine particles. As noted in the Integrated Science Assessment, the
enormous number and larger, collective surface area of ultrafine
particles are important considerations for focusing on this particle
size fraction in assessing potential public health impacts (U.S. EPA,
2009a, p. 6-83). Per unit mass, ultrafine particles may have more
opportunity to interact with cell surfaces due to their greater surface
area and their greater particle number compared with larger particles
(U.S. EPA, 2009a, p. 5-3). Greater surface area also increases the
potential for soluble components (e.g., transition metals, organics) to
adsorb to ultrafine particles and potentially cross cell membranes and
epithelial barriers (U.S. EPA, 2009a, p. 6-83). In addition, evidence
available in this review suggests that the ability of particles to
enhance allergic sensitization is associated more strongly with
particle number and surface area than with particle mass (U.S. EPA,
2009a, p. 6-127).
New evidence, primarily from controlled human exposure and
toxicological studies, expands our understanding of cardiovascular and
respiratory effects related to short-term ultrafine particle exposures.
However, the Policy Assessment concluded that this evidence was still
very limited and largely focused on exposure to diesel exhaust, for
which the Integrated Science Assessment concluded it was unclear
whether the effects observed are due to ultrafine particles, larger
particles within the PM2.5 mixture, or the gaseous
components of diesel exhaust (U.S. EPA, 2009a, p. 2-22). In addition,
the Integrated Science Assessment noted uncertainties associated with
the controlled human exposure studies using concentrated ambient
particle systems which have been shown to modify the composition of
ultrafine particles (U.S. EPA, 2009a, p. 2-22, see also section 1.5.3).
The Policy Assessment recognized that there are relatively few
epidemiological studies that have examined potential cardiovascular and
respiratory effects associated with short-term exposures to ultrafine
particles (U.S. EPA, 2011a, p. 2-51). These studies have reported
inconsistent and mixed results (U.S. EPA, 2009a, section 2.3.5).
Collectively, in considering the body of scientific evidence
available in this review, the Integrated Science Assessment concluded
that the currently available evidence was suggestive of a causal
relationship between short-term exposures to ultrafine particles and
cardiovascular and respiratory effects. Furthermore, the Integrated
Science Assessment concluded that evidence was inadequate to infer a
causal relationship between short-term exposure to ultrafine particles
and mortality as well as long-term exposure to ultrafine particles and
all outcomes evaluated (U.S. EPA, 2009a, sections 2.3.5, 6.2.12.3,
6.3.10.3, 6.5.3.3, 7.2.11.3, 7.3.9, 7.4.3.3, 7.5.4.3, and 7.6.5.3;
Table 2-6).
With respect to our understanding of ambient ultrafine particle
concentrations, at present, there is no national network of ultrafine
particle samplers; thus, only episodic and/or site-specific data sets
exist (U.S. EPA, 2009a, p. 2-2). Therefore, the Policy Assessment
recognized a national characterization of concentrations, temporal and
spatial patterns, and trends was not possible at this time, and the
availability of ambient ultrafine measurements to support health
studies was extremely limited (U.S. EPA, 2011a, p. 2-51). In general,
measurements of ultrafine particles are highly dependent on monitor
location and, therefore, more subject to exposure error than

[[Page 3122]]

accumulation mode particles (U.S. EPA, 2009a, p. 2-22). Furthermore,
the number of ultrafine particles generally decreases sharply downwind
from sources, as ultrafine particles may grow into the accumulation
mode by coagulation or condensation (U.S. EPA, 2009a, p. 3-89). Limited
studies of ambient ultrafine particle measurements have suggested that
these particles exhibit a high degree of spatial and temporal
heterogeneity driven primarily by differences in nearby source
characteristics (U.S. EPA, 2009a, p. 3-84). Internal combustion engines
and, therefore, roadways are a notable source of ultrafine particles,
so concentrations of these particles near roadways are generally
expected to be elevated (U.S. EPA, 2009a, p. 2-3). Concentrations of
ultrafine particles have been reported to drop off much more quickly
with distance from roadways than fine particles (U.S. EPA, 2009a, p. 3-
84).
In considering both the currently available health effects evidence
and the air quality data, the Policy Assessment concluded that this
information was still too limited to provide support for consideration
of a distinct PM standard for ultrafine particles (U.S. EPA, 2011a, p.
2-52).
In addressing the issue of particle composition, the Integrated
Science Assessment concluded that, ``[f]rom a mechanistic perspective,
it is highly plausible that the chemical composition of PM would be a
better predictor of health effects than particle size'' (U.S. EPA,
2009a, p. 6-202). Heterogeneity of ambient concentrations of
PM2.5 constituents (e.g., elemental carbon, organic carbon,
sulfates, nitrates) observed in different geographical regions as well
as regional heterogeneity in PM2.5-related health effects
reported in a number of epidemiological studies are consistent with
this hypothesis (U.S. EPA, 2009a, section 6.6).
With respect to the availability of ambient measurement data for
fine particle components in this review, the Policy Assessment noted
that there were now more extensive ambient PM2.5 speciation
measurement data available through the Chemical Speciation Network
(CSN) than in previous reviews (U.S. EPA, 2011a, section 1.3.2 and
Appendix B, section B.1.3). The Integrated Science Assessment observed
that data from the CSN provided further evidence of spatial and
seasonal variation in both PM2.5 mass and composition among
cities and geographic regions (U.S. EPA, 2009a, pp. 3-50 to 3-60;
Figures 3-12 to 3-18; Figure 3-47). Some of this variation may be
related to regional differences in meteorology, sources, and topography
(U.S. EPA, 2009a, p. 2-3).
The currently available epidemiological, toxicological, and
controlled human exposure studies evaluated in the Integrated Science
Assessment on the health effects associated with ambient
PM2.5 constituents and categories of fine particle sources
used a variety of quantitative methods applied to a broad set of
PM2.5 constituents, rather than selecting a few constituents
a priori (U.S. EPA, 2009a, p. 2-26). Epidemiological studies have used
measured ambient PM2.5 speciation data, including monitoring
data from the CSN, while all of the controlled human exposure and most
of the toxicological studies have used concentrated ambient particles
and analyzed the constituents therein (U.S. EPA, 2009a, p. 6-203).\63\
The CSN provides PM2.5 speciation measurements generally on
a one-in-three or one-in-six day sampling schedule and, thus, does not
capture data every day at most sites.\64\
---------------------------------------------------------------------------

\63\ Most studies considered between 7 to 20 ambient
PM2.5 constituents, with elemental carbon, organic
carbon, sulfates, nitrates, and metals most commonly measured. Many
of the studies grouped the constituents with various factorization
or source apportionment techniques to examine the relationship
between the grouped constituents and various health effects.
However, not all studies labeled the constituent groupings according
to their presumed source and a small number of controlled human
exposure and toxicological studies did not use any constituent
grouping. These differences across studies substantially limit any
integrative interpretation of these studies (U.S. EPA, 2009a, p. 6-
203).
\64\ To expand our understanding of the role of specific
PM2.5 components and sources with respect to the observed
health effects, researchers have expressed a strong interest in
having access to PM2.5 speciation measurements collected
more frequently (U.S. EPA, 2011a, p. 2-53, including footnote 47).
---------------------------------------------------------------------------

The Policy Assessment recognized that several new multi-city
studies evaluating short-term exposures to fine particle constituents
are now available. These studies continued to show an association
between mortality and cardiovascular and/or respiratory morbidity
effects and short-term exposures to various PM2.5 components
including nickel, vanadium, elemental carbon, organic carbon, nitrates,
and sulfates (U.S. EPA, 2011a, section 2.3.1; U.S. EPA, 2009a, sections
6.5.2.5 and 6.6).
Limited evidence is available to evaluate the health effects
associated with long-term exposures to PM2.5 components
(U.S. EPA, 2009a, section 7.6.2). The Policy Assessment noted the most
significant new evidence was provided by a study that evaluated
multiple PM2.5 components and an indicator of traffic
density in an assessment of health effects related to long-term
exposure to PM2.5 (Lipfert et al., 2006a). Using health data
from a cohort of U.S. military veterans and PM2.5
measurement data from the CSN, Lipfert et al. (2006a) reported positive
associations between mortality and long-term exposures to nitrates,
elemental carbon, nickel, and vanadium as well as traffic density and
peak ozone concentrations (U.S. EPA, 2011a, p. 2-54; U.S. EPA, 2009a,
pp. 7-89 to 7-90).
With respect to source categories of fine particles potentially
associated with a range of health endpoints, the Integrated Science
Assessment reported that the currently available evidence suggests
associations between cardiovascular effects and a number of specific
PM2.5-related source categories, including oil combustion,
wood or biomass burning, motor vehicle emissions, and crustal or road
dust sources (U.S. EPA, 2009a, section 6.6; Table 6-18). In addition, a
few studies have evaluated associations between PM2.5-
related source categories and mortality. For example, one study
reported an association between mortality and a PM2.5 coal
combustion factor (Laden et al., 2000), while other studies linked
mortality to a secondary sulfate long-range transport PM2.5
source (Ito et al., 2006; Mar et al., 2006) (U.S. EPA, 2009a, section
6.6.2.1). Other studies have looked at different components of
particulate matter. There was less consistency in associations observed
between selected sources of fine particles and respiratory health
endpoints, which may be partially due to the fact that fewer studies
have evaluated respiratory-related outcomes and measures. However,
there was some evidence for PM2.5-related associations with
secondary sulfate and decrements in lung function in asthmatic and
healthy adults (U.S. EPA, 2009a, p. 6-211; Gong et al., 2005; Lanki et
al., 2006). A couple of studies have observed an association between
respiratory endpoints in children and adults with asthma and surrogates
for the crustal/soil/road dust and traffic sources of PM (U.S. EPA,
2009a, p. 6-205; Gent et al., 2009; Penttinen et al., 2006).
Recent studies have shown that source apportionment methods have
the potential to add useful insights into which sources and/or PM
constituents may contribute to different health effects. Of particular
interest are several epidemiological studies that compared source
apportionment methods and reported consistent results across research
groups (U.S. EPA, 2009a, p. 6-211; Hopke et al., 2006; Ito et al.,
2006; Mar et al., 2006; Thurston et al., 2005).

[[Page 3123]]

These studies reported associations between total mortality and
secondary sulfate in two cities for two different lag times. The
sulfate effect was stronger for total mortality in Washington, DC and
for cardiovascular-related mortality in Phoenix (U.S. EPA, 2009a, p. 6-
204). These studies also found some evidence for associations with
mortality and a number of source categories (e.g., biomass/wood
combustion, traffic, copper smelter, coal combustion, sea salt) at
various lag times (U.S. EPA, 2009a, p. 6-204). Sarnat et al. (2008)
compared three different source apportionment methods and reported
consistent associations between emergency department visits for
cardiovascular diseases with mobile sources and biomass combustion as
well as increased respiratory-related emergency department visits
associated with secondary sulfate (U.S. EPA, 2009a, pp. 6-204 and 6-
211).
Collectively, in considering the currently available evidence for
health effects associated with specific PM2.5 components or
groups of components associated with any source categories of fine
particles as presented in the Integrated Science Assessment, the Policy
Assessment concluded that additional information available in this
review continues to provide evidence that many different constituents
of the fine particle mixture as well as groups of components associated
with specific source categories of fine particles are linked to adverse
health effects (U.S. EPA, 2011a, p. 2-55). However, as noted in the
Integrated Science Assessment, while ``[t]here is some evidence for
trends and patterns that link particular ambient PM constituents or
sources with specific health outcomes * * * there is insufficient
evidence to determine whether these patterns are consistent or robust''
(U.S. EPA, 2009a, p. 6-210). Assessing this information, the Integrated
Science Assessment concluded that ``the evidence is not yet sufficient
to allow differentiation of those constituents or sources that are more
closely related to specific health outcomes'' (U.S. EPA, 2009a, pp. 2-
26 and 6-212). Therefore, the Policy Assessment concluded that the
currently available evidence is not sufficient to support consideration
of a separate indicator for a specific PM2.5 component or
group of components associated with any source category of fine
particles. Furthermore, the Policy Assessment concluded that the
evidence is not sufficient to support eliminating any component or
group of components associated with any source categories of fine
particles from the mix of fine particles included in the
PM2.5 indicator (U.S. EPA, 2011a, p. 2-56).
The CASAC agreed with the EPA staff conclusions presented in the
Policy Assessment and concluded that it is appropriate to consider
retaining PM2.5 as the indicator for fine particles and
further asserted, ``There [is] insufficient peer-reviewed literature to
support any other indicator at this time'' (Samet, 2010c, p. 12). CASAC
expressed a strong desire for the EPA to ``look ahead to future review
cycles and reinvigorate support for the development of evidence that
might lead to newer indicators that may correlate better with the
health effects associated with ambient air concentrations of PM * * *''
(Samet, 2010c, p 2).
Consistent with the staff conclusions presented in the Policy
Assessment and CASAC advice, the Administrator proposed to retain
PM2.5 as the indicator for fine particles. Further, the
Administrator provisionally concluded that currently available
scientific information does not provide a sufficient basis for
supplementing mass-based, primary fine particle standards with
standards using a separate indicator for ultrafine particles or a
separate indicator for a specific PM2.5 component or group
of components associated with any source categories of fine particles.
In addition, the Administrator also provisionally concluded that the
currently available scientific information did not provide a sufficient
basis for eliminating any individual component or group of components
associated with any source categories from the mix of fine particles
included in the PM2.5 mass-based indicator.
The EPA received comparatively few public comments on issues
related to the indicator for fine particles.\65\ Some commenters
emphasized the need to conduct additional research to more fully
understand the effect of specific PM2.5 components and/or
sources on public health. These commenters expressed views about the
importance of evaluating health effect associations with various fine
particle components and types of source categories as a basis for
focusing ongoing and future research to reduce uncertainties in this
area and for considering whether alternative indicator(s) may be
appropriate to consider in future PM NAAQS reviews for standards
intended to protect against the array of health effects that have been
associated with fine particles as indexed by PM2.5. For
example, the PSR encouraged more research and monitoring related to
PM2.5 components and noted the importance of components
associated with coal combustion (PSR, 2012, pp. 5 to 6). EPRI asserted
that ``new'' studies support focusing on EC and OC and encouraged the
EPA to seriously consider the mass-based approach (EPRI, 2012, p. 2).
Likewise, Georgia Mining Association supported additional monitoring
and research efforts related to PM2.5 composition and
specifically encouraged the evaluation of using particle number (e.g.,
particle count) (GMA, 2012, pp. 2 to 3).
---------------------------------------------------------------------------

\65\ No public comments were submitted regarding the use of a
different size cut for fine particles.
---------------------------------------------------------------------------

The Administrator agrees with CASAC as well as these commenters
that the results of additional research and monitoring efforts will be
helpful for informing future PM NAAQS reviews. Information from such
studies could also help inform the development of strategies that
emphasize control of specific types of emission sources so as to
address particles of greatest concern to public health. However, based
upon the scientific information considered in the Integrated Science
Assessment as well as the public comments summarized above, the
Administrator continues to take note there is evidence that many
different constituents of the fine particle mixture as well as groups
of components associated with specific sources of fine particles are
linked to adverse health effects. Furthermore, she recognizes that the
evidence is not yet sufficient to differentiate those constituents or
sources that are most closely related to specific health outcomes nor
to exclude any PM2.5 components or sources of fine particles
from the mix of particles included in the PM2.5 indicator.
Having considered the public comments on this issue, the
Administrator concurs with the Policy Assessment conclusions and CASAC
recommendations and concludes that it is appropriate to retain
PM2.5 as the indicator for fine particles.
2. Averaging Time
In 1997, the EPA initially set both an annual standard, to provide
protection from health effects associated with both long- and short-
term exposures to PM2.5, and a 24-hour standard to
supplement the protection afforded by the annual standard (62 FR 38667
to 38668, July, 18, 1997). In the last review, the EPA retained both
annual and 24-hour averaging times (71 FR 61164, October 17, 2006).
These decisions were based, in part, on evidence of health effects
related to both long-term (from a year to several years) and short-term
(from less than one day to up to several days) measures of
PM2.5.

[[Page 3124]]

The overwhelming majority of studies conducted since the last
review continue to utilize annual (or multi-year) and 24-hour averaging
times, reflecting the averaging times of the current PM2.5
standards. These studies continue to provide evidence that health
effects are associated with annual and 24-hour averaging times.
Therefore, the Policy Assessment concluded it is appropriate to retain
the current annual and 24-hour averaging times to provide protection
from effects associated with both long- and short-term PM2.5
exposures (U.S. EPA, 2011a, p. 2-57).
In considering whether the information available in this review
supports consideration of different averaging times for
PM2.5 standards specifically with regard to considering a
standard with an averaging time less than 24 hours to address health
effects associated with sub-daily PM2.5 exposures, the
Policy Assessment noted there continues to be a growing body of studies
that provide additional evidence of effects associated with exposure
periods less than 24-hours (U.S. EPA, 2011a, p. 2-57). Relative to
information available in the last review, recent studies provide
additional evidence for cardiovascular effects associated with sub-
daily (e.g., one to several hours) exposure to PM, especially effects
related to cardiac ischemia, vasomotor function, and more subtle
changes in markers of systemic inflammation, hemostasis, thrombosis and
coagulation (U.S. EPA, 2009a, section 6.2). Because these studies have
used different indicators (e.g., PM2.5, PM10,
PM10-2.5, ultrafine particles), averaging times (e.g., 1, 2,
and 4 hours), and health outcomes, it is difficult to draw conclusions
about cardiovascular effects associated specifically with sub-daily
exposures to PM2.5.
With regard to respiratory effects associated with sub-daily
PM2.5 exposures, the currently available evidence was much
sparser than for cardiovascular effects and continues to be very
limited. The Integrated Science Assessment concluded that for several
studies of hospital admissions or medical visits for respiratory
diseases, the strongest associations were observed with 24-hour average
or longer exposures, not with less than 24-hour exposures (U.S. EPA,
2009a, section 6.3).
Collectively, the Policy Assessment concluded that this
information, when viewed as a whole, is too unclear, with respect to
the indicator, averaging time and health outcome, to serve as a basis
for consideration of establishing a primary PM2.5 standard
with an averaging time shorter than 24-hours at this time (U.S. EPA,
2011a, p. 2-57).
With regard to health effects associated with PM2.5
exposure across varying seasons in this review, Bell et al. (2008)
reported higher PM2.5 risk estimates for hospitalization for
cardiovascular and respiratory diseases in the winter compared to other
seasons. In comparison to the winter season, smaller statistically
significant associations were also reported between PM2.5
and cardiovascular morbidity for spring and autumn, and a positive, but
statistically non-significant association was observed for the summer
months. In the case of mortality, Zanobetti and Schwartz (2009)
reported a 4-fold higher effect estimate for PM2.5-
associated mortality for the spring as compared to the winter. Taken
together, these results provided emerging but limited evidence that
individuals may be at greater risk of dying from higher exposures to
PM2.5 in the warmer months and may be at greater risk of
PM2.5-associated hospitalization for cardiovascular and
respiratory diseases during colder months of the year (U.S. EPA, 2011a,
p. 2-58).
Overall, the Policy Assessment observed that there are few studies
presently available to deduce a general pattern in PM2.5-
related risk across seasons. In addition, these studies utilized 24-
hour exposure periods within each season to assess the
PM2.5-associated health effects and do not provide
information on health effects associated with a season-long exposure to
PM2.5. Due to these limitations in the currently available
evidence, the Policy Assessment concluded that there was no basis to
consider a seasonal averaging time separate from a 24-hour averaging
time.
Based on the above considerations, the Policy Assessment concluded
that the currently available information provided strong support for
consideration of retaining the current annual and 24-hour averaging
times but does not provide support for considering alternative
averaging times (U.S. EPA, 2011a, p. 2-58). In addition, CASAC
considered it appropriate to retain the current annual and 24-hour
averaging times for the primary PM2.5 standards (Samet,
2010c, pp. 2 to 3). At the time of the proposal, the Administrator
concurred with the staff conclusions and CASAC advice and proposed that
the averaging times for the primary PM2.5 standards should
continue to include annual and 24-hour averages to protect against
health effects associated with long- and short-term exposures.
Furthermore, the Administrator provisionally concluded, consistent with
conclusions reached in the Policy Assessment and by CASAC, that the
currently available information was too limited to support
consideration of alternative averaging times to establish a national
standard with a shorter-than 24-hour averaging time or with a seasonal
averaging time.
The EPA received no significant public comments on the issue of
averaging time for the PM2.5 primary standards. The
Administrator concurs with recommendations made by CASAC and the staff
conclusions presented in the Policy Assessment and concludes, as
proposed, that it is appropriate to retain the current annual and 24-
hour averaging times for the primary PM2.5 standards to
protect against health effects associated with long- and short-term
exposure periods.
3. Form
The ``form'' of a standard defines the air quality statistic that
is to be compared to the level of the standard in determining whether
an area attains the standard. In this review, the EPA considers whether
currently available information supports retaining or revising the
forms for the annual or 24-hour PM2.5 standards.
a. Annual Standard
In 1997, the EPA established the form of the annual
PM2.5 standard as an annual arithmetic mean, averaged over 3
years, from single or multiple community-oriented monitors. This form
was intended to represent a relatively stable measure of air quality
and to characterize longer-term area-wide PM2.5
concentrations, in conjunction with a 24-hour standard designed to
provide adequate protection against localized peak or seasonal
PM2.5 concentrations. The level of the standard was to be
compared to measurements made at each community-oriented monitoring
site, or, if specific criteria were met, measurements from multiple
community-oriented monitoring sites could be averaged (i.e., spatial
averaging) \66\ (62 FR 38671 to 38672, July 18, 1997). The constraints
were intended to ensure that spatial averaging would not result in
inequities in the level of protection provided by the standard (62 FR
38672, July 18, 1997). This approach was consistent with the
epidemiological studies on which the PM2.5 standard was
primarily based, in which air quality data were generally averaged
across multiple monitors in an

[[Page 3125]]

area or were taken from a single monitor that was selected to represent
community-wide exposures.
---------------------------------------------------------------------------

\66\ Spatial averaging as part of the form of the annual
PM2.5 standard is unique to this standard and is not used
with other PM standards nor with other NAAQS.
---------------------------------------------------------------------------

In the last review, the EPA tightened the criteria for use of
spatial averaging to provide increased protection for vulnerable
populations exposed to PM2.5. This change was based in part
on an analysis of the potential for disproportionate impacts on
potentially at-risk populations, which found that the highest
concentrations in an area tend to be measured at monitors located in
areas where the surrounding population is more likely to have lower
education and income levels and higher percentages of minority
populations (71 FR 61166/2, October 17, 2006; U.S. EPA, 2005, section
5.3.6.1).
In this review, as outlined in section III.B above and discussed
more fully in section III.B.3 of the proposal, there now exist more
health data such that the Integrated Science Assessment has identified
persons from lower socioeconomic strata as an at-risk population (U.S.
EPA, 2009a, section 8.1.7; U.S. EPA, 2011a, section 2.2.1). Moreover,
there now exist more years of PM2.5 air quality data than
were available in the last review. Consideration in the Policy
Assessment of the spatial variability across urban areas that was
revealed by this expanded data base has raised questions as to whether
an annual standard that allows for spatial averaging, even within
specified constraints as narrowed in 2006 (71 FR 61165 to 61167,
October 17, 2006), would provide appropriate public health protection.
In considering the potential for disproportionate impacts on at-
risk populations, the Policy Assessment considered an update of an air
quality analysis conducted for the last review (U.S. EPA, 2011a, pp. 2-
59 to 60; Schmidt, 2011, Analysis A). This analysis focused on
determining whether the spatial averaging provisions, as modified in
2006, could introduce inequities in protection for at-risk populations
exposed to PM2.5. Specifically, the Policy Assessment
considered whether persons of lower socioeconomic status, minority
groups, or different age groups (i.e., children or older adults) are
more likely than the general population to live in areas in which the
monitors recording the highest air quality values in an area are
located. Data used in this analysis included demographic parameters
measured at the Census Block or Census Block Group level, including
percent minority population, percent minority subgroup population,
percent of persons living below the poverty level, percent of persons
18 years of age or older, and percent of persons 65 years of age and
older. In each candidate geographic area, data from the Census Block(s)
or Census Block Group(s) surrounding the location of the monitoring
site (as delineated by radii buffers of 0.5, 1.0, 2.0, and 3.0 miles)
in which the highest air quality value was monitored were compared to
the average of monitored values in the area. This analysis looked
beyond areas that would meet the current spatial averaging criteria and
considered all urban areas (i.e., Core Based Statistical Areas or
CBSAs) with at least two valid annual design value monitors (Schmidt,
2011, Analysis A). Recognizing the limitations of such cross-sectional
analyses, the Policy Assessment observed that the highest
concentrations in an area tend to be measured at monitors located in
areas where the surrounding populations are more likely to live below
the poverty line and to have higher percentage of minorities (U.S. EPA,
2011a, p. 2-60).
Based upon the analysis described above, the Policy Assessment
concluded that the existing constraints on spatial averaging, as
modified in 2006, may be inadequate to avoid substantially greater
exposures in some areas, potentially resulting in disproportionate
impacts on at-risk populations of persons with lower SES levels as well
as minorities. Therefore, the Policy Assessment concluded that it was
appropriate to consider revising the form of the annual
PM2.5 standard such that it did not allow for the use of
spatial averaging across monitors. In doing so, the level of the annual
PM2.5 standard would be compared to measurements made at the
monitoring site that represents area-wide air quality recording the
highest PM2.5 concentrations \67\ (U.S. EPA, 2011a, p. 2-
60).
---------------------------------------------------------------------------

\67\ As discussed in section VIII.B.1 below, the EPA is revising
several terms associated with PM2.5 monitor placement.
Specifically, the EPA is revoking the term ``community-oriented''
and replacing it with the term ``area-wide'' monitoring.
---------------------------------------------------------------------------

The CASAC agreed with staff conclusions that it was ``reasonable''
for the EPA to eliminate the spatial averaging provisions (Samet,
2010d, p. 2). Further, in CASAC's comments on the first draft Policy
Assessment, it noted, ``Given mounting evidence showing that persons
with lower SES levels are a susceptible group for PM-related health
risks, CASAC recommends that the provisions that allow for spatial
averaging across monitors be eliminated for the reasons cited in the
(first draft) Policy Assessment'' (Samet, 2010c, p. 13). In its review
of the second draft Policy Assessment, CASAC recognized ``although much
of the epidemiological research has been conducted using community-wide
averages, several key studies reference the nearest measurement site,
so that some risk estimates are not necessarily biased by the averaging
process. Further, the number of such studies is likely to expand in the
future'' (Samet, 2010d, pp. 1 to 2).
Only two areas in the country used the initial spatial averaging
provisions for demonstrating attainment with the primary annual
PM2.5 standard set in 1997 (70 FR 19847, April 14, 2005;
U.S. EPA, 2006c). Since these provisions were tightened in 2006, no
area has used spatial averaging to demonstrate attainment. No areas in
the country are currently using the spatial averaging provisions to
demonstrate attainment with the current primary annual PM2.5
standard.
In considering the Policy Assessment's conclusions based on the
results of the analysis discussed above and concern over the evidence
of potential disproportionate impacts on at-risk populations as well as
CASAC advice, the Administrator proposed to revise the form of the
annual PM2.5 standard to eliminate the use of spatial
averaging. Thus, the Administrator proposed revising the form of the
annual PM2.5 standard to compare the level of the standard
with measurements from each ``appropriate'' monitor in an area \68\
with no allowance for spatial averaging. Thus, for an area with
multiple monitors, the appropriate reporting monitor with the highest
design value would determine the attainment status for that area.
---------------------------------------------------------------------------

\68\ As discussed in section VIII.B.2.b below, the EPA concludes
that PM2.5 monitoring sites at micro- and middle-scale
locations are comparable to the annual standard if the monitoring
site has been approved by the Regional Administrator as representing
an area-wide location.
---------------------------------------------------------------------------

Of the commenters noted in section III.D.2 above who supported a
more stringent annual PM2.5 standard, those who commented on
the form of the annual PM2.5 standard supported the EPA's
proposal to eliminate the spatial averaging provisions. These
commenters contended that the EPA's analyses of the potential impacts
of spatial averaging, discussed above and in the proposal (77 FR
38924), demonstrated that the current form results in uneven public
health protection leading to disproportionate impacts on at-risk
populations. Specifically, the ALA and other environmental and public
health commenters contended that ``spatial averaging allows exposure of
people to unhealthy levels of pollution at specific locales even within
an area meeting the standard'' (ALA et al., 2012, p. 23).

[[Page 3126]]

These commenters particularly focused on the importance for low-income
and minority populations of eliminating the spatial averaging
provisions. They concluded that spatial averaging ``is an environmental
justice concern because poor people are more likely to live near roads,
depots, factories, ports, and other pollution sources.'' Id. p. 24.
Other commenters (e.g., AAM, 2012; Dow, 2012) also supported the
elimination of spatial averaging in order to ``avoid potential
disproportionate impacts on at-risk populations'' and to maximize ``the
benefits to public health of reducing the annual PM2.5
standard.'' However, these groups expressed concern that the
elimination of spatial averaging, in combination with the requirement
for near road monitors (as discussed in section VIII.B.3.b.i of the
proposal), would effectively and inappropriately increase the
stringency of the annual PM2.5 standard.
This concern was also shared by other commenters who disagreed with
the elimination of spatial averaging. For example, the Class of '85 RRG
emphasized concerns about increasing the stringency of the standard
while providing few health benefits if spatial averaging is eliminated,
particularly in combination with the requirement for near-road
monitors. These commenters contended that ``[b]ecause EPA proposes to
use the readings from the highest single worst case monitor (rather
than the average of all community area monitors), and since roadway
monitoring locations will likely be worst case monitors, the proposed
NAAQS will become more stringent without targeting the PM2.5
species most harmful to human health'' (Class of '85 RRG, 2012, p. 6).
Several commenters also maintained that because spatial averaging
is consistent with how air quality data are considered in the
underlying epidemiological studies, such averaging should not be
eliminated. Specifically, commenters including NAM et al., AFPM, and
ACC pointed out that PM2.5 epidemiological studies use
spatially averaged multi-monitor concentrations, rather than the single
highest monitor, when evaluating health effects. Therefore, these
commenters contended that allowing spatial averaging would make the
PM2.5 standard more consistent with the approaches used in
the epidemiological studies upon which the standard is based. In
addition, some commenters also contended that the EPA failed to
consider whether modifying, rather than eliminating, the constraints on
spatial averaging would have been sufficient to protect the public
health. If so, these commenters argued that ``elimination of spatial
averaging would go beyond what is requisite to protect the public
health'' (NAM et al., 2012, p. 20).
In considering the public comments on the form of the annual
standard, the EPA recognizes a number of commenters agreed with the
basis for the EPA's proposal to eliminate spatial averaging. While
other commenters expressed disagreement or concern with the proposed
decision to eliminate the spatial averaging provisions, the Agency
notes that these commenters did not challenge the analyses or
considerations that provided the fundamental basis for the
Administrator's proposed decision. Rather, these commenters generally
raised concerns that eliminating the option for spatial averaging would
increase the stringency of the standard, especially in light of
additional monitoring sites in near-road environments (as discussed in
section VIII.B.3.b.1 below).
The EPA does not agree with the comment that siting some monitors
in near roadway environments makes the standard more stringent or
impermissibly more stringent. As discussed in section VIII.B.3.b.i
below, a significant fraction of the population lives in proximity to
major roads, and these exposures occur in locations that represent
ambient air. Monitoring in such areas does not make the standard more
stringent than warranted, but rather affords the intended protection to
the exposed populations, among them at-risk populations, exposed to
fine particles in these areas. Thus, in cases where monitors in near
roadway environments are deemed to be representative of area-wide air
quality they would be compared to the annual standard (as discussed
more fully in section VIII below). The 24-hour and annual NAAQS are
designed to protect the public with an adequate margin of safety, and
this siting provision is fully consistent with providing the protection
the standard is designed to provide and does not make the standard more
stringent or more stringent than necessary.
Monitors that are representative of area-wide air quality may be
compared to the annual standard. This is consistent with the use of
monitoring data in the epidemiological studies that provide the primary
basis for determining the level of the annual standard. In addition,
the EPA notes that the annual standard is designed to protect against
both long- and short-term exposures through controlling the broad
distribution of air quality across an area over time.\69\ It is fully
consistent with the protection the standard is designed to provide for
near road monitors to be compared to the annual standard if the monitor
is representative of area-wide air quality. This does not make the
standard either more stringent or impermissibly more stringent.
---------------------------------------------------------------------------

\69\ This is in contrast to the 24-hour standard which is
designed to provide supplemental protection, addressing peak
exposures that might not otherwise be addressed by the annual
standard. Consistent with this, monitors are not required to be
representative of area-wide air quality to be compared to the 24-
hour standard.
---------------------------------------------------------------------------

In further considering these comments, the EPA notes that the
stringency or level of protection provided by each NAAQS is not based
solely on the form of the standard; rather, the four elements of the
standard that together serve to define each standard (i.e., indicator,
averaging time, form, and level) must be considered collectively in
evaluating the protection afforded by each standard. Therefore, the EPA
considers these comments are also appropriate to discuss collectively
with other issues related to the appropriate level for annual standard,
and are discussed below in sections III.E.4.c-d.
In reaching a final decision on the form of the annual standard,
the Administrator considers the available analyses, CASAC advice, and
public comments on form as discussed above. She also considers related
issues in the public comments on the level of the annual standard as
discussed in section III.E.4.c below. She notes that even when the
annual PM2.5 standard was first set in 1997, the spatial
averaging provisions included constraints intended to ensure that
inequities in the level of protection would not result. These
constraints on spatial averaging were tightened in the last review,
based on an analysis showing the potential for spatial averaging to
allow higher PM2.5 concentrations in locations where
subgroups within the general population were potentially
disproportionately exposed and hence, at disproportionate risk (e.g.,
low income and minority communities). The Administrator notes that in
proposing to eliminate spatial averaging altogether in this review, she
has relied on further analyses in the current review (Schmidt, 2011,
Analysis A). As discussed above and in the proposal (77 FR 38924),
these analyses showed that the current constraints on spatial averaging
may be inadequate in some areas to avoid substantially greater
exposures for people living near monitors recording the highest
PM2.5 concentrations. Such exposures could result in

[[Page 3127]]

disproportionate impacts to at-risk populations, including low-income
populations as well as minority groups.
On this basis, the Administrator concludes that public health would
not be protected with an adequate margin of safety in all locations, as
required by law, if disproportionately higher exposure concentrations
in at-risk populations such as low income communities as well as
minority communities were averaged together with lower concentrations
measured at other sites in a large urban area. See ALA v. EPA, 134 F.
3d 388, 389 (D.C. Cir., 1998) (``this court has held that `NAAQS must
protect not only average healthy individuals, but also sensitive
citizens such as children,' and `if a pollutant adversely affects the
health of these sensitive individuals, EPA must strengthen the entire
national standard''') and Coalition of Battery Recyclers Association v.
EPA, 604 F 3d. 613, 617 (D.C. Cir., 2010) (``Petitioners' assertion
that the revised lead NAAQS is overprotective because it is more
stringent than necessary to protect the entire population of young U.S.
children ignores that the Clean Air Act allows protection of sensitive
subpopulations.'') In reaching this conclusion, the Administrator
further notes that her concern over possible disproportionate
PM2.5-related health impacts in at-risk populations extends
to populations living near important sources of PM2.5,
including the large populations that live near major roadways.\70\
---------------------------------------------------------------------------

\70\ Section VIII.B.3.b.i below discusses public comments
specifically related to the proposed requirement for near-road
monitors.
---------------------------------------------------------------------------

In light of all of the above considerations, including
consideration of available analyses, CASAC advice, and public comments,
the Administrator concludes that the current form of the annual
PM2.5 standard should be revised to eliminate spatial
averaging provisions. Thus, the level of the revised annual
PM2.5 standard established with this rule will be compared
with measurements from each appropriate monitor in an area, with no
allowance for spatial averaging. The Administrator's conclusions with
regard to the appropriate level of the annual PM2.5 standard
to set in conjunction with this form are discussed below in section
III.E.4.d.
b. 24-Hour Standard
In 1997, the EPA established the form of the 24-hour
PM2.5 standard as the 98th percentile of 24-hour
concentrations at each population-oriented monitor within an area,
averaged over three years (62 FR at 38671 to 38674, July 18, 1997). The
Agency selected the 98th percentile as an appropriate balance between
adequately limiting the occurrence of peak concentrations and providing
increased stability which, when averaged over 3 years, facilitated
effective health protection through the development of more stable
implementation programs. By basing the form of the standard on
concentrations measured at population-oriented monitoring sites, the
EPA intended to provide protection for people residing in or near
localized areas of elevated concentrations. In the last review, in
conjunction with lowering the level of the 24-hour standard, the EPA
retained this form based in part on a comparison with the 99th
percentile form.\71\
---------------------------------------------------------------------------

\71\ In reaching this final decision, the EPA recognized a
technical problem associated with a potential bias in the method
used to calculate the 98th percentile concentration for this form.
The EPA adjusted the sampling frequency requirement in order to
reduce this bias. Accordingly, the Agency modified the final
monitoring requirements such that areas that are within 5 percent of
the standards are required to increase the sampling frequency to
every day (71 FR 61164 to 61165, October 17, 2006).
---------------------------------------------------------------------------

In revisiting the stability of a 98th versus 99th percentile form
for a 24-hour standard intended to provide supplemental protection for
a generally controlling annual standard, an analysis presented in the
Policy Assessment considered air quality data reported in 2000 to 2008
to update our understanding of the ratio between peak-to-mean
PM2.5 concentrations. This analysis provided evidence that
the 98th percentile value was a more stable metric than the 99th
percentile (U.S. EPA, 2011a, Figure 2-2, p. 2-62).
At the time of the proposal, the Agency recognized that the
selection of the appropriate form of the 24-hour standard includes
maintaining adequate protection against peak 24-hour concentrations
while also providing a stable target for risk management programs,
which serves to provide for the most effective public health protection
in the long run.\72\ As in previous reviews, the EPA recognized that a
concentration-based form, compared to an exceedance-based form, was
more reflective of the health risks posed by elevated pollutant
concentrations because such a form gives proportionally greater weight
to days when concentrations are well above the level of the standard
than to days when the concentrations are just above the level of the
standard. Further, the Agency provisionally concluded that a
concentration-based form, when averaged over three years, provided an
appropriate balance between limiting peak pollutant concentrations and
providing a stable regulatory target, thus facilitating the development
of more stable implementation programs.
---------------------------------------------------------------------------

\72\ See ATA III, 283 F.3d at 374-376 which concludes that it is
legitimate for the EPA to consider overall stability of the standard
and its resulting promotion of overall effectiveness of NAAQS
control programs in setting a standard that is requisite to protect
the public health.
---------------------------------------------------------------------------

In considering the information provided in the Policy Assessment
and recognizing that the degree of public health protection likely to
be afforded by a standard is a result of the combination of the form
and the level of the standard, the Administrator proposed to retain the
98th percentile form of the 24-hour standard. The Administrator
provisionally concluded that the 98th percentile form represents an
appropriate balance between adequately limiting the occurrence of peak
concentrations and providing increased stability relative to an
alternative 99th percentile form.
Few public commenters commented specifically on the form of the 24-
hour standard. None of the public commenters raised objections to
continuing the use of a concentration-based form for the 24-hour
standard. Many of the individuals and groups who supported a more
stringent 24-hour PM2.5 standard noted in section III.D.2
above, however, recommended a more restrictive concentration-based
percentile form, specifically a 99th percentile form. The limited
number of these commenters who provided a specific rationale for this
recommendation generally expressed their concern that the 98th
percentile form could allow too many days where concentrations exceeded
the level of the standard, and thus fail to adequately protect public
health. Other public commenters representing state and local air
agencies and industry groups generally supported retaining the current
98th percentile form. In most cases, these groups expressed the overall
view that the current 24-hour PM2.5 standard, including the
form of the current standard, should be retained.
The EPA notes that the viewpoints represented in this review are
similar to comments submitted in the last review and through various
NAAQS reviews. The EPA recognizes that the selection of the appropriate
form includes maintaining adequate protection against peak 24-hour
values while also providing a stable target for risk management
programs, which serves to provide for the most effective public

[[Page 3128]]

health protection in the long run.\73\ Nothing in the commenters' views
has provided a reason to change the Administrator's previous conclusion
regarding the appropriate balance represented in the proposed form of
the 24-hour PM2.5 standard. Therefore, the Administrator
concurs with staff conclusions presented in the Policy Assessment and
CASAC recommendations and concludes that it is appropriate to retain
the 98th percentile form for the 24-hour PM2.5 standard.
---------------------------------------------------------------------------

\73\ As just noted above, it is legitimate for the EPA to
consider promotion of overall effectiveness of risk management
programs designed to attain the NAAQS, including their overall
stability, in setting a standard that is requisite to protect the
public health. The context for the court's discussion in ATA III is
identical to that here; whether to adopt a 98th percentile form for
a 24-hour standard intended to provide supplemental protection for a
generally controlling annual standard.
---------------------------------------------------------------------------

4. Level
In the last review, the EPA selected levels for the annual and the
24-hour PM2.5 standards using evidence of effects associated
with periods of exposure that were most closely matched to the
averaging time of each standard. Thus, as discussed in section III.A.1,
the EPA relied upon evidence from long-term exposure studies as the
principal basis for selecting the level of the annual PM2.5
standard that would protect against effects associated with long-term
exposures. The EPA relied upon evidence from the short-term exposure
studies as the principal basis for selecting the level of the 24-hour
PM2.5 standard that would protect against effects associated
with short-term exposures. As summarized in section III.A.2 above, the
2006 decision to retain the level of the annual PM2.5
standard at 15 [mu]g/m\3\ \74\ was challenged and on judicial review,
the DC Circuit remanded the primary annual PM2.5 standard to
the EPA, finding that EPA's explanation for its approach to setting the
level of the annual standard was inadequate.
---------------------------------------------------------------------------

\74\ Throughout this section, the annual standard levels are
denoted as integer values for simplicity, although, as noted above
in section II.B.1, Table 1, the annual standard level is defined to
one decimal place, such that the current annual standard level is
15.0 [mu]g/m\3\. Alternative annual standard levels discussed in
this section are similarly defined to one decimal place.
---------------------------------------------------------------------------

a. General Approach for Considering Standard Levels
Building upon the lessons learned in the previous PM NAAQS reviews,
in considering alternative standard levels supported by the currently
available scientific information, the Policy Assessment used an
approach that integrated evidence-based and risk-based considerations,
took into account CASAC advice, and considered the issues raised by the
court in remanding the primary annual PM2.5 standard.
Following the general approach outlined in section III.A.3 above, for
the reasons discussed below, the Policy Assessment concluded it was
appropriate to consider the protection afforded by the annual and 24-
hour standards taken together against mortality and morbidity effects
associated with both long- and short-term PM2.5 exposures.
This was consistent with the approach taken in the review completed in
1997 rather than considering each standard separately, as was done in
the review completed in 2006.
Beyond looking directly at the relevant epidemiologic evidence, the
Policy Assessment considered the extent to which specific alternative
PM2.5 standard levels were likely to reduce the nature and
magnitude of both long-term exposure-related mortality risk and short-
term exposure-related mortality and morbidity risk (U.S. EPA, 2011a,
section 2.3.4.2; U.S.EPA, 2010a, section 4.2.2). As noted in section
III.C above, patterns of increasing estimated risk reductions were
generally observed as either the annual or 24-hour standard, or both,
were reduced below the level of the current standards (U.S. 2011a,
Figures 2-11 and 2-12; U.S. EPA, 2010a, sections 4.2.2, 5.2.2, and
5.2.3).
Based on the quantitative risk assessment, the Policy Assessment
observed, as discussed in section III.A.3, that analyses conducted for
this and previous reviews demonstrated that much, if not most, of the
aggregate risk associated with short-term exposures results from the
large number of days during which the 24-hour average concentrations
are in the low-to mid-range, below the peak 24-hour concentrations
(U.S. EPA, 2011a, p. 2-9). Furthermore, as discussed in section III.C
above and in section III.C.3 of the proposal, the Risk Assessment
observed that alternative annual standard levels, when controlling,
resulted in more consistent risk reductions across urban study areas,
thereby potentially providing a more consistent degree of public health
protection (U.S. EPA, 2010a, pp. 5-15 to 5-16). In contrast, the Risk
Assessment noted that the results of simulating alternative suites of
PM2.5 standards including different combinations of
alternative annual and 24-hour standard levels suggested that an
alternative 24-hour standard level can produce additional estimated
risk reductions beyond that provided by an alternative annual standard
alone. However, the degree of estimated risk reduction provided by
alternative 24-hour standard levels was highly variable, in part due to
the choice of rollback approached used (U.S. EPA, 2010a, p. 5-17).
Based on its review of the second draft Policy Assessment, CASAC
agreed with the EPA staff's general approach for translating the
available epidemiological evidence, risk information, and air quality
information into the basis for reaching conclusions on alternative
standards for consideration. Furthermore, CASAC agreed ``that it is
appropriate to return to the strategy used in 1997 that considers the
annual and the short-term standards together, with the annual standard
as the controlling standard, and the short-term standard supplementing
the protection afforded by the annual standard'' and ``considers it
appropriate to place the greatest emphasis'' on health effects judged
to have evidence supportive of a causal or likely causal relationship
as presented in the Integrated Science Assessment (Samet, 2010d, p. 1).
Therefore, the Policy Assessment concluded, consistent with
specific CASAC advice, that it was appropriate to set a ``generally
controlling'' annual standard that will lower a wide range of ambient
24-hour concentrations. The Policy Assessment concluded this approach
would likely reduce aggregate risks associated with both long- and
short-term exposures with more consistency than a generally controlling
24-hour standard and would be the most effective and efficient way to
reduce total PM2.5-related population risk and so provide
appropriate protection. The staff believed this approach, in contrast
to one focusing on a generally controlling 24-hour standard, would
likely reduce aggregate risks associated with both long- and short-term
exposures with more consistency and would likely avoid setting national
standards that could result in relatively uneven protection across the
country due to setting standards that were either more or less
stringent than necessary in different geographical areas.
The Policy Assessment recognized that an annual standard intended
to serve as the primary means for providing protection against effects
associated with both long- and short-term PM2.5 exposures
cannot be expected to offer an adequate margin of safety against the
effects of all short-term PM2.5 exposures. As a result, in
conjunction with a generally controlling annual standard, the Policy
Assessment concluded it was appropriate to

[[Page 3129]]

consider setting a 24-hour standard to provide supplemental protection,
particularly for areas with high peak-to-mean ratios possibly
associated with strong local or seasonal sources, or PM2.5-
related effects that may be associated with shorter-than-daily exposure
periods.
At the time of the proposal, the Administrator agreed with the
approach discussed in the Policy Assessment as summarized in section
III.A.3 above, and supported by CASAC, of considering the protection
afforded by the annual and 24-hour standards taken together for
mortality and morbidity effects associated with both long- and short-
term exposures to PM2.5. Furthermore, based on the evidence
and quantitative risk assessment, the Administrator provisionally
concluded it was appropriate to set a ``generally controlling'' annual
standard that will lower a wide range of ambient 24-hour
concentrations, with a 24-hour standard focused on providing
supplemental protection, particularly for areas with high peak-to-mean
ratios possibly associated with strong local or seasonal sources, or
PM2.5-related effects that may be associated with shorter-
than daily exposure periods. The Administrator provisionally concluded
this approach would likely reduce aggregate risks associated with both
long- and short-term exposures more consistently than a generally
controlling 24-hour standard and would be the most effective and
efficient way to reduce total PM2-5-related population risk.
The Administrator is mindful that considering what standards are
requisite to protect public health with an adequate margin of safety
requires public health policy judgments that neither overstate nor
understate the strength and limitations of the evidence or the
appropriate inferences to be drawn from the evidence. At the time of
the proposal, in considering how to translate the available information
into appropriate standard levels, the Administrator weighed the
available scientific information and associated uncertainties and
limitations. For the purpose of determining what standard levels were
appropriate to propose, the Administrator recognized, as did the EPA
staff in the Policy Assessment, that there was no single factor or
criterion that comprised the ``correct'' approach to weighing the
various types of available evidence and information, but rather there
were various approaches that were appropriate to consider. The
Administrator further recognized that different evaluations of the
evidence and other information before the Administrator could reflect
placing different weight on the relative strengths and limitations of
the scientific information, and different judgments could be made as to
how such information should appropriately be used in making public
health policy decisions on standard levels. This recognition led the
Administrator to consider various approaches to weighing the evidence
so as to identify appropriate standard levels to propose. In so doing,
the Administrator encouraged extensive public comment on alternative
approaches to weighing the evidence and other information so as to
inform her public health policy judgments before reaching final
decisions on appropriate standard levels.
b. Proposed Decisions on Standard Levels
i. Consideration of the Alternative Standard Levels in the Policy
Assessment
In recognizing the absence of a discernible population threshold
below which effects would not occur, the Policy Assessment's general
approach for identifying alternative annual standard levels that were
appropriate to consider focused on characterizing the part of the
distribution of PM2.5 concentrations in which we had the
most confidence in the associations reported in the epidemiological
studies and conversely where our confidence in the association became
appreciably lower. The most direct approach to address this issue,
consistent with CASAC advice (Samet, 2010c, p. 10), was to consider
epidemiological studies reporting confidence intervals around
concentration-response relationships (U.S. EPA, 2011a, p. 2-63). Based
on a thorough search of the available evidence, the Policy Assessment
identified only one study (Schwartz et al., 2008) that conducted a
multi-model analysis to characterize confidence intervals around the
estimated concentration-response relationship. The Policy Assessment
concluded that this single relevant analysis was too limited to serve
as the principal basis for identifying alternative standard levels in
this review (U.S. EPA, 2011a, p. 2-70).
The Policy Assessment explored other approaches to characterize the
part of the distributions of long-term mean PM2.5
concentrations that were most influential in generating health effect
estimates in long- and short-term epidemiological studies, and placed
greatest weight on those studies that reported positive and
statistically significant associations (U.S. EPA, 2011a, p. 2-63).
First, as discussed in section III.A.3 above, the Policy Assessment
considered the statistical metric used in previous reviews. This
approach recognized the EPA's views that the strongest evidence of
associations occurs at concentrations around the long-term mean
concentration. Thus, in earlier reviews, the EPA focused on identifying
standard levels that were somewhat below the long-term mean
concentrations reported in PM2.5 epidemiological studies.
The long-term mean concentrations represented air quality data
typically used in epidemiological analyses and provided a direct link
between PM2.5 concentrations and the observed health
effects. Further, these data were available for all long- and short-
term exposure studies analyzed and, therefore, represented the data set
available for the broadest set of epidemiological studies.

[[Page 3130]]

However, consistent with CASAC's comments on the second draft
Policy Assessment \75\ (Samet, 2010d, p. 2), in preparing the final
Policy Assessment, the EPA staff explored ways to take into account
additional information from epidemiological studies, when available
(Rajan et al., 2011). These analyses focused on evaluating different
statistical metrics, beyond the long-term mean concentration, to
characterize the part of the distribution of PM2.5
concentrations in which staff continued to have confidence in the
associations observed in epidemiological studies and below which there
was a comparative lack of data such that the staff's confidence in the
relationship was appreciably less. This would also be the part of the
distribution of PM2.5 concentrations which had the most
influence on generating the health effect estimates reported in
epidemiological studies. As discussed in section III.A.3 above, the
Policy Assessment recognized there was no one percentile value within a
given distribution that was the most appropriate or ``correct'' way to
characterize where our confidence in the associations becomes
appreciably lower. The Policy Assessment concluded that focusing on
concentrations within the lower quartile of a distribution, such as the
range from the 25th to the 10th percentile, was reasonable to consider
as a region within which we begin to have appreciably less confidence
in the associations observed in epidemiological studies.\76\ In the EPA
staff's view, considering lower PM2.5 concentrations, down
to the lowest concentration observed in a study, would be a highly
uncertain basis for selecting alternative standard levels (U.S. EPA,
2009a, p. 2-71).
---------------------------------------------------------------------------

\75\ While CASAC expressed the view that it would be most
desirable to have information on concentration-response
relationships, they recognized that it would also be ``preferable to
have information on the concentrations that were most influential in
generating the health effect estimates in individual studies''
(Samet, 2010d, p. 2).
\76\ In the last review, staff believed it was appropriate to
consider a level for an annual PM2.5 standard that was
somewhat below the averages of the long-term concentrations across
the cities in each of the key long-term exposures studies,
recognizing that the evidence of an association in any such study
was strongest at and around the long-term average where the data in
the study are most concentrated. For example, the interquartile
range of long-term average concentrations within a study and a range
within one standard deviation around the study mean were considered
reasonable approaches for characterizing the range over which the
evidence of association is strongest (U.S. EPA, 2005, pp. 5-22 to 5-
23). In this review, the Policy Assessment noted the
interrelatedness of the distributional statistics and a range of one
standard deviation around the mean which contains approximately 68
percent of normally distributed data, in that one standard deviation
below the mean falls between the 25th and 10th percentiles (U.S.
EPA, 2011a, p. 2-71).
---------------------------------------------------------------------------

As outlined in section III.A.3 above, the Policy Assessment
recognized that there were two types of population-level information to
consider in identifying the range of PM2.5 concentrations
which have the most influence on generating the health effect estimates
reported in epidemiological studies. The most relevant information to
consider was the number of health events (e.g., deaths,
hospitalizations) occurring within a study population in relation to
the distribution of PM2.5 concentrations likely experienced
by study participants. However, in recognizing that access to health
event data may be restricted, and consistent with advice from CASAC
(Samet 2010d, p. 2), EPA staff also considered the number of
participants within each study area, in relation to the distribution of
PM2.5 concentrations (i.e., study population data), as a
surrogate for health event data.
In applying this approach, the Policy Assessment focused on
identifying the part of the distribution of PM2.5
concentrations which had the most influence on generating health effect
estimates in epidemiological studies, as discussed in section III.A.3
above. As discussed below, in working with study investigators, the EPA
staff was able to obtain health event data for three large multi-city
studies (Krewski et al., 2009; Zanobetti and Schwartz, 2009; Bell et
al., 2008) and population data for the same three studies and one
additional long-term exposure study (Miller et al., 2007), as
documented in a staff memorandum (Rajan et al., 2011).\77\ For the
three studies for which both health event and study population data
were available, the EPA staff analyzed the reliability of using study
population data as a surrogate for health event data. Based on these
analyses, the EPA staff recognized that the 10th and 25th percentiles
of the health event and study population distributions are nearly
identical and concluded that the distribution of population data can be
a useful surrogate for event data, providing support for consideration
of the study population data for Miller et al. (2007), for which health
event data were not available (Rajan et al., 2011, Analysis 1 and
Analysis 2, in particular, Table 1 and Figures 1 and 2).
---------------------------------------------------------------------------

\77\ The distributional statistical analysis of population-level
data built upon an earlier analysis that evaluated the distributions
of air quality and associated population data for three long-term
exposure studies and three short-term exposure studies (Schmidt et
al., 2010, Analysis 2).
---------------------------------------------------------------------------

With regard to the long-term mean PM2.5 concentrations
which are relevant to the first approach, Figures 1 through 3 (U.S.
EPA, 2011a, Figures 2-4, 2-5, 2-6, and 2-8) summarize data available
for multi-city, long- and short-term exposure studies that evaluated
endpoints classified in the Integrated Science Assessment as having
evidence of a causal or likely causal relationship or evidence
suggestive of a causal relationship, showing the studies with long-term
mean PM2.5 concentrations below 17 [mu]g/m\3\.\78\ As
discussed in more detail in section III.E.4.b of the proposal, Figures
1 and 3 summarize the health outcomes evaluated, relative risk
estimates, air quality data, and geographic scope for long- and short-
term exposure studies, respectively, that evaluated mortality (evidence
of a causal relationship); cardiovascular effects (evidence of a causal
relationship); and respiratory effects (evidence of a likely causal
relationship) in the general population, as well as in older adults, an
at-risk population. Figure 2 provides this same summary information for
long-term exposure studies that evaluated respiratory effects (evidence
of a likely causal relationship) in children, an at-risk population, as
well as developmental effects (evidence suggestive of a causal
relationship).
---------------------------------------------------------------------------

\78\ Additional studies presented and assessed in the Integrated
Science Assessment report effects at higher long-term mean
PM2.5 concentrations (e.g., U.S. EPA, 2009a, Figures 2-1,
2-2, 7-6, and 7-7).
\79\ The long-term mean PM2.5 concentrations reported
by the study authors for the Miller et al. (2007) and Lipfert et al.
(2006a) studies are discussed more fully in the Response to Comments
document (U.S. EPA, 2012a).

---------------------------------------------------------------------------

[[Page 3131]]

[GRAPHIC] [TIFF OMITTED] TR15JA13.000

[[Page 3132]]

[GRAPHIC] [TIFF OMITTED] TR15JA13.001

[[Page 3133]]

[GRAPHIC] [TIFF OMITTED] TR15JA13.002

With regard to consideration of additional information from
epidemiological studies which was relevant to the second approach, the
EPA staff compiled a summary of the range of PM2.5
concentrations

[[Page 3134]]

corresponding with the 25th to 10th percentiles of health event or
study population data from the four multi-city studies, for which
distributional statistics are available \80\ (U.S. EPA, 2011a, Figure
2-7; Rajan et al., 2011, Table 1). By considering this approach, one
could focus on the range of PM2.5 concentrations below the
long-term mean ambient concentrations over which we continue to have
confidence in the associations observed in epidemiological studies
(e.g., above the 25th percentile) where commensurate public health
protection could be obtained for PM2.5-related effects and,
conversely, identify the range in the distribution below which our
confidence in the associations is appreciably less, to identify
alternative annual standard levels.
---------------------------------------------------------------------------

\80\ The EPA staff obtained health event data (e.g., number of
deaths, hospitalizations) occurring in a study population for three
multi-city studies (Krewski et al., 2009; Zanobetti and Schwartz,
2009; Bell et al., 2008) and study population data were obtained for
the same three studies and one additional study (Miller et al.,
2007) (U.S. EPA, 2011a, p. 2-71). If health event or study
population data were available for additional studies, the EPA could
employ distributional statistics to identify the broader range of
PM2.5 concentrations that were most influential in
generating health effect estimates in those studies.
---------------------------------------------------------------------------

The mean PM2.5 concentrations associated with the
studies summarized in Figures 1, 2, and 3 and with the distributional
statistics analyses (Rajan et al., 2011) are based on concentrations
averaged across ambient monitors within each area included in a given
study and then averaged across study areas to calculate an overall
study mean concentration, as discussed above. Figure 4, discussed in
more detail in section III.E.4.a of the proposal, summarizes
statistical metrics for those key studies \81\ included in Figures 1,
2, and 3 that provide evidence of positive and generally statistically
significant PM2.5-related effects, which are relevant to the
two approaches for translating epidemiological evidence into potential
standard levels as discussed above. The top of Figure 4 includes
information for long-term exposure studies evaluating health outcomes
classified as having evidence of a causal or likely causal relationship
with PM2.5 exposures (long-term mean PM2.5
concentrations indicated by diamond symbols). The middle of Figure 4
includes information for short-term exposure studies evaluating health
outcomes classified as having evidence of a causal or likely causal
relationship with PM2.5 exposures (long-term mean
PM2.5 concentrations indicated by triangle symbols). The
bottom of Figure 4 includes information for long-term exposures studies
evaluating health outcomes classified as having evidence suggestive of
a causal relationship (long-term mean PM2.5 concentrations
indicated by square symbols). Figure 4 also summarizes the range of
PM2.5 concentrations corresponding with the 25th (indicated
by solid circles) to 10th (indicated by open circles) percentiles of
the health event or study population data from the four multi-city
studies (highlighted in bold text) for which distributional statistics
are available.
---------------------------------------------------------------------------

\81\ Long- and short-term exposure studies considered ``key''
studies for consideration are summarized in Figure 4 and include
those studies observing effects for which the evidence supported a
causal or likely causal association. This figure represents the
subset of multi-city studies included in Figures 1 through 3 that
provided evidence of positive and generally statistically
significant effects associated in whole or in part with more recent
air quality data, generally representing health effects associated
with lower PM2.5 concentrations than had previously been
considered in the last review. The EPA notes that many of these
studies evaluated multiple health endpoints, and not all of the
effects evaluated provided evidence of positive and statistically
significant effects. For purposes of informing the Administrator's
decision on the appropriate standard levels, the Agency considers
the full body of scientific evidence and focuses on those aspects of
the key studies that provided evidence of positive and generally
statistically significant effects.
\82\ The long-term mean PM2.5 concentrations reported
by the study authors for the Miller et al. (2007) and Lipfert et al.
(2006a) studies are discussed more fully in the Response to Comments
document (U.S. EPA, 2012a).

---------------------------------------------------------------------------

[[Page 3135]]

[GRAPHIC] [TIFF OMITTED] TR15JA13.003

In considering the evidence, the Policy Assessment recognized that
NAAQS are standards set so as to provide requisite protection, neither
more nor less stringent than necessary to protect public health, with
an adequate margin of safety. This judgment, ultimately made by the
Administrator, involves weighing the strength of the evidence and the
inherent uncertainties and limitations of that evidence. Therefore,
depending on

[[Page 3136]]

the weight placed on different aspects of the evidence and inherent
uncertainties, consideration of different alternative standard levels
could be supported.
Given the currently available evidence discussed in more detail in
section III.E.4.b of the proposal and considering the various
approaches discussed above, the Policy Assessment concluded it was
appropriate to focus on an annual standard level within a range of
about 12 to 11 [mu]g/m\3\ (U.S. EPA, 2011a, pp. 2-82, 2-101, and 2-
106). As illustrated in Figure 4, the Policy Assessment recognized that
a standard level of 12 [mu]g/m\3\, at the upper end of this range, was
somewhat below the long-term mean PM2.5 concentrations
reported in all the multi-city, long- and short-term exposure studies
that provided evidence of positive and statistically significant
associations with health effects classified as having evidence of a
causal or likely causal relationship, including premature mortality and
hospitalizations and emergency department visits for cardiovascular and
respiratory effects as well as respiratory effects in children.
Further, a level of 12 [mu]g/m\3\ would reflect consideration of
additional population-level information from such epidemiological
studies in that it generally corresponded with approximately the 25th
percentile of the available distributions of health events data in the
studies for which population-level information was available. In
addition, a level of 12 [mu]g/m\3\ would reflect some consideration of
studies that provided more limited evidence of reproductive and
developmental effects, which were suggestive of a causal relationship,
in that it was about at the same level as the lowest long-term mean
PM2.5 concentrations reported in such studies (see Figure
4).
Alternatively the Policy Assessment recognized that an annual
standard level of 11 [mu]g/m\3\, at the lower end of this range, was
well below the lowest long-term mean PM2.5 concentrations
reported in all multi-city long- and short-term exposure studies that
provide evidence of positive and statistically significant associations
with health effects classified as having evidence of a causal or likely
causal relationship. A level of 11 [mu]g/m\3\ would reflect placing
more weight on the distributions of health event and population data,
in that this level was within the range of PM2.5
concentrations corresponding to the 25th and 10th percentiles of all
the available distributions of such data. In addition, a level of 11
[mu]g/m\3\ was somewhat below the lowest long-term mean
PM2.5 concentrations reported in reproductive and
developmental effects studies that are suggestive of a causal
relationship. Thus, a level of 11 [mu]g/m\3\ would reflect an approach
to translating the available evidence that places relatively more
emphasis on margin of safety considerations and less certain causal
relationships than would a standard set at a higher level. Such a
policy approach would tend to weigh uncertainties in the evidence in
such a way as to avoid potentially underestimating PM2.5-
related risks to public health. Further, recognizing the uncertainties
inherent in identifying any particular point at which our confidence in
reported associations becomes appreciably less, the Policy Assessment
concluded that the available evidence did not provide a sufficient
basis to consider alternative annual standard levels below 11 [mu]g/
m\3\ (U.S. EPA, 2011a, p. 2-81).
The Policy Assessment also considered the extent to which the
available evidence provided a basis for considering alternative annual
standard levels above 12 [mu]g/m\3\. As discussed below, the Policy
Assessment concluded that it could be reasonable to consider a standard
level up to 13 [mu]g/m\3\ based on a policy approach that weighed
uncertainties in the evidence in such a way as to avoid potentially
overestimating PM2.5-related risks to public health,
especially to the extent that primary emphasis was placed on long-term
exposure studies as a basis for an annual standard level. A level of 13
[mu]g/m\3\ was somewhat below the long-term mean PM2.5
concentrations reported in all but one of the long-term exposure
studies providing evidence of positive and statistically significant
associations with PM2.5-related health effects classified as
having a causal or likely causal relationship. As shown in Figure 4,
the one long-term exposure study with a long-term mean PM2.5
concentration just below 13 [mu]g/m\3\ was the Miller et al., (2007)
study. However, as noted in section III.D.1.a of the proposal and
discussed in more detail in the Response to Comments document, the
Policy Assessment observed that in comparison to other long-term
exposure studies, the Miller et al. study was more limited in that it
was based on only one year of air quality data and the one year was
after the health outcomes were reported (U.S. EPA, 2011a, pp. 2-81 to
2-82). Thus, to the extent that less weight was placed on the Miller et
al. study than on other long-term exposure studies with more robust air
quality data, a level of 13 [mu]g/m\3\ could be considered as being
protective of long-term exposure related effects classified as having a
causal or likely causal relationship. In also considering short-term
exposure studies, however, the Policy Assessment noted that a level of
13 [mu]g/m\3\ was below the long-term mean PM2.5
concentrations reported in most but not all such studies. In
particular, two studies--Burnett et al. (2004) and Bell et al. (2008)--
reported long-term mean PM2.5 concentrations of 12.8 and
12.9 [mu]g/m\3\, respectively. In considering these studies, the Policy
Assessment found no basis to conclude that these two studies were any
more limited or uncertain than the other short-term exposure studies
shown in Figures 3 and 4 (U.S. EPA, 2011a, p. 2-82). On this basis, as
discussed below, the Policy Assessment concluded that consideration of
an annual standard level of 13 [mu]g/m\3\ would have implications for
the degree of protection that would need to be provided by the 24-hour
standard, in order that the suite of PM2.5 standards, taken
together, would provide appropriate protection from effects on public
health related to short-term exposure to PM2.5 (U.S. EPA,
2011a, p. 2-82).
The Policy Assessment also noted that a standard level of 13 [mu]g/
m\3\ would reflect a judgment that the uncertainties in the
epidemiological evidence as summarized in section III.B above and
discussed in more detail in section III.B.2 of the proposal, including
uncertainties related to the heterogeneity observed in the
epidemiological studies in the eastern versus western parts of the
U.S., the relative toxicity of PM2.5 components, and the
potential role of co-pollutants, are too great to warrant placing any
weight on the distributions of health event and population data that
extend down below the long-term mean concentrations into the lower
quartile of the data. This level would also reflect a judgment that the
evidence from reproductive and developmental effects studies that is
suggestive of a causal relationship was too uncertain to support
consideration of any lower level.
Beyond evidence-based considerations, the Policy Assessment also
considered the extent to which the quantitative risk assessment
supported consideration of these alternative standard levels or
provided support for lower levels. In considering simulations of just
meeting alternative annual standard levels within the range of 13 to 11
[mu]g/m\3\ (in conjunction with the current 24-hour standard level of
35 [mu]g/m\3\), the Policy Assessment concluded that important public
health improvements are associated with risk

[[Page 3137]]

reductions estimated for standard levels of 13 and 12 [mu]g/m\3\ and
noted that the level of 11 [mu]g/m\3\ was not included in the
quantitative risk assessment. The Policy Assessment noted that the
overall confidence in the quantitative risk estimates varied for the
different alternative standard levels evaluated and was stronger for
the higher levels and substantially lower for the lowest level
evaluated (i.e., 10 [mu]g/m\3\). Based on the above considerations, the
Policy Assessment concluded that the quantitative risk assessment
provided support for considering alternative annual standard levels
within a range of 13 to 11 [mu]g/m\3\, but did not provide strong
support for considering lower alternative standard levels (U.S. EPA,
2011a, pp. 2-102 to 2-103).
Taken together, the Policy Assessment concluded that consideration
of alternative annual standard levels in the range of 13 to 11 [mu]g/
m\3\ may be appropriate. Furthermore, the Policy Assessment concluded
that the currently available evidence most strongly supported
consideration of an alternative annual standard level in the range of
12 to 11 [mu]g/m\3\ (U.S. EPA, 2011a, p. 2-82). The Policy Assessment
concluded that an alternative level within the range of 12 to 11 [mu]g/
m\3\ would more fully take into consideration the available information
from all long- and short-term PM2.5 exposure studies,
including studies of at-risk populations, than would a higher level.
This range also reflected placing weight on information from studies
that helped to characterize the range of PM2.5
concentrations over which we continue to have confidence in the
associations observed in epidemiological studies, as well as the extent
to which our confidence in the associations was appreciably less at
lower concentrations.
As recognized in sections III.A.3 and III.E.4.a above, an annual
standard intended to serve as the primary means for providing
protection from effects associated with both long- and short-term
PM2.5 exposures is not expected to provide appropriate
protection against the effects of all short-term PM2.5
exposures (unless established at a level so low as to undoubtedly
provide more protection than necessary for long-term exposures). Of
particular concern are areas with high peak-to-mean ratios possibly
associated with strong local or seasonal sources, or PM2.5-
related effects that may be associated with shorter-than-daily exposure
periods. As a result, the Policy Assessment concluded that it was
appropriate to consider alternative 24-hour PM2.5 standard
levels that would supplement the protection provided by an annual
standard.
As outlined in section III.A.3 above, the Policy Assessment
considered the available evidence from short-term PM2.5
exposure studies, as well as the uncertainties and limitations in that
evidence, to assess the degree to which alternative annual and 24-hour
PM2.5 standards can be expected to reduce the estimated
risks attributed to short-term fine particle exposures. In considering
the available epidemiological evidence, the Policy Assessment took into
account information from multi-city studies as well as single-city
studies. The Policy Assessment considered the distributions of 24-hour
PM2.5 concentrations reported in short-term exposure
studies, focusing on the 98th percentile concentrations to match the
form of the 24-hour standard as discussed in section III.E.3.b above.
In recognizing that the annual and 24-hour standards work together to
provide protection from effects associated with short-term
PM2.5 exposures, the Policy Assessment also considered
information on the long-term mean PM2.5 concentrations from
these studies.
In addition to considering the epidemiological evidence, the Policy
Assessment considered air quality information, specifically peak-to-
mean ratios using county-level 24-hour and annual design values, to
characterize air quality patterns in areas possibly associated with
strong local or seasonal sources. These patterns helped in
understanding the extent to which different combinations of annual and
24-hour standards would be consistent with the policy goal of setting a
generally controlling annual standard with a 24-hour standard that
provides supplemental protection especially for areas with high peak-
to-mean ratios (U.S. EPA, 2011a, p. 2-14).
In considering the information provided by the short-term exposure
studies, the Policy Assessment recognized that to the extent these
studies were conducted in areas that likely did not meet one or both of
the current standards, such studies did not help inform the
characterization of the potential public health improvements of
alternative standards set at lower levels. Therefore, in considering
the short-term exposure studies to inform staff conclusions regarding
levels of the 24-hour standard that are appropriate to consider, the
Policy Assessment placed greatest weight on studies conducted in areas
that likely met both the current annual and 24-hour standards.
With regard to multi-city studies that evaluated effects associated
with short-term PM2.5 exposures, as summarized in Figure 3
above and discussed in more detail in section III.E.4.c of the
proposal, the Policy Assessment noted that, to the extent air quality
distributions were reduced to reflect just meeting the current 24-hour
standard, additional protection would be anticipated for the effects
observed in the three multi-city studies with 98th percentile values
greater than 35 [mu]g/m\3\ (Burnett et al., 2004; Burnett and Goldberg,
2003; Franklin et al., 2008). In the three additional studies with 98th
percentile values below 35 [mu]g/m\3\, specifically 98th percentile
concentrations of 34.2, 34.3, and 34.8 [mu]g/m\3\, the Policy
Assessment noted that these studies reported long-term mean
PM2.5 concentrations of 12.9, 13.2, and 13.4 [mu]g/m\3\,
respectively (Bell et al., 2008; Zanobetti and Schwartz, 2009; Dominici
et al., 2006a). To the extent that consideration was given to revising
the level of the annual standard, as discussed in section III.E.4.b of
the proposal, the Policy Assessment recognized that potential changes
associated with meeting such an alternative annual standard would
result in lowering risks associated with both long- and short-term
PM2.5 exposures. Consequently, in considering a 24-hour
standard that would operate in conjunction with an annual standard to
provide appropriate public health protection, the Policy Assessment
noted that to the extent that the level of the annual standard was
revised to within a range of 13 to 11 [mu]g/m\3\, in particular in the
range of 12 to 11 [mu]g/m\3\, additional protection would be provided
for the long-term effects observed in these multi-city studies (U.S.
EPA, 2011a, p. 2-84).
Based on this information, the Policy Assessment concluded that the
multi-city, short-term exposure studies generally provided support for
retaining the 24-hour standard level at 35 [mu]g/m\3\ so long as the
standard is in conjunction with an annual standard level revised to
within a range of 12 to 11 [mu]g/m\3\ (U.S. EPA, 2011a, p. 2-84).
Alternatively, in conjunction with an annual standard level of 13
[mu]g/m\3\, the Policy Assessment concluded that the multi-city studies
provided limited support for revising the 24-hour standard level
somewhat below 35 [mu]g/m\3\, such as down to 30 [mu]g/m\3\, based on
one study (Bell et al., 2008) that reported positive and statistically
significant effects with an overall 98th percentile value below the
level of the current 24-hour standard and an overall long-term mean
concentration slightly less than 13 [mu]g/m\3\ (Figure 3; U.S. EPA,
2011a, p. 2-84).
In reaching staff conclusions regarding alternative 24-hour
standard levels that were appropriate to consider,

[[Page 3138]]

the Policy Assessment also took into account relevant information from
single-city studies that evaluated effects associated with short-term
PM2.5 exposures. The Policy Assessment recognized that these
studies may provide additional insights regarding impacts on at-risk
populations and/or on areas with isolated peak concentrations.
As discussed in more detail in section III.E.4.c of the proposal,
although a number of single-city studies reported effects at
appreciably lower PM2.5 concentrations than multi-city
short-term exposure studies, the uncertainties and limitations
associated with the single-city studies were considerably greater than
those associated with the multi-city studies and, thus, the Policy
Assessment concluded there was less confidence in using these studies
as a basis for setting the level of a standard. Therefore, the Policy
Assessment concluded that the multi-city short-term exposure studies
provided the strongest evidence to inform decisions on the level of the
24-hour standard, and the single-city studies did not warrant
consideration of 24-hour standard levels different from those supported
by the multi-city studies (U.S. EPA, 2011a, p. 2-88).
In addition to considering the epidemiological evidence, the Policy
Assessment took into account air quality information based on county-
level 24-hour and annual design values to understand the public health
implications of the alternative standard levels supported by the
currently available scientific evidence, as discussed in this section.
Consistent with the general approach discussed in section III.A.3
above, the Policy Assessment considered the extent to which different
combinations of alternative annual and 24-hour standard levels based on
the evidence would support the policy goal of lowering annual and 24-
hour air quality distributions by using the annual standard to be the
``generally controlling'' standard in conjunction with setting the 24-
hour standard to provide supplemental protection (U.S. EPA, 2011a, pp
2-88 to 2-91, Figure 2-10).
Using information on the relationship of the 24-hour and annual
design values, the Policy Assessment examined the implications of three
alternative suites of PM2.5 standards identified as
appropriate to consider based on the currently available scientific
evidence, as discussed above. The Policy Assessment concluded that an
alternative suite of PM2.5 standards that would include an
annual standard level of 11 or 12 [mu]g/m\3\ and a 24-hour standard
with a level of 35 [mu]g/m\3\ (i.e., 11/35 or 12/35) would result in
the annual standard being the generally controlling standard in most
areas although the 24-hour standard would continue to be the generally
controlling standard in the Northwest (U.S. EPA, 2011a, pp. 2-89 to 2-
91 and Figure 2-10). These Northwest counties generally represented
areas where the annual mean PM2.5 concentrations have
historically been low but where relatively high 24-hour concentrations
occur, often related to seasonal wood smoke emissions. Alternatively,
combining an alternative annual standard of 13 [mu]g/m\3\ with a 24-
hour standard of 30 [mu]g/m\3\ would result in many more areas across
the country in which the 24-hour standard would likely become the
controlling standard (the standard driving air quality distributions
lower) than if an alternative annual standard of 12 or 11 [mu]g/m\3\
were paired with the current level of the 24-hour standard (i.e., 35
[mu]g/m\3\).
The Policy Assessment concluded that consideration of retaining the
24-hour standard level at 35 [mu]g/m\3\ would reflect placing greatest
weight on evidence from multi-city studies that reported positive and
statistically significant associations with health effects classified
as having a causal or likely causal relationship. In conjunction with
lowering the annual standard level, especially within a range of 12 to
11 [mu]g/m\3\, this alternative recognized additional public health
protection against effects associated with short-term PM2.5
exposures which would be provided by lowering the annual standard such
that revision to the 24-hour standard would not be warranted (U.S. EPA,
2011a, p. 2-91).
Beyond evidence-based considerations, the Policy Assessment also
considered the extent to which the quantitative risk assessment
supported consideration of retaining the current 24-hour standard level
or provided support for lower standard levels. In considering
simulations of just meeting the current 24-hour standard level of 35
[mu]g/m\3\ or alternative levels of 30 or 25 [mu]g/m\3\ (in conjunction
with alternative annual standard levels within a range of 13 to 11
[mu]g/m\3\), the Policy Assessment noted that the overall confidence in
the quantitative risk estimates varied for the different standard
levels evaluated and was stronger for the higher levels and
substantially lower for the lowest level evaluated (i.e., 25 [mu]g/
m\3\). Based on this information, the Policy Assessment concluded that
the quantitative risk assessment provided support for considering a 24-
hour standard level of 35 or 30 [mu]g/m\3\ (in conjunction with an
alternative standard level within a range of 13 to 11 [mu]g/m\3\) but
did not provide strong support for considering lower alternative 24-
hour standard levels (U.S. EPA, 2011a, pp. 2-102 to 2-103).
Taken together, the Policy Assessment concluded that while it was
appropriate to consider an alternative 24-hour standard level within a
range of 35 to 30 [mu]g/m\3\, the currently available evidence most
strongly supported consideration for retaining the current 24-hour
standard level at 35 [mu]g/m\3\ in conjunction with lowering the level
of the annual standard within a range of 12 to 11 [mu]g/m\3\ (U.S. EPA,
2011a, p. 2-92).
ii. CASAC Advice
Based on its review of the second draft Policy Assessment, CASAC
agreed with the general approach for translating the available
epidemiological evidence, risk information, and air quality information
into the basis for reaching conclusions on alternative standards for
consideration. Furthermore, CASAC agreed ``that it is appropriate to
return to the strategy used in 1997 that considers the annual and the
short-term standards together, with the annual standard as the
controlling standard, and the short-term standard supplementing the
protection afforded by the annual standard'' and ``considers it
appropriate to place the greatest emphasis'' on health effects judged
to have evidence supportive of a causal or likely causal relationship
as presented in the Integrated Science Assessment (Samet, 2010d, p. 1).
CASAC concluded that the range of levels presented in the second
draft Policy Assessment (i.e., alternative annual standard levels
within a range of 13 to 11 [mu]g/m\3\ and alternative 24-hour standard
levels within a range of 35 to 30 [mu]g/m\3\) ``are supported by the
epidemiological and toxicological evidence, as well as by the risk and
air quality information compiled'' in the Integrated Science
Assessment, Risk Assessment, and second draft Policy Assessment. CASAC
further noted that ``[a]lthough there is increasing uncertainty at
lower levels, there is no evidence of a threshold (i.e., a level below
which there is no risk for adverse health effects)'' (Samet, 2010d, p.
ii).
Although CASAC supported the alternative standard level ranges
presented in the second draft Policy Assessment, it did not express
support for any specific levels or combinations of standards. Rather,
CASAC encouraged the EPA to develop a clearer rationale in the final
Policy Assessment for staff conclusions regarding annual

[[Page 3139]]

and 24-hour standards that were appropriate to consider, including
consideration of the combination of these standards supported by the
available information (Samet, 2010d, p. ii). Specifically, in
commenting on a distributional statistical analysis of air quality and
associated population data presented in the second draft Policy
Assessment, CASAC encouraged staff to focus on information related to
the concentrations that were most influential in generating the health
effect estimates in individual studies to inform alternative standard
levels. CASAC urged that the EPA redo that analysis using health event
or study population data (Samet, 2010d, p. 2). CASAC also commented
that the approach presented in the second draft Policy Assessment to
identify alternative 24-hour standard levels which focused on peak-to-
mean ratios was not relevant for informing the actual level (Samet
2010d, p. 4). Further, they expressed the concern that the combinations
of annual and 24-hour standard levels discussed in the second draft
Policy Assessment (i.e., in the range of 13 to 11 [mu]g/m\3\ for the
annual standard, in conjunction with retaining the current 24-hour
PM2.5 standard level of 35 [mu]g/m\3\; alternatively,
revising the level of the 24-hour standard to 30 [mu]g/m\3\ in
conjunction with an annual standard level of 11 [mu]g/m\3\) ``may not
be adequately inclusive'' and ``[i]t was not clear why, for example a
daily standard of 30 [mu]g/m\3\ should only be considered in
combination with an annual level of 11 [mu]g/m\3\'' (Samet, 2010d, p.
ii). CASAC encouraged the EPA to more clearly explain its rationale for
identifying the 24-hour/annual combinations that are appropriate for
consideration (Samet 2010d, p. ii).
In considering CASAC's advice as well as public comment on the
second draft Policy Assessment, the EPA staff conducted additional
analyses and modified their conclusions regarding alternative standard
levels that were appropriate to consider. The staff conclusions in the
final Policy Assessment (U.S. EPA, 2011a, section 2.3.4.4) differed
somewhat from the alternative standard levels discussed in the second
draft Policy Assessment (U.S. EPA, 2010f, section 2.3.4.3), upon which
CASAC based its advice. Changes made in the final Policy Assessment
were primarily focused on improving and clarifying the approach for
translating the epidemiological evidence into a basis for staff
conclusions on the broadest range of alternative standard levels
supported by the available scientific information and more clearly
articulating the rationale for the staff's conclusions (Wegman, 2011,
pp. 1 to 2). Consistent with CASAC's advice to consider more
information from epidemiological studies, as discussed in section
III.E.4.b.1 above, the EPA analyzed additional population-level data
obtained from several study authors (Rajan et al., 2011). In
transmitting the final Policy Assessment to CASAC, the Agency notified
CASAC that the final staff conclusions reflected consideration of
CASAC's advice and that those staff conclusions were based, in part, on
the specific distributional analysis that CASAC had urged the EPA to
conduct (Wegman, 2011, p.2). Thus, CASAC had an opportunity to comment
on the final Policy Assessment, but chose not to provide any additional
comments or advice after receiving it.
iii. Administrator's Proposed Decisions on the Primary PM2.5
Standard Levels
In reaching her conclusions regarding appropriate alternative
standard levels to consider, the Administrator considered the
epidemiological and other scientific evidence, estimates of risk
reductions associated with just meeting alternative annual and/or 24-
hour standards, air quality analyses, related limitations and
uncertainties, staff conclusions as presented in the Policy Assessment,
and the advice of CASAC. As an initial matter, the Administrator agreed
with the general approach discussed in the Policy Assessment as
summarized in sections III.A.3 and III.E.4.a above, and supported by
CASAC, of considering the protection afforded by the annual and 24-hour
standards taken together for mortality and morbidity effects associated
with both long- and short-term exposures to PM2.5 (77 FR
38939). Furthermore, based on the evidence and quantitative risk
assessment, the Administrator provisionally concluded it is appropriate
to set a ``generally controlling'' annual standard that will lower a
wide range of ambient 24-hour concentrations, with a 24-hour standard
focused on providing supplemental protection, particularly for areas
with high peak-to-mean ratios possibly associated with strong local or
seasonal sources, or PM2.5-related effects that may be
associated with shorter-than daily exposure periods. The Administrator
provisionally concluded this approach would likely reduce aggregate
risks associated with both long- and short-term exposures more
consistently than a generally controlling 24-hour standard and would be
the most effective and efficient way to reduce total PM2.5-
related population risk. Id.
In reaching decisions on alternative standard levels to propose,
the Administrator judged that it was most appropriate to examine where
the evidence of associations observed in the epidemiological studies
was strongest and, conversely, where she had appreciably less
confidence in the associations observed in the epidemiological studies.
Based on the characterization and assessment of the epidemiological and
other studies presented and assessed in the Integrated Science
Assessment and the Policy Assessment, the Administrator recognized the
substantial increase in the number and diversity of studies available
in this review including extended analyses of the seminal studies of
long-term PM2.5 exposures (i.e., ACS and Harvard Six Cities
studies) as well as important new long-term exposure studies (as
summarized in Figures 1 and 2). Collectively, the Administrator noted
that these studies, along with evidence available in the last review,
provided consistent and stronger evidence of an association with
premature mortality, with the strongest evidence related to
cardiovascular-related mortality, at lower ambient concentrations than
previously observed. The Administrator also recognized the availability
of stronger evidence of morbidity effects associated with long-term
PM2.5 exposures, including evidence of cardiovascular
effects from the WHI study and respiratory effects, including decreased
lung function growth, from the extended analyses for the Southern
California Children's Health Study. Furthermore, the Administrator
recognized new U.S. multi-city studies that greatly expanded and
reinforced our understanding of mortality and morbidity effects
associated with short-term PM2.5 exposures, providing
stronger evidence of associations at ambient concentrations similar to
those previously observed (as summarized in Figure 3). Id. at 38939-40.
The newly available scientific evidence built upon the previous
scientific data base to provide evidence of generally robust
associations and to provide a basis for greater confidence in the
reported associations than in the last review. The Administrator
recognized that the weight of evidence, as evaluated in the Integrated
Science Assessment, was strongest for health endpoints classified as
having evidence of a causal relationship. These relationships included
those between long- and short-term PM2.5 exposures and
mortality and cardiovascular effects. She recognized that the weight of
evidence was also

[[Page 3140]]

strong for health endpoints classified as having evidence of a likely
causal relationship, which included those between long- and short-term
PM2.5 exposures and respiratory effects. In addition, the
Administrator made note of the much more limited evidence for health
endpoints classified as having evidence suggestive of a causal
relationship, including developmental, reproductive and carcinogenic
effects. Id. at 38940.
Based on information discussed and presented in the Integrated
Science Assessment, the Administrator recognized that health effects
may occur over the full range of concentrations observed in the long-
and short-term epidemiological studies and that no discernible
threshold for any effects can be identified based on the currently
available evidence (U.S. EPA, 2009a, section 2.4.3). She also
recognized, in taking note of CASAC advice and the distributional
statistics analysis discussed in section III.E.4.b.i above and in the
Policy Assessment, that there was significantly greater confidence in
observed associations over certain parts of the air quality
distributions in the studies, and conversely, that there was
significantly diminished confidence in ascribing effects to
concentrations toward the lower part of the distributions.
Consistent with the general approach summarized in section III.A.3
above, and supported by CASAC as discussed in section III.E.4.a above,
the Administrator generally agreed that it was appropriate to consider
a level for an annual standard that was somewhat below the long-term
mean PM2.5 concentrations reported in long- and short-term
exposure studies. In recognizing that the evidence of an association in
any such study was strongest at and around the long-term average where
the data in the study are most concentrated, she understood that this
approach did not provide a bright line for reaching decisions about
appropriate standard levels. The Administrator noted that long-term
mean PM2.5 concentrations were available for each study
considered and, therefore, represented the most robust data set to
inform her decisions on appropriate annual standard levels. She also
noted that the overall study mean PM2.5 concentrations were
generally calculated based on monitored concentrations averaged across
monitors in each study area with multiple monitors, referred to as a
composite monitor concentration, in contrast to the highest
concentration monitored in each study area, referred to as a maximum
monitor concentration, which are used to determine whether an area
meets a given standard. In considering such long-term mean
concentrations, the Administrator understood that it was appropriate to
consider the weight of evidence for the health endpoints evaluated in
such studies in giving weight to this information. Id.
Based on the information summarized in Figure 4 above and presented
in more detail in the Policy Assessment (U.S. EPA, 2011a, chapter 2)
for effects classified in the Integrated Science Assessment as having a
causal or likely causal relationship with PM2.5 exposures,
the Administrator observed an overall pattern of statistically
significant associations reported in studies of long-term
PM2.5 exposures with long-term mean concentrations ranging
from somewhat above the current standard level of 15 [mu]g/m\3\ down to
the lowest mean concentration in such studies of 12.9 [mu]g/m\3\ (in
Miller et al., 2007).\83\ She observed a similar pattern of
statistically significant associations in studies of short-term
PM2.5 exposures with long-term mean concentrations ranging
from around 15 [mu]g/m\3\ down to 12.8 [mu]g/m\3\ (in Burnett et al.,
2004). With regard to effects classified as providing evidence
suggestive of a causal relationship, the Administrator observed a small
number of long-term exposure studies related to developmental and
reproductive effects that reported statistically significant
associations with overall study mean PM2.5 concentrations
down to 11.9 [mu]g/m\3\ (in Bell et al., 2007).\84\ Id.
---------------------------------------------------------------------------

\83\ The EPA notes that the Miller et al., (2007) study provides
strong evidence of cardiovascular related effects associated with
long-term PM2.5 exposures. At the time of the proposal,
the EPA recognized the limited nature of the air quality data
considered in this study (77 FR 38918, fn. 62). The EPA has reviewed
those limitations, in conjunction with consideration of public
comments received on the proposal as discussed in section III.E.4.c,
in conjunction with reaching a final decision on the level of the
annual standard.
\84\ With respect to suggestive evidence related to cancer,
mutagenic, and genotoxic effects, the PM2.5
concentrations reported in studies generally included ambient
concentrations that are equal to or greater than ambient
concentrations observed in studies that reported mortality and
cardiovascular and respiratory effects (U.S. EPA, 2009a, section
7.5), such that in selecting alternative standard levels that
provide protection from mortality and cardiovascular and respiratory
effects, it is reasonable to anticipate that protection will also be
provided for carcinogenic effects.
---------------------------------------------------------------------------

The Administrator also considered additional information from
epidemiological studies, consistent with CASAC advice, to take into
account the broader distribution of PM2.5 concentrations and
the degree of confidence in the observed associations over the broader
air quality distribution. In considering this additional information,
she understood that the Policy Assessment presented information on the
25th and 10th percentiles of the distributions of PM2.5
concentrations available from four multi-city studies to provide a
general frame of reference as to the part of the distribution in which
the data become appreciably more sparse and, thus, where her confidence
in the associations observed in epidemiological studies would become
appreciably less.
As summarized in Figure 4 above, the Administrator took note of
additional population-level data that were available for four studies
(Krewski et al., 2009; Miller et al., 2007; Bell et al., 2008;
Zanobetti and Schwartz, 2009), each of which reported statistically
significant associations with health endpoints classified as having
evidence of a causal relationship. In considering the long-term
PM2.5 concentrations associated with the 25th percentile
values of the population-level data for these four studies, she
observed that these values ranged from somewhat above to somewhat below
12 [mu]g/m\3\. The Administrator recognized that these studies include
some of the strongest evidence available within the overall body of
scientific evidence and noted that three of these studies (Krewski et
al., 2009; Bell et al., 2008; Zanobetti and Schwartz, 2009) were used
as the basis for concentration-response functions used in the
quantitative risk assessment (U.S. EPA, 2010a, section 3.3.3).
In considering this information, the Administrator noted that CASAC
advised that information about the long-term PM2.5
concentrations that were most influential in generating the health
effect estimates in epidemiological studies can help to inform
selection of an appropriate annual standard level. However, the
Administrator also recognized that additional population-level data
were available for only these four studies and, therefore, she believed
that these studies comprised a more limited data set than one based on
long-term mean PM2.5 concentrations for which data were
available for all studies considered, as discussed above.
The Administrator recognized, as summarized in section III.B above,
that important uncertainties remain in the evidence and information
considered in this review of the primary fine particle standards. These
uncertainties are generally related to understanding the relative
toxicity of the different components in the fine particle mixture, the
role of PM2.5 in the complex ambient mixture, exposure
measurement errors inherent in epidemiological studies based on
concentrations measured at

[[Page 3141]]

fixed monitor sites, and the nature, magnitude, and confidence in
estimated risks related to increasingly lower ambient PM2.5
concentrations. Furthermore, the Administrator noted that
epidemiological studies have reported heterogeneity in responses both
within and between cities and geographic regions across the U.S. She
recognized that this heterogeneity may be attributed, in part, to
differences in fine particle composition in different regions and
cities. The Administrator also recognized that there are additional
limitations associated with evidence for reproductive and developmental
effects, identified as being suggestive of a causal relationship with
long-term PM2.5 exposures, including: the limited number of
studies evaluating such effects; uncertainties related to identifying
the relevant exposure time periods of concern; and limited
toxicological evidence providing little information on the mode of
action(s) or biological plausibility for an association between long-
term PM2.5 exposures and adverse birth outcomes. Id. at
38941.
The Administrator was mindful that considering what standards were
requisite to protect public health with an adequate margin of safety
required public health policy judgments that neither overstated nor
understated the strength and limitations of the evidence or the
appropriate inferences to be drawn from the evidence. In considering
how to translate the available information into appropriate standard
levels, the Administrator weighed the available scientific information
and associated uncertainties and limitations. For the purpose of
determining what standard levels were appropriate to propose, the
Administrator recognized, as did EPA staff in the Policy Assessment,
that there was no single factor or criterion that comprised the sole
``correct'' approach to weighing the various types of available
evidence and information, but rather there were various approaches that
are appropriate to consider. The Administrator further recognized that
different evaluations of the evidence and other information before the
Administrator could reflect placing different weight on the relative
strengths and limitations of the scientific information, and different
judgments could be made as to how such information should appropriately
be used in making public health policy decisions on standard levels.
This recognition led the Administrator to consider various approaches
to weighing the evidence so as to identify appropriate standard levels
to propose. In so doing, the Administrator encouraged extensive public
comment on alternative approaches to weighing the evidence and other
information so as to inform her public health policy judgments before
reaching final decisions on appropriate standard levels.
In considering the available information, the Administrator noted
the advice of CASAC that the currently available scientific
information, including epidemiological and toxicological evidence as
well as risk and air quality information, provided support for
considering an annual standard level within a range of 13 to 11 [mu]g/
m\3\ and a 24-hour standard level within a range of 35 to 30 [mu]g/
m\3\. In addition, the Administrator recognized that the Policy
Assessment concluded that the available evidence and risk-based
information support consideration of annual standard levels in the
range of 13 to 11 [mu]g/m\3\, and that the Policy Assessment also
concluded that the evidence most strongly supported consideration of an
annual standard level in the range of 12 to 11 [mu]g/m\3\. In
considering how the annual and 24-hour standards work together to
provide appropriate public health protection, the Administrator
observed that CASAC did not express support for any specific levels or
combinations of standards within these ranges. Nor did CASAC choose to
comment on additional information and analyses presented in the final
Policy Assessment prepared in response to CASAC's recommendations on
the second draft Policy Assessment (Wegman, 2011).
In considering the extent to which the currently available evidence
and information provided support for specific standard levels within
the ranges identified by CASAC and the Policy Assessment as appropriate
for consideration, the Administrator initially considered standard
levels within the range of 13 to 11 [mu]g/m\3\ for the annual standard.
In so doing, the Administrator first considered the long-term mean
PM2.5 concentrations reported in studies of effects
classified as having evidence of a causal or likely causal
relationship, as summarized in Figure 4 above and discussed more
broadly above. She noted that a level at the upper end of this range
would be below most but not all the overall study mean concentrations
from the multi-city studies of long- and short-term exposures, whereas
somewhat lower levels within this range would be below all such overall
study mean concentrations. In considering the appropriate weight to
place on this information, the Administrator again noted that the
evidence of an association in any such study was strongest at and
around the long-term average where the data in the study are most
concentrated, and that long-term mean PM2.5 concentrations
were available for each study considered and, therefore, represented
the most robust data set to inform her decisions on appropriate annual
standard levels. Further, she was mindful that this approach did not
provide a bright line for reaching decisions about appropriate standard
levels. Id.
In considering the long-term mean PM2.5 concentrations
reported in studies of effects classified as having evidence suggestive
of a causal relationship, as summarized in Figure 4 for reproductive
and developmental effects, the Administrator noted that a level at the
upper end of this range would be below the overall study mean
concentration in one of the three studies, while levels in the mid- to
lower part of this range would be below the overall study mean
concentrations in two or three of these studies. In considering the
appropriate weight to place on this information, the Administrator
noted the very limited nature of this evidence of such effects and the
additional uncertainties in these epidemiological studies relative to
the studies that provide evidence of causal or likely causal
relationships.
The Administrator also considered the distributional analyses of
population-level information that were available from four of the
epidemiological studies that provide evidence of effects identified as
having a causal relationship with long- or short-term PM2.5
concentrations for annual standard levels within the same range of 13
to 11 [mu]g/m\3\. In so doing, the Administrator first noted that a
level in the mid-part of this range generally corresponds with
approximately the 25th percentile of the distributions of health events
data available in three of these studies. The Administrator also noted
that standard levels toward the upper part of this range would reflect
placing substantially less weight on this information, whereas standard
levels toward the lower part of this range would reflect placing
substantially more weight on this information. In considering this
information, the Administrator noted that there was no bright line that
delineates the part of the distribution of PM2.5
concentrations within which the data become appreciably more sparse
and, thus, where her confidence in the associations observed in
epidemiological studies became appreciably less.
In considering mean PM2.5 concentrations and
distributional

[[Page 3142]]

analyses from the various sets of epidemiological studies noted above,
the Administrator was mindful, as noted above, that such studies
typically report concentrations based on composite monitor
distributions, in which concentrations may be averaged across multiple
ambient monitors that may be present within each area included in a
given study. Thus, a policy approach that used data based on composite
monitors to identify potential alternative standard levels would
inherently build in a margin of safety of some degree relative to an
alternative standard level based on measurements at the monitor within
an area that records the highest concentration, or the maximum monitor,
since once a standard was set, concentrations at appropriate maximum
monitors within an area were generally used to determine whether an
area meets a given standard.
The Administrator also recognized that judgments about the
appropriate weight to place on any of the factors discussed above
should reflect consideration not only of the relative strength of the
evidence but also on the important uncertainties that remained in the
evidence and information being considered in this review. The
Administrator noted that the extent to which these uncertainties
influenced judgments about appropriate annual standard levels within
the range of 13 to 11 [mu]g/m\3\ would likely be greater for standard
levels in the lower part of this range which would necessarily be based
on fewer available studies than would higher levels within this range.
Based on the above considerations, the Administrator concluded that
it was appropriate to propose to set a level for the primary annual
PM2.5 standard within the range of 12 to 13 [mu]g/m\3\. The
Administrator provisionally concluded that a standard set within this
range would reflect alternative approaches to appropriately placing the
most weight on the strongest available evidence, while placing less
weight on much more limited evidence and on more uncertain analyses of
information available from a relatively small number of studies.
Further, she provisionally concluded that a standard level within this
range would reflect alternative approaches to appropriately providing
an adequate margin of safety for the populations at risk for the
serious health effects classified as having evidence of a causal or
likely causal relationship, depending in part on the emphasis placed on
margin of safety considerations. The Administrator recognized that
setting an annual standard level at the lower end of this range would
reflect an approach that placed more emphasis on the entire body of the
evidence, including the analysis of the distribution of air quality
concentrations most influential in generating health effect estimates
in the studies, and on margin of safety considerations, than would
setting a level at the upper end of the range. Conversely, an approach
that would support a level at the upper end of this range would
generally support a view that the uncertainties remaining in the
evidence are such that the evidence does not warrant setting a lower
annual standard level. Id. at 38942.
At the time of the proposal, while the Administrator recognized
that CASAC advised, and the Policy Assessment concluded, that the
available scientific information provided support for considering a
range that extended down to 11 [mu]g/m\3\, she concluded that proposing
such an extended range would reflect a public health policy approach
that placed more weight on relatively limited evidence and more
uncertain information and analyses than she considered appropriate at
this time. Nonetheless, the Administrator solicited comment on a level
down to 11 [mu]g/m\3\ as well as on approaches for translating
scientific evidence and rationales that would support such a level.
Such an approach might reflect a view that the uncertainties associated
with the available scientific information warrant a highly
precautionary public health policy response that would incorporate a
large margin of safety.
The Administrator recognized that potential air quality changes
associated with meeting an annual standard set at a level within the
range of 12 to 13 [mu]g/m\3\ will result in lowering risks associated
with both long- and short-term PM2.5 exposures. However, the
Administrator recognized that such an annual standard intended to serve
as the primary means for providing protection from effects associated
with both long- and short-term PM2.5 exposures would not by
itself be expected to offer requisite protection with an adequate
margin of safety against the effects of all short-term PM2.5
exposures. As a result, in conjunction with proposing an annual
standard level in the range of 12 to 13 [mu]g/m\3\, the Administrator
provisionally concluded that it was appropriate to continue to provide
supplemental protection by means of a 24-hour standard set at the
appropriate level, particularly for areas with high peak-to-mean ratios
possibly associated with strong local or seasonal sources, or for
PM2.5-related effects that may be associated with shorter-
than-daily exposure periods.
Based on the approach discussed in section III.A.3 above, at the
time of the proposal the Administrator relied upon evidence from the
short-term exposure studies as the principal basis for selecting the
level of the 24-hour standard. In considering these studies as a basis
for the level of a 24-hour standard, and having selected a 98th
percentile form for the standard, the Administrator agreed with the
focus in the Policy Assessment of looking at the 98th percentile
values, as well as at the long-term mean PM2.5
concentrations in these studies.
In considering the information provided by the short-term exposure
studies, the Administrator recognized that to the extent these studies
were conducted in areas that likely did not meet one or both of the
current standards, such studies did not help inform the
characterization of the potential public health improvements of
alternative standards set at lower levels. By reducing the
PM2.5 concentrations in such areas to just meet the current
standards, the Administrator anticipated that additional public health
protection would occur. Therefore, the Administrator focused on studies
that reported positive and statistically significant associations in
areas that would likely have met both the current 24-hour and annual
standards. She also considered whether or not these studies were
conducted in areas that would likely have met an annual standard level
of 12 to 13 [mu]g/m\3\ to inform her decision regarding an appropriate
24-hour standard level. As discussed in section III.E.4.a, consistent
with the Policy Assessment, the Administrator concluded that multi-
city, short-term exposure studies provided the strongest data set for
informing her decisions on appropriate 24-hour standard levels. The
Administrator viewed the single-city, short-term exposure studies as a
much more limited data set providing mixed results and, therefore, she
had less confidence in using those studies as a basis for setting the
level of a 24-hour standard. With regard to the limited number of
single-city studies that reported positive and statistically
significant associations for a range of health endpoints related to
short-term PM2.5 concentrations in areas that would likely
have met the current suite of PM2.5 standards, the
Administrator recognized that many of those studies had significant
limitations (e.g., limited statistical power, limited exposure data) or
equivocal results (mixed results within the same study area) that made
them unsuitable to form the basis for setting the level of a 24-hour
standard.

[[Page 3143]]

With regard to multi-city studies that evaluated effects associated
with short-term PM2.5 exposures, the Administrator observed
an overall pattern of positive and statistically significant
associations in studies with 98th percentile values averaged across
study areas in the range of 45.8 to 34.2 [mu]g/m\3\ (Burnett et al.,
2004; Zanobetti and Schwartz, 2009; Bell et al., 2008; Dominici et al.,
2006a, Burnett and Goldberg, 2003; Franklin et al., 2008). The
Administrator noted that, to the extent air quality distributions were
reduced to reflect just meeting the current 24-hour standard,
additional protection would be anticipated for the effects observed in
the three multi-city studies with 98th percentile values greater than
35 [mu]g/m\3\ (Burnett et al., 2004; Burnett and Goldberg, 2003;
Franklin et al., 2008). In the three additional studies with 98th
percentile values below 35 [mu]g/m\3\, specifically 98th percentile
concentrations of 34.2, 34.3, and 34.8 [mu]g/m\3\, the Administrator
noted that these studies reported long-term mean PM2.5
concentrations of 12.9, 13.2, and 13.4 [mu]g/m\3\, respectively (Bell
et al., 2008; Zanobetti and Schwartz, 2009; Dominici et al., 2006a).
In proposing to revise the level of the annual standard to within
the range of 12 to 13 [mu]g/m\3\, as discussed above, the Administrator
recognized that additional protection would be provided for the short-
term effects observed in these multi-city studies in conjunction with
an annual standard level of 12 [mu]g/m\3\, and in two of these three
studies in conjunction with an annual standard level of 13 [mu]g/m\3\.
She noted that the study-wide mean concentrations were based on
averaging across monitors within study areas and that compliance with
the standard would be based on concentrations measured at the monitor
reporting the highest concentration within each area. The Administrator
believed it would be reasonable to conclude that revision to the 24-
hour standard would not be appropriate in conjunction with an annual
standard within this range. Based on the above considerations related
to the epidemiological evidence, the Administrator provisionally
concluded that it was appropriate to retain the level of the 24-hour
standard at 35 [mu]g/m\3\, in conjunction with a revised annual
standard level in the proposed range of 12 to 13 [mu]g/m\3\.
In addition to considering the epidemiological evidence, the
Administrator also took into account air quality information based on
county-level 24-hour and annual design values to understand the public
health implications of retaining the 24-hour standard level at 35
[mu]g/m\3\ in conjunction with an annual standard level within the
proposed range of 12 to 13 [mu]g/m\3\. She considered whether these
suites of standards would meet a public health policy goal which
included setting the annual standard to be the ``generally
controlling'' standard in conjunction with setting the 24-hour standard
to provide supplemental protection to the extent that additional
protection is warranted. As discussed above, the Administrator
provisionally concluded that this approach was the most effective and
efficient way to reduce total population risk associated with both
long- and short-term PM2.5 exposures, resulting in more
uniform protection across the U.S. than the alternative of setting the
24-hour standard to be the controlling standard.
In considering the air quality information, the Administrator first
recognized that there was no annual standard within the proposed range
of levels, when combined with a 24-hour standard at the proposed level
of 35 [mu]g/m\3\, for which the annual standard would be the generally
controlling standard in all areas of the country. She further observed
that such a suite of PM2.5 standards with an annual standard
level of 12 [mu]g/m\3\ would result in the annual standard as the
generally controlling standard in most regions across the country,
except for certain areas in the Northwest, where the annual mean
PM2.5 concentrations have historically been low but where
relatively high 24-hour concentrations occur, often related to seasonal
wood smoke emissions (U.S. EPA, 2011a, pp. 2-89 to 2-91, Figure 2-10).
Although not explicitly delineated on Figure 2-10 in the Policy
Assessment, an annual standard of 13 [mu]g/m\3\ would be somewhat less
likely to be the generally controlling standard in some regions of the
U.S. outside the Northwest in conjunction with a 24-hour standard level
of 35 [mu]g/m\3\.
Taking the above considerations into account, the Administrator
proposed to revise the level of the primary annual PM2.5
standard from 15.0 [mu]g/m\3\ to within the range of 12.0 to 13.0
[mu]g/m\3\ and to retain the 24-hour standard level at 35 [mu]g/m\3\.
In the Administrator's judgment, such a suite of primary
PM2.5 standards and the rationale supporting such levels
could reasonably be judged to reflect alternative approaches to the
appropriate consideration of the strength of the available evidence and
other information and their associated uncertainties and the advice of
CASAC.
The Administrator recognized that the final suite of standards
selected from within the proposed range of annual standard levels, or
the broader range of annual standard levels on which public comment was
solicited, must be clearly responsive to the issues raised by the DC
Circuit's remand of the 2006 primary annual PM2.5 standard.
Furthermore, at the time of the proposal, she recognized that the final
suite of standards will reflect her ultimate judgment in the final
rulemaking as to the suite of primary PM2.5 standards that
would be requisite to protect the public health with an adequate margin
of safety from effects associated with fine particle exposures. The
final judgment to be made by the Administrator will appropriately
consider the requirement for a standard that is neither more nor less
stringent than necessary and will recognize that the CAA does not
require that primary standards be set at a zero-risk level, but rather
at a level that reduces risk sufficiently so as to protect public
health with an adequate margin of safety.
At the time of the proposal, having reached her provisional
judgment to propose revising the annual standard level from 15.0 to
within a range of 12.0 to 13.0 [mu]g/m\3\ and to propose retaining the
24-hour standard level at 35 [mu]g/m\3\, the Administrator solicited
public comment on this range of levels and on approaches to considering
the available evidence and information that would support the choice of
levels within this range. The Administrator also solicited public
comment on alternative annual standard levels down to 11 [mu]g/m\3\ and
on the combination of annual and 24-hour standards that commenters may
believe is appropriate, along with the approaches and rationales used
to support such levels. In addition, given the importance the evidence
from epidemiologic studies played in considering the appropriate annual
and 24-hour levels, the Administrator solicited public comment on
issues related to translating epidemiological evidence into standards,
including approaches for addressing the uncertainties and limitations
associated with this evidence.
c. Comments on Standard Levels
This section addresses comments that relate to consideration of the
appropriate levels of the primary annual and 24-hour PM2.5
standards, including comments on the general approach used by the EPA
to translate the available scientific information into standard levels
and how specific PM2.5 exposure studies should be considered
as a basis for the standard levels. These comments on standard levels
expand upon the more general comments that either supported or opposed
any change to the current suite of primary PM2.5

[[Page 3144]]

standards, which are addressed above in section III.D.2.\85\ As
explained there, one group of commenters generally opposed any change
to the current primary PM2.5 standards and more specifically
disagreed with the basis for the EPA's proposal to revise the annual
standard level. Another group of commenters supported revising the
current suite of primary PM2.5 standards to provide
increased public health protection. Some commenters in this second
group argued that both the annual and 24-hour standard levels should be
lowered while other commenters in this group agreed with the EPA's
proposal to retain the level of the 24-hour standard in conjunction
with revising the level of the annual standard. While generally
supporting the EPA's proposal to lower the level of the annual
standard, many commenters in this group disagreed that a level within
the EPA's proposed range was adequately protective and supported a
level of 11 [mu]g/m\3\ or below.
---------------------------------------------------------------------------

\85\ Specific comments on the forms of the annual and 24-hour
standards are addressed in section III.E.3.a and III.E.3.b,
respectively.
---------------------------------------------------------------------------

i. Annual Standard Level
The group of commenters opposed to any change to the current suite
of primary PM2.5 standards generally raised questions
regarding the underlying scientific evidence, including the causal
determinations reached in the Integrated Science Assessment, and
focused strongly on the uncertainties they saw in the scientific
evidence as a basis for their conclusion that no changes to the current
standard levels were warranted. In commenting on the proposed standard
levels, these commenters typically relied on the arguments summarized
and addressed above in section III.D.2 as to why they believed it was
inappropriate for the EPA to make any revisions to the suite of primary
PM2.5 standards. That is, they asserted that the EPA's
causal determinations were not adequately supported by the underlying
scientific information; the biological plausibility of health effects
observed in epidemiological studies has not been demonstrated in
controlled human exposure and toxicological studies; uncertainties in
the underlying health science are as great or greater than in 2006;
there is no evidence of greater risk since the last review to justify
tightening the current annual PM2.5 standard; and ``new''
studies not included in the Integrated Science Assessment continue to
increase uncertainty about possible health risks associated with
exposure to PM2.5.
With regard to the level of the annual standard, these commenters
strongly disagreed with the Agency's proposed decision to revise the
level to within a range of 12 to 13 [mu]g/m\3\ and argued that the
current standard level of 15 [mu]g/m\3\ should be retained. For
example, UARG, API, and other commenters in this group raised a number
of issues that they asserted called into question the EPA's
interpretation of the epidemiological evidence to support revising the
annual standard level. These commenters raised specific questions
related to the general approach used by the EPA to translate the air
quality and other information from specific epidemiological studies
into standard levels, including: (1) The EPA's approach for using
composite monitor air quality distributions reported in epidemiological
studies to select a standard level that would be compared to
measurements at the monitor recording the highest value in an area to
determine compliance with the standard; (2) the appropriate exposure
period for effects observed in long-term exposure mortality studies;
and (3) the use of the EPA's analysis of distributions of underlying
population-level data (i.e., health event and study population data)
for those epidemiological studies for which such information was
available. These commenters also raised questions regarding the EPA's
consideration of specific scientific evidence as a basis for setting a
standard level, including: (4) evidence of respiratory morbidity
effects in long-term exposure studies and (5) more limited evidence of
health effects which have been categorized in the Integrated Science
Assessment as suggestive of a causal relationship (i.e., developmental
and reproductive outcomes). These comments are discussed in turn below.
(1) Some commenters in this group argued that one reason why they
believe there is no basis for setting a standard level below 15 [mu]g/
m\3\ is that the air quality metric from epidemiological studies that
the EPA relied on in the proposal is not the same metric that will be
compared to the level of the standard to determine compliance with the
standard. That is, commenters noted that the long-term mean
PM2.5 concentrations that the EPA considered, shown in
Figure 4 above, are composite monitor mean concentrations (i.e.,
concentrations averaged across multiple monitors within areas with more
than one monitor), whereas the PM2.5 concentrations that
will be compared to the level of the standard are maximum monitor
concentrations (i.e., the concentration measured by the monitor within
an area reporting the highest concentration). This comment was
presented most specifically in UARG's comments (UARG, 2012, Attachment
1, pp. 2 to 6), which raised two overarching issues as discussed below.
First, the commenter noted that the EPA's approach of considering
composite monitor mean PM2.5 concentrations in selecting a
standard level, and then comparing the maximum monitor mean
PM2.5 concentration in each area to the standard level when
the standard is implemented, was characterized in the proposal as
inherently having the potential to build in a margin of safety (UARG,
2012, Attachment 1, p. 4, citing 77 FR 38905). The commenter asserted
that the Administrator is ignoring this distinction between composite
and maximum monitor concentrations, and that this approach creates an
unwarranted case for lowering the standard level, since in the
commenter's view, it would result in a margin of safety that would be
arbitrary, not based on evidence, and unquantified (UARG, 2012,
Attachment 1, p. 4). In support of this view, the commenter asserted
that there is a significant difference between composite monitor mean
PM2.5 concentrations and maximum monitor mean
PM2.5 concentrations. The commenter asserted that the
maximum monitor value will always be higher than the composite monitor
value (except in areas that contain only a single monitor), such that
when an area just attains the NAAQS, that area's composite monitor
long-term mean PM2.5 concentration will be lower than the
level of the standard (UARG, 2012, Attachment 1, p. 3).
Second, the commenter asserted that a more ``reasoned and
consistent approach would be to decide on a mean composite monitor
PM2.5 level that should be achieved and then identify the
maximum monitor level that would result in that composite value''
(UARG, 2012, Attachment 1, p. 4). The commenter conducted an analysis
of maximum monitor versus composite monitor annual mean
PM2.5 concentrations using monitoring data \86\ from 2006 to
2008 and presented results averaged across areas within two groups
(i.e., those with design values \87\ above the current standard level
and those with design values just below the

[[Page 3145]]

current standard level) to illustrate their suggested alternative
approach. The commenter interpreted this analysis as showing that the
composite monitor long-term mean PM2.5 concentrations from
the subset of the epidemiological studies shown in Figure 4 (of the
proposal and above) that the commenter considered to be an appropriate
focus for this analysis would be achieved across the U.S. if the
current annual NAAQS of 15 [mu]g/m\3\ is retained and attained. The
commenter considered the subset of epidemiological studies that
included only long-term exposures studies of effects for which the
evidence is categorized as causal or likely causal, but did not
consider short-term exposure studies. On this basis, the commenter
asserted that attaining the current annual PM2.5 standard
would result in composite monitor long-term mean concentrations in all
areas that would be generally within or below the range of the
composite monitor long-term mean concentrations from such studies and,
as a result, there is no reason to lower the level of the current
annual NAAQS.
---------------------------------------------------------------------------

\86\ The commenter indicated that this analysis was based on
monitoring data for every core based statistical area (CBSA) in the
EPA's Air Quality System (AQS) database.
\87\ The design value is the air quality statistic that is
compared to the level of the NAAQS to determine the attainment
status of a given area.
---------------------------------------------------------------------------

In considering the first issue related to the EPA's approach, the
EPA notes that in proposing to revise both the form and level of the
annual standard, the Administrator clearly took into account the
distinction between the composite monitor long-term mean
PM2.5 concentrations from the epidemiological studies,
considered as a basis for selecting an annual standard level, and
maximum monitor long-term mean PM2.5 concentrations. In
deciding to focus on the composite monitor long-term mean
concentrations in selecting the standard level, and on the maximum
monitor concentrations in selecting the form of the standard (i.e.,
consistent with proposing to eliminate the option for spatial averaging
across monitors within an area when implementing the standard \88\),
the Administrator reasonably considered the distinction between these
metrics in a manner that was consistent with advice from CASAC (Samet
et al., 2010d, pp. 2 to 3).
---------------------------------------------------------------------------

\88\ As discussed above in section III.E.3.a.
---------------------------------------------------------------------------

As noted above in section III.A.3, the EPA recognizes that a
statistical metric (e.g., the mean of a distribution) based on maximum
monitor concentrations may be identical to or above the same
statistical metric based on composite monitor concentrations. More
specifically, many areas have only one monitor, in which case the
composite and maximum monitor concentrations are identical. Based on
the most recent data from the EPA's AQS from 2009 to 2011 in the 331
CBSAs in which valid PM2.5 data are available, as discussed
in Frank (2012a, Table 5), there were 208 such areas (with design
values ranging up to about 15 [mu]g/m\3\). Frank (2012a) also observed
that other areas have multiple monitors with composite and maximum
monitor mean PM2.5 concentrations that were the same or
relatively close, with 57 areas in which the maximum monitor mean
concentration was no more than 0.5 [mu]g/m\3\ higher than the composite
monitor mean concentration and 56 areas in which the difference was
between 0.6 and 2 [mu]g/m\3\. Further, there were only a few other
areas in which the maximum monitor mean concentration was appreciably
higher than the composite monitor mean concentration, such as areas in
which some monitors may be separately impacted by local sources. There
were only 10 such areas in the country in which the maximum monitor
mean concentration was between 2 to 6 [mu]g/m\3\ higher than the
composite monitor concentration (Frank, 2012a, Table 4).\89\ Thus, the
EPA does not agree that there is a significant difference between
composite monitor mean PM2.5 concentrations and maximum
monitor mean PM2.5 concentrations in the large majority of
areas across the country.
---------------------------------------------------------------------------

\89\ The average difference between the maximum and composite
design value among the 123 CBSAs with two or more monitors is 0.8
[mu]g/m\3\ and the median difference is 0.6 [mu]g/m\3\. The 25th and
75th percentiles are 0.3 and 1.0 [mu]g/m\3\, respectively (Frank,
2012a, p. 4).
---------------------------------------------------------------------------

In proposing to revise the form of the annual PM2.5
standard, as discussed above in section III.E.3.a, the EPA noted that
when an annual PM2.5 standard was first set in 1997, the
form of the standard included the option for averaging across
measurements at appropriate monitoring sites within an area, generally
consistent with the composite monitor approach used in epidemiological
studies, with some constraints intended to ensure that spatial
averaging would not result in inequities in the level of protection for
communities within large metropolitan areas. In the last review the EPA
tightened the constraints on spatial averaging, and in this review has
eliminated the option altogether, on the basis of analyses in each
review that showed that such constraints may be inadequate to avoid
substantially greater exposures for people living in locations around
the monitors recording the highest PM2.5 concentrations in
some areas, potentially resulting in disproportionate impacts on at-
risk populations of persons with lower SES levels as well as
minorities. In light of these analyses, and consistent with the
Administrator's decision to revise the form of the annual
PM2.5 standard by eliminating the option for spatial
averaging, the EPA continues to conclude that a standard level based on
consideration of long-term mean concentrations from composite monitors,
and applied at each monitor within an area including the monitor
measuring the highest concentration, is the appropriate approach to use
in setting a standard that will protect public health, including the
health of at-risk populations, with an adequate margin of safety, as
required by the CAA.
The EPA acknowledges that at proposal, the Agency characterized the
approach of using maximum monitor concentrations to determine
compliance with the standard, while selecting the standard level based
on consideration of composite monitor concentrations, as one that
inherently had the potential to build in a margin of safety (77 FR
38905), and CASAC reiterated that view in supporting the EPA's approach
(Samet, 2010d, p. 3). Nonetheless, in light of the discussion above,
the EPA more specifically recognizes that this approach does not build
in any margin of safety in the large number of areas across the country
with only one monitor. Further, based on the analyses done to inform
consideration of the form of the standard (Schmidt, 2011, Analysis A),
the EPA concludes that this approach does not provide a margin of
safety for the at-risk populations that live around the monitor
measuring the highest concentration, such as in those few areas in
which the maximum monitor concentration is appreciably higher than the
composite monitor concentration. Rather, this approach properly treats
those at-risk populations the same way it does the broader populations
that live in areas with only one monitor, by providing the same degree
of protection for those at-risk populations that would otherwise be
disproportionately impacted as it does for the broader populations in
other areas, While the EPA recognizes that this approach can result in
some additional margin of safety for the subset of areas with multiple
monitors in which at-risk populations may not be disproportionately
represented in areas around the maximum monitor, which may be the case
in areas with relatively small differences between the maximum and
composite monitor concentrations, the EPA notes that this margin would
be relatively small in such areas.
Based on the above considerations, the EPA does not agree that the
Agency's approach of using maximum monitor concentrations to determine
compliance with the standard, while

[[Page 3146]]

selecting the standard level based on consideration of composite
monitor concentrations creates an unwarranted case for lowering the
standard level based on a margin of safety that would be arbitrary, not
based on evidence, or lack quantification. The EPA recognizes that
setting a standard to protect public health, including the health of
at-risk populations, with an adequate margin of safety, depends upon
selecting a standard level sufficiently below where the EPA has found
the strongest evidence of health effects so as to provide such
protection, and that the EPA's approach regarding consideration of
composite and maximum monitor concentrations is intended to, and does,
serve to address this requirement as part of and not separate from the
selection of an appropriate standard level based on the health effects
evidence.
In considering the second issue related to the commenter's
suggested alternative approach, the EPA strongly disagrees with the
commenter's view that a more ``reasoned and consistent approach would
be to decide on a mean composite monitor PM2.5 level that
should be achieved and then identify the maximum monitor level that
would result in that composite value'' (UARG, 2012, Attachment 1, p.
4). As discussed above, the EPA notes that for areas with only one
monitor, or with multiple monitors that measure concentrations that are
very close in magnitude, the maximum monitor level that would limit the
composite monitor PM2.5 level to be no greater than the
level that should be achieved to protect public health with an adequate
margin of safety, would essentially be the same as that composite
monitor level. Further, as discussed above, even for areas in which the
maximum monitor concentration is appreciably higher than other monitor
concentrations within the same area, public health would not be
protected with an adequate margin of safety if the disproportionately
higher exposures of at-risk, susceptible populations around the monitor
measuring the highest concentration were in essence averaged away with
measurements from monitors in other locations within large urban areas.
Further, the commenter's suggested approach would be based on annual
average PM2.5 concentrations that have been measured over
some past time period. Such an approach would reflect the air quality
that existed in the past, but it would not necessarily provide
appropriate constraints on the range of concentrations that would be
allowed by such a standard in the future, when relationships between
maximum and composite monitor concentrations in areas across the
country may be different. For these reasons, the EPA fundamentally
rejects the commenter's suggested approach because in the EPA's view it
would not protect public health, including providing protection for at-
risk populations, with an adequate margin of safety in areas across the
country.
More specifically, in further considering the commenter's analysis
of design values based on maximum versus composite monitor annual mean
PM2.5 concentrations using monitoring data from 2006 to 2008
which they assert supports retaining the current standard level of 15
[mu]g/m\3\, the EPA finds flaws with the numerical results and the
scope of the analysis, as well as flaws in the commenter's translation
of the analysis results into the basis for selecting an annual standard
level.
In considering the commenter's analysis, the EPA notes that the
analysis compared maximum versus composite monitor annual mean
PM2.5 concentrations, averaged over 3 years, for two groups
of areas: (1) Areas with design values that exceed the current annual
standard level (i.e., greater than 15.0 [mu]g/m\3\) and (2) areas with
design values that are just attaining the current annual standard
(i.e., between 14.5 and 15.0 [mu]g/m\3\).\90\ The commenter indicated
that they used the full body of PM2.5 monitoring data from
the EPA's AQS database (UARG, 2012, Attachment 1, p. 4), In attempting
to reproduce the commenter's results, the EPA repeated the calculations
using only valid air quality data (i.e., data that meet data
completeness and monitor siting criteria) from the AQS database for the
same time period (Frank, 2012a).\91\ Based on this corrected analysis,
the EPA finds that the composite monitor concentrations averaged across
the areas within each group are somewhat higher than those calculated
by the commenter, and the average differences between the maximum and
composite monitor concentrations are somewhat smaller (Frank, 2012a,
Table 3).\92\ Notably, the difference between the maximum and composite
monitor average concentrations for the second group of areas is
substantially reduced in the corrected analysis, such that the
difference (averaged across the 10 areas with valid data in the second
group) is approximately 0.5 [mu]g/m\3\, not 1.2 [mu]g/m\3\ as in the
commenter's analysis. In addition, the commenter's analysis compared
the average of the composite monitors to the average of the maximum
monitors for each subset of areas. This comparison of averages across
all the areas in each subset masks the fact that the large majority of
areas across the country have only one monitor, with the composite
monitor and maximum monitor values the same for such areas, and many
other areas have a maximum monitor value that is close to the composite
monitor value. As discussed above, these circumstances have a major
impact on the protection that would be achieved by the approach
suggested by the commenter.
---------------------------------------------------------------------------

\90\ For the first group of areas (which included 33 areas),
this analysis calculated an average across the areas of maximum
monitor annual mean PM2.5 concentrations, averaged over 3
years, of 17.2 [mu]g/m\3\ compared to an average of composite
monitor concentrations of 14.3 [mu]g/m\3\. For the second group of
areas (which included 11 areas), this analysis calculated an average
across the areas of maximum monitor annual mean concentrations,
averaged over 3 years, of 14.8 [mu]g/m\3\ compared to an average of
composite monitor concentrations of 13.6 [mu]g/m\3\ (UARG, 2012,
Attachment 1, Table 1).
\91\ The EPA notes that the Frank (2012a) analysis is similar to
an earlier EPA staff analysis (Hassett-Sipple et al., 2010), which
used air quality data from EPA's AQS database to compare maximum
versus composite monitor long-term mean PM2.5
concentrations across the study areas in six selected multi-city
epidemiological studies.
\92\ The EPA's analysis was intended to repeat the commenter's
analysis, but using only valid air quality data (from 2006 to 2008).
For the first group of areas (which included 21 areas with valid
data), the EPA's analysis calculated an average across the areas of
maximum monitor annual mean concentrations, averaged of 3 years, of
16.8 [mu]g/m\3\ compared to an average of composite monitor
concentrations of 14.8 [mu]g/m\3\. For the second group of areas
(which included 10 areas with valid data), the EPA's analysis
calculated an average across the areas of maximum monitor annual
mean concentrations, averaged over 3 years, of 14.8 [mu]g/m\3\
compared to an average of composite monitor concentrations of 14.2
[mu]g/m\3\ (Frank, 2012a, Table 3).
---------------------------------------------------------------------------

With regard to the scope of the commenter's analysis, the EPA finds
that by limiting the scope to a small subset of areas with design
values above or just below the current annual standard level of 15
[mu]g/m\3\, the analysis ignores the large number of areas across the
country with lower design values that are relevant to consider in light
of the epidemiological evidence of serious health effects at lower
concentrations, well below the level of the current standard.
In translating the analysis results into the basis for selecting an
annual standard level, the commenter's translation is premised on the
view that the ``natural focal point'' for setting an annual
PM2.5 standard level should be somewhere within the range of
the long-term mean PM2.5 concentrations from the subset of
epidemiological studies that included only long-term exposure studies
of effects for which the evidence is categorized as causal or likely
causal, but not for effects categorized as suggestive of causality, nor
did it

[[Page 3147]]

include short-term exposure studies (which are included in Figure 4 of
the proposal notice and above). Such a view is not consistent with
setting a standard that would provide sufficient protection from the
serious health effects reported even in the limited subset of studies
considered by the commenter, including protecting public health with an
adequate margin of safety. As discussed below, the EPA does not agree
with the commenter's view as to the appropriate focal point for
selecting the level of an annual PM2.5 standard, or with the
limited set of studies considered by the commenter as a basis for
selecting the level of the annual PM2.5 standard.
Regarding an appropriate focal point for selecting the level of the
annual standard, as discussed in the proposal and as advised by CASAC,
the EPA has focused on PM2.5 concentrations somewhat below
the lowest long-term mean concentrations from each of the key studies
of both long- and short-term exposures of effects for which the
evidence is causal or likely causal, as considered by the EPA (i.e.,
the first two sets of studies shown in Figure 4). If the level of the
annual standard was set just somewhere within the range of the long-
term mean concentrations from the various long-term exposure studies,
then one or more of the studies would have a long-term mean
concentration below the selected level of the standard. Absent some
reason to ignore or discount these studies, which the commenter does
not provide (and of which the EPA is unaware), setting such a standard
would allow that level of air quality, where the evidence of health
effects is strongest, and its associated risk of PM2.5-
related mortality and/or morbidity effects to continue. Selecting such
a standard level could not be considered sufficient to protect the
public health with an adequate margin of safety.
Further, focusing on just the long-term mean PM2.5
concentrations in the key epidemiological studies--even the lowest
long-term mean concentration from the set of key studies--is not
appropriate. Concentrations at and around the long-term mean
concentrations represent the part of the air quality distribution where
the data in any given study are most concentrated and, thus, where the
confidence in the magnitude and significance of an association in such
study is strongest. However, the evidence of an association with
adverse health effects in the studies is not limited to the
PM2.5 concentrations just at and around the long-term mean,
but rather extends more broadly to a lower part of the distribution,
recognizing that no discernible population-level threshold for any such
effects can be identified based on the available evidence. This broader
region of the distribution of PM2.5 concentrations should be
considered to the extent relevant information is available, recognizing
that the degree of confidence in the association identified in a study
would become lower as one moves below concentrations at and around the
long-term mean concentration in any given study. The commenter's
approach ignores this fundamental consideration.
Regarding the set of studies that is appropriate to inform the
selection of the level of the annual PM2.5 standard, the EPA
finds that limiting consideration only to the long-term exposure
studies, as this commenter suggests, would be tantamount to ignoring
the short-term exposure studies,\93\ which provide some of the
strongest evidence from the entire body of epidemiological studies.
Thus, selecting an annual standard level using the limited set of
studies suggested by the commenter would fail to provide a degree of
protection that would be sufficient to protect public health with an
adequate margin of safety.
---------------------------------------------------------------------------

\93\ The commenter suggests that the EPA should not place
significant reliance on the long-term mean concentrations from
short-term exposure studies because ``[T]he short[hyphen]term
studies did not use the annual average of PM2.5 to
develop their associations; they used the daily 24-hour averages of
PM2.5. Thus, short-term studies do not provide a natural
indicator for the appropriate level of an annual standard * * *.''
(UARG, 2012, Attachment 1, p. 3). The EPA finds this argument
unpersuasive. Quite simply, effects were observed in these studies
with an air quality distribution that can meaningfully be
characterized by these long-term mean concentrations. Indeed, in
remanding the 2006 standard, the D.C. Circuit discussed at length
the interrelationship of the long- and short-term standards and
studies, and remanded the 2006 standard to the EPA, in part, for
ignoring those relationships without adequate explanation. American
Farm Bureau Federation v. EPA. 559 F. 3d at 522-24.
---------------------------------------------------------------------------

For all the reasons discussed above, the EPA finds the commenter's
concerns with the EPA's approach to considering composite and maximum
monitor PM2.5 concentrations in selecting the level of the
annual PM2.5 standard to be without merit. Further, the EPA
finds no support in the commenter's analysis for their suggested
alternative approach.
(2) With respect to the appropriate exposure period for mortality
effects observed in long-term exposure studies, some commenters in this
group generally expressed views consistent with comments from UARG that
argued that these studies ``are most likely detecting health risk from
earlier, higher PM2.5 levels and misattributing those risks
to more recent, lower PM2.5 levels'' (UARG, 2012, Attachment
1, p 7). Further, this commenter asserted that ``there is no knowledge
or evidence indicating whether premature deaths are the result of
PM2.5 exposures in the most recent year; or due to physical
damages incurred from PM2.5 exposures much earlier in life
(with the impact on lifespan only emerging later in life); or due to
total accumulated PM2.5 exposure over many years.'' Id. In
addition, the commenter asserted that the long-term exposure studies of
mortality are central to the EPA's basis for proposing to set a lower
annual standard level, since most of the estimated benefits associated
with a lower annual PM2.5 standard are based on reductions
in mortality related to long-term exposures to PM2.5.
As an initial matter, the EPA has recognized the challenge in
distinguishing between PM2.5-associated effects due to past
and recent long-term exposures, and in identifying the relevant latency
period for long-term exposure to PM and resultant health effects (U.S.
EPA, 2009a, section 7.6.4; 77 FR 38941/1). While the EPA has
acknowledged that there remain important uncertainties related to
characterizing the most relevant exposure periods in long-term exposure
studies, the assertion that there is ``no knowledge or evidence'' that
helps to inform this issue is not correct, as discussed below.
Both in the last review and in the current review, the EPA has
assessed studies that used different air quality periods for estimating
long-term exposure and tested associations with mortality for the
different exposure periods (U.S. EPA, 2004, section 8.2.3.5; U.S. EPA
2009a, section 7.6.4). In this review, the Integrated Science
Assessment discussed studies available since the last review that have
assessed the relationship between long-term exposure to
PM2.5 and mortality to explore the issue of the latency
period between exposure to PM2.5 and death (U.S. EPA, 2009a,
section 7.6.4).
Notably, in a recent analysis of the extended Harvard Six Cities
Study, Schwartz et al. (2008) used model averaging (i.e., multiple
models were averaged and weighted by probability of accuracy) to assess
exposure periods prospectively (77 FR 38907/1-2). The exposure periods
were estimated across a range of unconstrained distributed lag models
(i.e., same year, one year prior, two years prior to death). In
comparing lags, the authors reported that the effects of changes in
exposure to PM2.5 on mortality were strongest within a 2-
year period prior to death (U.S. EPA, 2009a, p. 7-92, Figure 7-9).
Similarly, a large

[[Page 3148]]

multi-city study of the elderly found that the mortality risk
associated with long-term exposure to PM10 reported
cumulative effects that extended over the years that deaths were
observed in the study population (i.e., the follow-up period) and for
the 3-year period prior to death (Zanobetti et al., 2008).
Further, in a study of two locations that experienced an abrupt
decline in PM2.5 concentrations (i.e., Utah Steel Strike,
coal ban in Ireland), R[ouml][ouml]sli et al. (2005) reported that
approximately 75 percent of health benefits were observed in the first
5 years (U.S. EPA, 2009a, Table 7-9). Schwartz et al. (2008) and Puett
et al. (2008) found, in a comparison of exposure periods ranging from 1
month to 48 months prior to death, that exposure to PM10 24
months prior to death exhibited the strongest association, and the
weakest association was reported for exposure in the time period of 1
month prior to death.
Overall, the EPA notes that the available evidence for determining
the exposure period that is causally related to the mortality effects
of long-term PM2.5 exposures, as discussed above, cannot
specifically disentangle the effects observed in long-term exposure
studies associated with more recent air quality measurements from
effects that may have been associated with earlier, and most likely
higher, PM2.5 exposures. While the evidence suggests that a
latency period of up to five years would account for the majority of
deaths, it does not provide a basis for concluding that it is solely
recent PM2.5 concentrations that account for the mortality
risk observed in such studies. Nonetheless, the more recent air quality
data does well at explaining the relationships observed between long-
term exposures to PM2.5 and mortality, with the strongest
association observed in the two years prior to death. Further, the EPA
recognizes that there is no discernible population-level threshold
below which effects would not occur, such that it is reasonable to
consider that health effects may occur over the full range of
concentrations observed in the epidemiological studies, including the
lower concentrations in the latter years.
In light of this evidence and these considerations, the EPA
concludes that it is appropriate to consider air quality concentrations
that are generally contemporaneous with the collection of health event
data (i.e., collected over the same time period) as being causally
associated with at least some proportion of the deaths assessed in a
long-term exposure study. This would include long-term mean
PM2.5 concentrations from most of the key long-term exposure
studies of effects with causal or likely causal evidence shown in
Figure 4 above, which reported long-term mean PM2.5
concentrations ranging from 13.6 [micro]g/m\3\ to 14.3 [micro]g/m\3\.
These studies include studies of mortality by Eftim et al. (2008),
which separately analyzed the ACS and Harvard Six City sites, Zeger et
al. (2008), and Lipfert et al. (2006a), as well as studies of morbidity
endpoints by Goss et al. (2004), McConnell et al. (2003) and Gauderman
et al. (2004), and Dockery et al. (1996) and Razienne et al. (1996).
The EPA acknowledges that uncertainty in the relevant exposure period
is most notable in two other long-term exposure studies of mortality.
The Miller et al. (2007) reported a long-term mean PM2.5
concentration for a 1-year exposure period that post-dated the follow-
up period in which health event data were collected by two years. Also,
the Krewski et al. (2009) study reported a long-term mean
PM2.5 concentration for an exposure period that included
only the last two years of the 18-year follow-up period. Based on these
considerations, the EPA does not now consider it appropriate to put
weight on the reported long-term mean concentrations from these two
studies for the purpose of translating the information from the long-
term mortality studies into a basis for selecting the level of the
annual PM2.5 standard.\94\
---------------------------------------------------------------------------

\94\ Nonetheless, the EPA notes that the Krewski et al. (2009)
and Miller et al. (2007) studies provide strong evidence of
mortality and cardiovascular-related effects associated with long-
term PM2.5 exposures to inform causality determinations
reached in the Integrated Science Assessment (U.S. EPA, 2009a,
sections 7.2.11 and 7.6).
---------------------------------------------------------------------------

In addition, the EPA acknowledges that exposure periods that extend
at least a couple years prior to the follow-up period in which health
event data were collected would likely more fully capture the PM-
related deaths in such studies. To explore how much higher the long-
term mean PM2.5 concentrations would likely have been had
air quality data prior to the follow-up years of the studies been
included, the EPA conducted a sensitivity analysis of long-term mean
PM2.5 concentrations (Schmidt, 2012a) particularly
considering studies that only included deaths from a relatively recent
follow-up period. As examples of such studies, this analysis considered
the Eftim et al. (2008) study of mortality in the ACS sites and the
Harvard Six Cities sites, as well as sites in the eastern region in the
Zeger et al. (2008) study. Using data from the EPA's AQS database, the
analysis added the two years of air quality data just prior to the
follow-up period in each study, which was 2000 to 2002 in Eftim et al.
(2008) and 2000 to 2005 in Zeger et al. (2008). The analysis then
calculated the extended long-term mean PM2.5 concentration
for each study. As discussed in Schmidt (2012a), in each case the long-
term mean PM2.5 concentration averaged over the extended
exposure period was less than 0.4 [micro]g/m\3\ higher than the long-
term mean PM2.5 concentration averaged over the follow-up
period. The EPA finds it reasonable to conclude that such a relatively
small difference in long-term mean PM2.5 concentrations
would likely apply for other long-term exposure studies that used
similarly recent follow-up periods as well (e.g., Goss et al., 2004;
Lipfert et al., 2006a).
Based on the above considerations, the EPA concludes that it is
appropriate to consider the available air quality information from the
long-term exposure studies, while taking into account the uncertainties
in the relevant long-term exposure periods in weighing the information
from these studies. The EPA recognizes that considering such
information in selecting an appropriate annual standard level has the
potential to build in some margin of safety. The EPA further concludes
that it is appropriate to consider the air quality information from the
set of long-term exposure studies discussed above in the context of the
broader array of epidemiological studies that inform the EPA's
consideration of the level of the annual PM2.5 standard.
The EPA also notes that while the long-term exposure studies are an
important component of the epidemiological evidence that informs the
Agency's consideration of the level of the annual standard, they do not
provide the only relevant information, nor are they the set of studies
for which the relevant long-term mean PM2.5 concentrations
are the lowest. As discussed in the proposal, the EPA also considers
the long-term mean PM2.5 concentrations from the short-term
mortality and morbidity studies as providing important information in
considering the level of the annual standard. As discussed above, a
large proportion of the aggregate risk associated with short-term
exposures results from the large number of days during which the 24-
hour average concentrations are in the low- to mid-range of the
concentrations observed in the studies. Thus, setting the level of the
annual standard based on long-term mean concentrations, as well as the
distribution of concentrations below the mean, in the short-term
exposure studies is the most effective and efficient way to reduce
total PM2.5-

[[Page 3149]]

related risk from the broad array of mortality and morbidity effects
associated with short-term exposures.
Further, the EPA notes that the relevant exposure period for the
short-term exposure studies is the period contemporaneous with the
collection of health event data, and that this exposure period is not
subject to the uncertainties discussed above related to the long-term
exposure studies. Recognizing that the long-term mean PM2.5
concentrations from several of the multi-city short-term exposure
studies shown in Figure 4 are below the long-term mean PM2.5
concentrations from the long-term exposure studies (with the exception
of Miller et al., 2007).\95\ It is reasonable that in selecting the
level of the annual standard primary consideration should be given to
the information from this set of short-term exposure studies. There is
no reasonable basis to discount the long-term mean concentrations of
the short-term exposure studies for purposes of setting the level of
the annual standard. Thus, the commenter is incorrect in asserting that
the long-term exposure studies, not the short-term exposure studies,
would be central in the Administrator's decision on the level of the
annual standard. The standard is ultimately intended to protect not
just against the single type of effect that contributes the most to
quantitative estimates of risk to public health, but rather to the
broad array of effects, including mortality and morbidity effects from
long- and short-term exposures across the range of at-risk populations
impacted by PM2.5-related effects.
---------------------------------------------------------------------------

\95\ As noted above, the EPA is not placing weight on the
reported long-term mean concentrations from the Miller et al. (2007)
study for the purpose of translating the information from the long-
term mortality studies into a basis for selecting the level of the
annual PM2.5 standard.
---------------------------------------------------------------------------

(3) With regard to the EPA's analysis of distributions of
underlying population-level data (i.e., health event and study
population data) and corresponding air quality data from each study
area in certain key multi-city epidemiological studies (Rajan et al.,
2011), some commenters in this group raised a number of issues related
to this analysis (API, 2012, Attachment 1 pp. 5 to 6; McClellan, 2012,
pp.2 to 4). Some commenters noted the limited number of studies for
which health event and study population data were available, and
questioned whether these distributions would apply to other studies.
Commenters expressed concerns that this analysis had not been formally
reviewed by CASAC and was not published in the peer-review literature.
Based on such concerns, some commenters asserted that the EPA should
not consider this information as a basis for selecting a standard
level.
As an initial matter, as discussed in section III.E.4.b above, the
EPA agrees with CASAC's advice that it is appropriate to consider
additional data beyond the mean PM2.5 concentrations in key
multi-city studies to help inform selection of the level of the annual
PM2.5 standard. As both the EPA and CASAC recognize, in the
absence of a discernible threshold, health effects may occur over the
full range of concentrations observed in the epidemiological studies.
Nonetheless, the EPA recognizes that confidence in the magnitude and
significance of an association is highest at and around the long-term
mean PM2.5 concentrations reported in the studies and the
degree of confidence becomes lower at lower concentrations within any
given study. Following CASAC's advice (Samet, 2010d, p.2), the EPA used
additional population-level and air quality data made available by
study authors to conduct an analysis of the distributions of such data,
to help inform consideration of how the degree of confidence in the
magnitude and significance of observed associations varies across the
range of long-term mean PM2.5 concentrations in study areas
within key multi-city epidemiological studies. In the EPA's view, such
consideration is important in selecting a level for an annual standard
that will protect public health with an adequate margin of safety.
With regard to the number of multi-city studies for which an
analysis of the distributions of population-level data across the study
areas and the corresponding annual mean PM2.5 concentrations
was done, the EPA noted at proposal that data for such an analysis were
made available from study authors for four studies, including two long-
term exposure studies and two short-term exposure studies.\96\ The EPA
recognized that access to health event data can be restricted due to
confidentiality issues, such that it is not reasonable to expect that
such information could be made available from all studies. In
considering the information from these four studies, the EPA has
further taken into consideration uncertainties discussed in response to
the above comment related to the appropriate exposure period for long-
term exposure studies. Based on these considerations, as noted above,
the EPA concludes that such uncertainties are an important factor in
evaluating the usefulness of the air quality information from the two
long-term exposure studies in this analysis (Krewski et al., 2009;
Miller et al., 2007) and that it would not be appropriate to place
weight on the distributional analysis of health event and air quality
data from these two studies specifically for the purpose of translating
the information from the long-term mortality studies into a basis for
selecting the level of the annual PM2.5 standard. Such
uncertainties are not relevant to the short-term exposure studies, and
thus, the Agency focuses on the two short-term exposure studies in this
analysis (Bell et al., 2008; Zanobetti and Schwartz, (2009).
---------------------------------------------------------------------------

\96\ Health event data and study population data were available
from two short-term exposure studies (Bell et al. 2008; Zanobetti
and Schwartz, 2009) and one long-term exposure study (Krewski et
al., 2009). Only study population data were available from another
long-term exposure study (Miller et al., 2007).
---------------------------------------------------------------------------

In focusing on these two short-term exposure studies, the EPA first
notes that these studies are key multi-city studies that reported
positive and statistically significant associations between mortality
and cardiovascular-related hospital admissions across a large number of
areas throughout the U.S. (112 U.S. cities in Zanobetti and Schwartz,
2009; 202 U.S. counties in Bell et al., 2008) using relatively recent
air quality and health event data (i.e., 1999 through 2005 in both
studies). The EPA considers this to be a modest but important data set
to use for this distributional analysis to help inform consideration of
how much below the long-term mean PM2.5 concentrations in
key multi-city long- and short-term exposure studies the annual
PM2.5 standard level should be set. While the EPA
acknowledges that having such data available from more studies would
have been useful, the Agency finds the information from this limited
set of studies to be an important consideration in selecting an annual
standard level, consistent with CASAC advice to consider such
information.
In considering the results of this distributional analysis, as
discussed more fully in the Response to Comments document, the EPA
considers PM2.5 concentrations between the 25th and 10th
percentiles of the distribution of health events to be a reasonable
range for providing a general frame of reference for that part of the
distribution in which confidence in the magnitude and significance of
the association may be appreciably lower than confidence at and around
the long-term mean concentration. For the two short-term exposure
studies included in this analysis, the EPA notes that the
PM2.5 concentrations corresponding to the 25th percentiles
of the distributions of

[[Page 3150]]

health events were 12.5 [micro]g/m\3\ and 11.5 [micro]g/m\3\,
respectively, for Zanobetti and Schwartz (2009) and for Bell et al.
(2008), with the 10th percentiles being lower by approximately 2
[micro]g/m\3\ in each study (Rajan et al., 2011, Table 1). In
considering this information, the EPA recognizes, however, that there
is no clear dividing line or single percentile within a given
distribution (including both above and below the 25th percentile)
provided by the scientific evidence that is most appropriate or
`correct' to use to characterize where the degree of confidence in the
associations warrants setting the annual standard level. The decision
as to the appropriate standard level below the long-term mean
concentrations of the key studies is largely a public health policy
judgment to be made by the Administrator, taking into account all of
the evidence and its related uncertainties, as discussed in section
III.E.4.d below.
In response to concerns that this analysis was not reviewed by
CASAC nor published in the peer-reviewed literature, the EPA notes that
this analysis was conducted to directly respond to advice from CASAC,
as discussed in section III.E.4.b.i above, in conjunction with their
review of the Policy Assessment. The EPA notes that the same type of
distributional analysis was presented in the second draft Policy
Assessment based on air quality data, as well as population-weighted
air quality data, rather than health event or study population data. In
considering that distributional information, CASAC urged that the EPA
redo the analysis using health event or study population data, which is
exactly what the EPA did and presented in the final Policy Assessment.
The EPA provided CASAC with the final Policy Assessment and
communicated how the final staff conclusions reflected consideration of
its advice and that those staff conclusions were based in part on the
specific distributional analysis that CASAC had urged the EPA to
conduct (Wegman, 2011, Attachment p. 2). CASAC did not choose to
provide any additional comments or advice after receiving the final
Policy Assessment. The EPA considers this distributional analysis to be
the product of the peer review conducted by CASAC of the Policy
Assessment, and thus does not agree with commenters' characterization
that the analysis lacked appropriate peer review. The EPA's final
analysis was based on the comments provided by CASAC, the peer review
committee established pursuant to the CAA, on the draft analysis, such
that the final analysis stems directly from CASAC's advice and the
EPA's response to its comments.
Based on the above considerations, the EPA continues to conclude
that its analysis of distributions of health event and air quality data
from two key multi-city epidemiological studies provides important
information related to understanding the associations between health
events observed in each city (e.g., deaths, hospitalizations) and the
corresponding long-term mean PM2.5 concentrations observed
in the studies. While recognizing that this is a relatively modest data
set, the EPA further concludes that such information can appropriately
help to inform the selection of the level of an annual standard that
will protect public health with an adequate margin of safety from these
types of health effects which are causally related to long- and short-
term exposures to PM2.5.
(4) Some commenters in this group asserted there were limitations
in the long-term exposure studies of morbidity, including studies
evaluating respiratory effects in children. For example, one commenter
(UARG, 2012, p. 12, Attachment 1, pp. 14 to 16) asserted there were
serious limitations in the long-term exposure studies of respiratory
morbidity in each of the studies considered by the EPA (including
McConnell et al., 2003; Gauderman et al., 2004; Dockery et al., 1996;
Raizenne et al., 1996; and Goss et al., 2004) and argued that this
evidence provides only a ``weak association'' with PM2.5
exposures. This commenter asserted that many of these long-term
exposure studies evaluating respiratory effects were considered at the
time the EPA reaffirmed the current annual standard level of 15
[micro]g/m\3\ in 2006, that the Administrator in the last review
determined that the information they provided ``was too limited to
serve as the basis for setting a level of a national standard,'' and
that they should be given little weight in setting the level of the
annual standard in this review (UARG, 2012, Attachment 1, p. 14).
More specifically, this commenter asserted that the McConnell et
al. (2003) and Gauderman et al. (2004) studies reported mixed results
for associations with PM2.5 and stronger associations with
NO2 (API, 2012, Attachment 1, pp. 14 to 15). Similarly, this
commenter argued that the Dockery et al. (1996) and Raizenne et al.
(1996) studies showed stronger associations with acidity than with fine
particles (measured as PM2.1). Id. pp. 15 to 16. With regard
to the cystic fibrosis study, this commenter noted that the association
between pulmonary exacerbations and PM2.5 in this study was
no longer statistically significant when the model adjusted for each
individual's baseline lung function. The commenters referred to the
data on lung function as an ``important explanatory variable,'' and
suggested that the EPA should rely on results from the model that
included individual baseline lung function information. Id. p. 16. For
the reasons discussed below and in more detail in the Response to
Comments document, the EPA disagrees with the commenters'
interpretation of these studies.
As an initial matter, the EPA notes that three of these studies
(McConnell et al., 2003; Dockery et al., 1996; Raizenne et al., 1996)
as well as the initial studies from the Southern California Children's
Health Study (Peters et al., 1999; McConnell et al., 1999; Gauderman et
al., 2000, 2002; Avol et al., 2001) were discussed and considered in
the 2004 Air Quality Criteria Document (U.S. EPA, 2004) and, thus,
considered within the air quality criteria supporting the EPA's final
decisions in the review completed in 2006. Two additional studies
(Gauderman et al., 2004; Goss et al., 2004) were discussed and
considered in the provisional science assessment conducted for the last
review (U.S. EPA, 2006a). The EPA concluded that ``new'' studies
considered in the provisional assessment completed in 2006 did not
materially change any of the broad scientific conclusions regarding the
health effects of PM exposure made in the Criteria Document (71 FR
61148 to 61149, October 17, 2006). All of these studies were considered
in the Integrated Science Assessment that informs the current review
(U.S. EPA, 2009a).
With regard to the Southern California Children's Health Study,
extended analyses considered in the Integrated Science Assessment
provided evidence that clinically important deficits in lung function
\97\ associated with long-term exposure to PM2.5 persist
into early adulthood (U.S. EPA, 2009a, p. 7-27; Gauderman et al.,
2004). These effects remained positive in copollutant models.\98\
Additional analyses of the

[[Page 3151]]

Southern California Children's Health Study cohort reported an
association between long-term PM2.5 exposure and bronchitic
symptoms (U.S. EPA, 2009a, p. 7-23 to 7-24; McConnell et al., 2003,
long-term mean concentration of 13.8 [micro]g/m\3\) that remained
positive in co-pollutant models, with the PM2.5 effect
estimates increasing in magnitude in some models and decreasing in
others, and a strong modifying effect of PM2.5 on the
association between lung function and asthma incidence (U.S. EPA,
2009a, 7-24; Islam et al., 2007). The outcomes observed in the more
recent reports from the Southern California Children's Health Study,
including evaluation of a broader range of endpoints and longer follow-
up periods, were larger in magnitude and more precise than reported in
the initial version of the study. Supporting these results were new
longitudinal cohort studies conducted by other researchers in varying
locations using different methods (U.S. EPA, 2009a, section 7.3.9.1).
The EPA, therefore, disagrees with the commenters that the studies by
McConnell et al. (2003) and Gauderman et al. (2004) are flawed and
should not be used in the PM NAAQS review process.
---------------------------------------------------------------------------

\97\ Clinical significance was defined as an FEV1
below 80 percent of the predicted value, a criterion commonly used
in clinical settings to identify persons at increased risk for
adverse respiratory conditions (U.S. EPA, 2009a, p. 7-29 to 7-30).
The primary NAAQS for sulfur dioxide (SO2) also included
this interpretation for FEV1 (75 FR 35525, June 22,
2010).
\98\ Gauderman et al. (2004) clearly stated throughout their
analysis that NO2 was one component of a highly
correlated mixture that contains PM2.5. Gauderman et al.
(2004) did not present the results from copollutants models but
stated ``two-pollutant models for any pair of pollutants did not
provide a significantly better fit to the data than the
corresponding single-pollutant models.''
---------------------------------------------------------------------------

The 24-City study \99\ by Dockery et al. (1996) (long-term mean
concentration of 14.5 [micro]g/m\3\) was considered in the current as
well as two previous reviews (U.S. EPA, 2009a; U.S. EPA, 2004; U.S.
EPA, 1996). This study observed that PM, specifically ``particle strong
acidity'' and sulfate particles (indicators of fine particles), were
associated with reports of bronchitis in the previous year. Similarly,
the magnitude of the associations between bronchitis and
PM10 and PM2.1 were similar to those for acidic
aerosols and sulfate particles, though the confidence intervals for the
PM10 and PM2.1 associations were slightly wider
and the associations were not statistically significant. Acid aerosols,
sulfate, and fine particles are formed in secondary reactions of the
emissions from incomplete combustion and these pollutants have similar
regional and temporal distributions. As noted by the study authors,
``the strong correlations of several pollutants in this study,
especially particle strong acidity with sulfate (r=0.90) and
PM2.1 (r=0.82), make it difficult to distinguish the agent
of interest'' (Dockery et al., 1996, p. 505). Overall, Dockery et al.
(1996) (and, similarly, Raizenne et al., 1996) observed similar
associations between respiratory health effects and acid aerosols,
sulfate, PM10 and PM2.1 concentrations. The
commenters noted that the associations with particle acidity were
sensitive to the inclusion of the six Canadian sites. The EPA notes
that none of these Canadian cities were in the ``sulfate belt'' where
particle strong acidity was highest. Thus, the change in the effect
estimate when the six Canadian cities were excluded from the analysis
is likely due to the lower prevalence of bronchitis and the lower
concentrations of acid aerosols in these cities, and not due to some
difference in susceptibility to bronchitis between the U.S. and
Canadian populations that is not due to air pollution, as suggested by
the commenters (UARG, 2012, Attachment 1, p. 15). In fact, contrary to
the statements made by the commenters, the authors did not observe any
subgroups that appeared to be markedly more susceptible to the risk of
bronchitis.
---------------------------------------------------------------------------

\99\ The 24-City study conducted by Dockery et al. (1996)
included 18 sites in the U.S. and 6 sites in Canada. The Raizenne et
al. (1996) study considered 22 of these 24 study areas. Athens, OH
and South Brunswick, NJ were not included in this study.
---------------------------------------------------------------------------

The Goss et al. (2004) study considered a U.S. cohort of cystic
fibrosis patients and provided evidence of association between long-
term PM2.5 exposures and exacerbations of respiratory
symptoms resulting in hospital admissions or use of home intravenous
antibiotics (U.S. EPA, 2009a, p. 7-25; long-term mean concentration of
13.7 [micro]g/m\3\). The commenters noted that the association between
pulmonary exacerbations and PM2.5 in this study was no
longer statistically significant when the model adjusted for each
individual's baseline lung function. The commenters referred to the
data on lung function as an ``important explanatory variable,'' and
suggested that the EPA should rely on results from the model that
included individual baseline lung function information. The EPA
disagrees with the commenters' interpretation of this study. The Agency
concludes it is unlikely that lung function is a potential confounder
or an important explanatory variable in this study. In fact, the
authors noted that ``it is more likely that lung function decline may
be intimately associated with chronic exposure to air pollutants and
may be part of the causal pathway in worsening prognosis in CF [cystic
fibrosis]; in support of this explanation, we found both cross-
sectional and longitudinal strong inverse relationships between
FEV1 and PM levels'' (Goss et al., 2004, p. 819). The EPA
notes that adjusting for a variable that is on the causal pathway can
lead to overadjustment bias, which is likely to attenuate the
association (Schisterman et al. 2009); this is likely what was observed
by the authors. Thus, the EPA continues to believe it is appropriate to
focus on the results reported in Goss et al. (2004) that did not
include individual baseline lung function in the model.
In addition, the EPA disagrees with commenters' reliance solely on
statistical significance when interpreting the study results from
individual study results and the collective evidence across studies. As
discussed in section III.D.2 above, statistical significance of
individual study findings has played an important role in the EPA's
evaluation of the study's results and the EPA has placed greater
emphasis on studies reporting statistically significant results.
However, in the broader evaluation of the evidence from many
epidemiological studies, and subsequently during the process of forming
causality determinations in the Integrated Science Assessment by
integrating evidence from across epidemiological, controlled human
exposure, and toxicological studies, the EPA has emphasized the pattern
of results across epidemiological studies and whether the effects
observed were coherent across the scientific disciplines for drawing
conclusions on the relationship between PM2.5 and different
health outcomes.
As noted in section III.B.1.a of the proposal, with regard to
respiratory effects, the Integrated Science Assessment concluded that
extended analyses of studies available in the last review as well as
new epidemiological studies conducted in the U.S. and abroad provided
stronger evidence of respiratory-related morbidity associated with
long-term PM2.5 exposure (77 FR 38918). The strongest
evidence for respiratory-related effects available in this review was
from epidemiological studies that evaluated decrements in lung function
growth in children and increased respiratory symptoms and disease
incidence in adults (U.S. EPA, 2009a, sections 2.3.1.2, 7.3.1.1, and
7.3.2.1).
In considering the collective evidence from epidemiological,
toxicological, and controlled human exposure studies, including the
studies discussed above, the EPA recognizes that the Integrated Science
Assessment concluded that a causal relationship is likely to exist
between long-term PM2.5 exposures and respiratory effects
(U.S. EPA, 2009a, p. 2-12, pp. 7-42 to 7-43). CASAC concurred with this
causality determination (Samet, 2009f, p.9).

[[Page 3152]]

The commenter's assertion that the EPA should adhere to its
assessment of these studies as it did in the review completed in 2006
is significantly mistaken. Most obviously, the EPA's final decision in
the last review was held to be deficient by the DC Circuit in remanding
the 2006 primary annual PM2.5 standard. As discussed in
section III.A.2 above, the DC Circuit specifically held that the EPA
did not provide a reasonable explanation of why certain morbidity
studies, including an earlier study from the Southern California
Children's Health Study (Gauderman et al., 2000, long-term mean
PM2.5 concentration approximately 15 [mu]g/m\3\) and the 24-
Cities Study (Raizenne et al., 1996, long-term mean concentrations
approximately 14.5 [micro]g/m\3\) did not warrant a more stringent
annual PM2.5 standard when the long-term mean
PM2.5 concentrations reported in those studies were at or
lower than the level of the annual standard. American Farm Bureau
Federation v. EPA. 559 F. 3d at 525. Indeed, the court found that,
viewed together, the Gauderman et al. (2000) and Raizenne et al.,
(1996) studies ``are related and together indicate a significant public
health risk * * * On this record, therefore, it appears the EPA too
hastily discounted the Gauderman and 24-Cities studies as lacking in
significance.'' Id.
In this review, the EPA recognizes a significant amount of evidence
beyond these two studies that expands our understanding of respiratory
effects associated with long-term PM2.5 exposures. This body
of scientific evidence includes an extended and new analyses from the
Southern California Children's Health Study (Gauderman et al., 2004;
Islam et al., 2007; Stanojevic et al., 2008) as well as additional
studies that examined these health effects (Kim et al., 2004; Goss et
al., 2004). Thus, even more so than in the last review, the evidence
indicates a ``significant public health risk'' to children from long-
term PM2.5 exposures at concentrations below the level of
the current annual standard. A standard that does not reflect
appropriate consideration of this evidence would not be requisite to
protect public health with an adequate margin of safety.
(5) With regard to the use of studies of health effects for which
the EPA finds the evidence to be ``suggestive'' of a causal
relationship, some commenters argued that such studies ``do not merit
any weight in the setting of the annual NAAQS'' (e.g., UARG, 2012,
Appendix 1, p. 3).
The EPA disagrees with the commenter's view that studies of health
effects for which the evidence is suggestive of a causal relationship,
rather than studies of health effects for which the evidence supports a
causal or likely causal relationship, merit no weight at all in setting
the NAAQS. To place no weight at all on such evidence would in essence
treat such evidence as though it had been categorized as ``not likely
to be a causal relationship.'' To do so would ignore the important
distinctions in the nature of the evidence supporting these different
causality determinations in the Integrated Science Assessment. It would
also ignore the CAA requirement that primary standards are to be set to
provide protection with an adequate margin of safety, including
providing protection for at-risk populations. Thus, ignoring this
information in making decisions on the appropriate standard level would
not be appropriate.\100\ Nonetheless, in considering studies of health
effects for which the evidence is suggestive of a causal relationship,
the EPA does believe that it is appropriate to place less weight on
such studies than on studies of health effects for which there is
evidence of a causal or likely causal relationship.
---------------------------------------------------------------------------

\100\ As discussed in section II.A above, the requirement that
primary standards provide an adequate margin of safety was intended
to address uncertainties associated with inconclusive scientific and
technical information available at the time of standard setting. I
was also intended to provide a reasonable degree of protection
against hazards that research has not yet identified. This certainly
encompasses consideration of effects for which there is evidence
suggestive of a causal relationship.
---------------------------------------------------------------------------

A second group of commenters supported revising the suite of
primary PM2.5 standards to provide increased public health
protection. These commenters found the available scientific information
and technical analyses to be stronger and more compelling than in the
last review. These commenters generally placed substantial weight on
CASAC advice and on the EPA staff analyses presented in the final
Policy Assessment, which concluded that the evidence most strongly
supported an annual standard level within a range of 11 to 12 [mu]g/
m\3\ (U.S. EPA, 2011a, p. 2-206). While some of these commenters felt
that the level should be set within the proposed range (12 to 13 [mu]g/
m\3\), most of these commenters advocated a level of 11 [mu]g/
m\3\.\101\ For example, ALA et al., asserted:
---------------------------------------------------------------------------

\101\ As discussed in section III.E.4.c.ii, many of these
commenters also supported lowering the level of the primary 24-hour
PM2.5 standard.

The EPA's proposed PM2.5 standards, while a step in
the right direction are insufficient to protect public health,
including the health of susceptible populations, with an adequate
margin of safety as required by the Clean Air Act * * *we will
discuss the enormous gap in public health protection afforded by an
annual standard of 13 [micro]g/m\3\, at the upper end of the
proposed range, compared to the more protective 11 [micro]g/m\3\, as
---------------------------------------------------------------------------
advocated by our organizations (ALA et al., 2012, p. 6).

In general, these commenters expressed the view that given the
strength of the available scientific evidence, the serious nature of
the health effects associated with PM2.5 exposures, the
large size of the at-risk populations, the risks associated with long-
and short-term PM2.5 exposures, and the important
uncertainties inherently present in the evidence, the EPA should follow
a highly precautionary policy response by selecting an annual standard
level that incorporates a large margin of safety.
More specifically, these commenters offered a range of comments
related to the general approach used by the EPA to select standard
levels, including: (1) The EPA's approach for setting a generally
controlling annual standard; (2) the importance of the greatly expanded
and stronger overall scientific data base; (3) consideration of the
distributional statistical analysis conducted by the EPA and other
approaches for translating the air quality information from specific
epidemiological studies into standard levels; and (4) the significance
of the PM2.5-related public health impacts, especially
potential impacts on at-risk populations, including children, in
reaching judgments on setting standards that provide protection with an
adequate margin of safety. These comments are discussed in turn below.
(1) Some of these commenters disagreed with the EPA's approach for
setting a ``generally controlling'' annual standard in conjunction with
a 24-hour standard providing supplemental protection particularly for
areas with high peak-to-mean ratios. These commenters argued this
approach would lead to ``regional inequities'' as demonstrated in the
EPA's analyses contained in Appendix C of the Policy Assessment (ALA et
al., pp. 26 to 27). Specifically, these commenters argued:

There is no basis in the Clean Air Act for such a determination.
The Clean Air Act requires only that the NAAQS achieve public health
protection with an adequate margin of safety. It is well-documented
that both long- and short-term exposures to PM2.5 have
serious and sometimes irreversible health impacts. There is no
health protection reason to argue that one standard should be
``controlling'' as a matter of policy without regard to the health
consequences of such a policy. To adopt such a policy ignores the
obligation to provide equal protection under

[[Page 3153]]

the law to all Americans because it would result in uneven
protection from air pollution in different localities and regions of
the country (ALA et al., 2012, p. 26).

The EPA believes these commenters misunderstood the basis for the
EPA's policy goal of setting a ``generally controlling'' annual
standard. This approach relates exclusively to setting standards that
will provide requisite protection against effects associated with both
long- and short-term PM2.5 exposures. It does so by lowering
the overall air quality distributions across an area, recognizing that
changes in PM2.5 air quality designed to meet an annual
standard would likely result not only in lower annual mean
PM2.5 concentrations but also in fewer and lower peak 24-
hour PM2.5 concentrations. As discussed in section III.A.3
in the proposal and above, the EPA recognizes that there are various
ways to combine the two primary PM2.5 standards to achieve
an appropriate degree of public health protection. Furthermore, the
extent to which these two standards are interrelated in any given area
depends in large part on the relative levels of the standards, the
peak-to-mean ratios that characterize air quality patterns in an area,
and whether changes in air quality designed to meet a given suite of
standards are likely to be of a more regional or more localized nature.
In focusing on an approach of setting a generally controlling
annual standard, the EPA's intent is in fact to avoid the potential
``regional inequities'' that are of concern to the commenters. The EPA
judges that the most appropriate way to set standards that provide more
consistent public health protection is by using the approach of setting
a generally controlling annual standard. This judgment builds upon
information presented in the Policy Assessment as discussed in section
III.A.3 above. More specifically, the Policy Assessment recognized that
the short-term exposure studies primarily evaluated daily variations in
health effects with monitor(s) that measured the variation in daily
PM2.5 concentrations over the course of several years. The
strength of the associations observed in these epidemiological studies
was demonstrably in the numerous ``typical'' days within the air
quality distribution, not in the peak days (U.S. EPA, 2011a, p. 2-9).
In addition, the quantitative risk assessments conducted for this and
previous reviews demonstrated the same point, that is, much, if not
most, of the aggregate risk associated with short-term PM2.5
exposures results from the large number of days during which the 24-
hour average concentrations are in the low-to mid-range, below the peak
24-hour concentrations (U.S. EPA, 2011a, section 2.2.2; U.S. EPA,
2010a, section 3.1.2.2). In addition, there was no evidence suggesting
that risks associated with long-term exposures were likely to be
disproportionately driven by peak 24-hour concentrations.\102\
---------------------------------------------------------------------------

\102\ In confirmation, a number of studies have presented
analyses excluding higher PM concentration days and reported a
limited effect on the magnitude of the effect estimates or
statistical significance of the association (e.g., Dominici, 2006b;
Schwartz et al., 1996; Pope and Dockery, 1992).
---------------------------------------------------------------------------

For these reasons, the Policy Assessment concluded that strategies
that focused primarily on reducing peak days were less likely to
achieve reductions in the PM2.5 concentrations that were
most strongly associated with the observed health effects. Furthermore,
the Policy Assessment concluded that an approach that focused on
reducing peak exposures would most likely result in more uneven public
health protection across the U.S. by either providing inadequate
protection in some areas or overprotecting in other areas (U.S. EPA,
2011a, p. 2-9; U.S. EPA, 2010a, section 5.2.3). This is because
reductions based on control of peak days are less likely to control the
bulk of the air quality distribution.
As a result, the EPA believes an approach that focuses on a
generally controlling annual standard would likely reduce aggregate
risks associated with both long- and short-term exposures more
consistently than a generally controlling 24-hour standard and,
therefore, would be the most effective and efficient way to reduce
total PM2.5-related population risk. The CASAC agreed with
this approach and considered it was ``appropriate to return to the
strategy used in 1997 that considers the annual and the short-term
standards together, with the annual standard as the controlling
standard, and the short-term standard supplementing the protection
afforded by the annual standard'' (Samet, 2010d, p. 1). For the reasons
discussed above, the EPA disagrees with the comments that this approach
will result in the concerns raised by the commenters; rather the EPA
concludes that this approach will help to address these concerns.
(2) Many of these commenters asserted that the currently available
scientific information is greatly expanded and stronger compared to the
last review. Some of these commenters highlighted the availability of
multiple, multi-city long- and short-term exposure studies providing
``repeated, consistent evidence of effects below the current annual
standard level'' (ALA et al., 2012, pp. 39 to 49) and, more
specifically, ``significant evidence of harm with strong confidence
well below EPA's proposed annual standard range of 12-13 [mu]g/m\3\''
(AHA et al., 2012, pp. 10 to 12).
The EPA recognizes that in setting standards that are requisite to
protect public health with an adequate margin of safety, the
Administrator must weigh the various types of available scientific
information in reaching public health policy judgments that neither
overstate nor understate the strength and limitations of this
information or the appropriate inferences to be drawn from the
available science.
In general, the EPA agrees with these commenters' views that the
currently available scientific evidence is stronger ``because of its
breadth and the substantiation of previously observed health effects''
(77 FR 38906/2) and provides ``greater confidence in the reported
associations than in the last review'' (77 FR 38940/1). The EPA also
agrees with the commenters' position that it is appropriate to consider
the regions within the broader air quality distributions where we have
the strongest confidence in the associations reported in
epidemiological studies in setting the level of the annual standard.
However, as discussed in section III.E.4.d below, in weighing the
available evidence and technical analyses, as well as the associated
uncertainties and limitations in that information, the EPA disagrees
with the commenters' views regarding the extent to which the available
scientific information provides support for considering an annual
standard level below the proposed range (i.e., below 12 to 13 [mu]g/
m\3\). In particular, the EPA disagrees with the degree to which these
commenters place more weight on the relatively more uncertain evidence
that is suggestive of a causal relationship (e.g., low birth weight).
Consistent with CASAC advice (Samet, 2010d, p. 1), the Agency concludes
it is appropriate and reasonable to place the greatest emphasis on
health effects for which the Integrated Science Assessment concluded
there is evidence of a causal or likely causal relationship and to
place less weight on the health effects that provide evidence that is
only suggestive of a causal relationship.
(3) With regard to using the air quality information from
epidemiological studies to inform decisions on standard levels,
commenters in this group generally supported the EPA's efforts to
explore different statistical metrics from

[[Page 3154]]

epidemiological studies to inform the Administrator's decisions. These
commenters argued that by considering different analytic measures--
either concentrations one standard deviation below the long-term means
reported in the epidemiological studies or the EPA's distributional
statistical analysis of population-level data that extends the approach
used in previous PM NAAQS reviews to consider information beyond a
single statistical metric--``the annual standard must be significantly
lower than EPA has proposed'' (ALA et al., 2012, pp. 50 to 61).
Furthermore, with regard to characterizing the PM2.5 air
quality at which associations have been observed, some of these
commenters highlighted CASAC's recommendation that ``[f]urther
consideration should be given to using the 10th percentile as a level
for assessing various scenarios of levels for the PM NAAQS'' (Samet,
2010c, p. 11) (ALA et al., 2012, p. 55). Other commenters urged that
the EPA extend the distributional analysis to include additional
studies. For example, CHPAC urged the EPA to also conduct
distributional analysis for children's health studies to better inform
standards that would protect both children and adults from adverse
health outcomes (CHPAC, 2012, p. 3).
The EPA agrees with these commenters' views that it is appropriate
to take into account different statistical metrics from epidemiological
studies to inform the decisions on standard levels that are appropriate
to consider in setting a standard that will protect public health with
an adequate margin of safety. In the development of the Policy
Assessment, the EPA staff explored various approaches for using
information from epidemiological studies in setting the standards. The
general approach used in the final Policy Assessment, discussed in
sections III.A.3 and III.E.4.a above, reflects consideration of CASAC
advice (Samet, 2010c, d) and public comments on multiple drafts of the
Policy Assessment.
With regard to using the distributional statistical analysis to
characterize the confidence in the associations, the EPA emphasizes
that there is no clear dividing line provided by the scientific
evidence, and that choosing how far below the long-term mean
concentrations from the epidemiological studies is appropriate to
identify a standard level that will provide protection for the public
health with an adequate margin of safety is largely a public health
policy judgment. The EPA considers the region from approximately the
25th to 10th percentiles to be a reasonable range for providing a
general frame of reference as to the part of the distribution over
which our confidence in the magnitude and significance of the
associations observed in epidemiological studies is appreciably lower.
Based on these considerations, the EPA concludes that it is not
appropriate to place as much confidence in the magnitude and
significance of the associations over the lower percentiles of the
distributions in each study as at and around the long-term mean
concentrations. Thus, the EPA disagrees with the commenters' views that
this analysis compels placing more emphasis on the lower part of this
range in selecting a level for an annual standard that will protect
public health with an adequate margin of safety. The EPA recognizes
that this information comes primarily from two short-term exposure
studies, a relatively modest data set. In light of the limited nature
of this information, and in recognition of more general uncertainties
inherent in the epidemiological evidence, the Administrator deems it
reasonable not to place more emphasis on concentrations in the lower
part of this range, as discussed below in section III.E.4.d.
With regard to the scope of the distributional statistical
analysis, the EPA requested additional population-level data from the
study authors for a group of six multi-city studies for which previous
air quality analyses had been conducted (Hassett-Sipple et al., 2010;
Schmidt et al., 2010, Analysis 2). These six studies were originally
selected because they considered multiple locations representing
varying geographic regions across multiple years. Thus, these studies
provided evidence on the influence of different particle mixtures on
health effects associated with long- and short-term PM2.5
exposures. In addition, these multi-city studies considered relatively
more recent health events and air quality conditions (1999 to 2005). As
discussed in section III.E.4.b.i above, the EPA received and analyzed
population-level data for four of the six studies (Rajan et al., 2011).
Three of these four studies (Krewski et al., 2009; Bell et al., 2008;
Zanobetti and Schwartz, 2009) served as the basis for the
concentration-response functions used to develop the core risk
estimates (U.S. EPA, 2010a, section 3.3.3). While, the EPA agrees that
it would be useful to have such data from more studies, the Agency
believes that the additional data that was requested and received from
study authors provide useful information to help inform the
Administrator's selection of the annual standard level.
(4) Many commenters in this group highlighted PM2.5-
related impacts on at-risk populations, including potential impacts on
children, older adults, persons with pre-existing heart and lung
disease, and low-income populations, to support their views that the
annual standard should be revised to a level of 11 [mu]g/m\3\ or lower
(CHPAC, 2012; AHA et al., 2012; ALA, 2012, pp. 29 to 38; Rom et al.,
2012; Air Alliance Houston, et al., 2012). These commenters urged the
EPA to adopt a policy approach that placed less weight on the remaining
uncertainties and limitations in the evidence and placed more emphasis
on margin of safety considerations, including providing protection
against effects for which there is more limited scientific evidence.
For example, CHPAC urged the EPA ``to place the same weight on studies
examining impacts on children's health as that of adult studies. * * *
The fact that there may be stronger evidence from adult studies does
not mean that standards based on adult studies will be protective for
children and consequently will meet the standard requisite to protect
public health with an adequate margin of safety'' (CHPAC, 2012 p. 3).
Furthermore, with regard to the EPA's approach for weighing
uncertainties, some of these commenters stated that ``we find no
justification in the preamble for an annual standard level as high as
13 [mu]g/m\3\, other than the vague assertion that uncertainties
increase at lower concentrations. Further, the final proposal
completely failed to address the Policy Assessment recommendations that
if 13 [mu]g/m\3\ was proposed, the 24-hour standard should be
strengthened as well'' (ALA et al., p. 7).
The EPA has carefully evaluated and considered evidence of effects
in at-risk populations. With regard to effects classified as having
evidence of a causal or likely causal relationship with long- or short-
term PM2.5 exposures (i.e., premature mortality,
cardiovascular effects, and respiratory effects), the Agency takes note
that it considered the full range of studies evaluating these effects,
including studies of at-risk populations, to inform its review of the
primary PM2.5 standards. Specific multi-city studies
summarized in Figures 1, 2, and 3 above highlight evidence of effects
observed in two different lifestages--children and older adults--that
have been identified as at-risk populations. Thus, the EPA places as
much weight on studies that explored effects in children for which the
evidence is causal or likely causal in

[[Page 3155]]

nature as on studies of such effects in adults, including older adults.
As discussed above in responses to commenters supporting the retention
of the current standards, in setting the standard, the EPA has focused
on considering PM2.5 concentrations somewhat below the
lowest long-term mean concentrations from each of the key studies of
both long- and short-term exposures of effects for which the evidence
supports a causal or likely causal relationship (i.e., the first two
sets of studies shown in Figure 4). Absent some reason to ignore or
discount these studies, which the commenter does not provide (and of
which the EPA is unaware), the EPA considers the available evidence of
effects in children as well as other at-risk populations.
With respect to the EPA's consideration of more limited studies
providing evidence suggestive of a causal relationship (e.g.,
developmental and reproductive effects), as noted above in responding
to comments from the first group of commenters, the Agency agrees that
it is important to place some weight on this body of evidence in
setting standards that provide protection for at-risk populations, as
required by the CAA. However, the Agency does not agree that the same
weight must be placed on this information as on the body of scientific
information for which there is evidence of a causal or likely causal
relationship. To do so would ignore the difference in the breadth and
strength of the evidence supporting the different causality
determinations reached in the Integrated Science Assessment.
With regard to weighing the uncertainties and limitations remaining
in the evidence and technical analyses, as discussed in section II.A
above, the EPA recognizes that in setting a primary NAAQS that provides
an adequate margin of safety, the Administrator must consider a number
of factors including the nature and severity of the health effects
involved, the size of sensitive population(s) at risk, and the kind and
degree of the uncertainties that remain. As discussed in section
III.E.4.d below, the Agency agrees with these commenters that, in
weighing the available evidence and technical analyses including the
uncertainties and limitations in this scientific information, there is
no justification for setting a primary PM2.5 annual standard
level as high as 13 [mu]g/m\3\.
Finally, some commenters in both groups also identified ``new''
studies that were not included in the Integrated Science Assessment as
providing further support for their views on the level of the annual
standard. As discussed in section II.B.3 above, the EPA completed a
provisional review and assessment of ``new'' studies published since
the close of the Integrated Science Assessment, including ``new''
studies submitted by commenters (U.S. EPA, 2012b). The provisional
assessment found that the ``new'' studies expand the scientific
information considered in the Integrated Science Assessment and provide
important insights on the relationship between PM2.5
exposure and health effects of PM (U.S. EPA, 2012b). However, the EPA
notes that the provisional assessment found that the ``new'' science
did not materially change the conclusions reached in the Integrated
Science Assessment. The EPA notes that, as in past NAAQS reviews, the
Agency is basing the final decisions in this review on the studies and
related information included in the Integrated Science Assessment that
have undergone CASAC and public review, and will consider newly
published studies for purposes of decision making in the next PM NAAQS
review.
ii. 24-Hour Standard Level
With respect to the level of the 24-hour standard, the EPA received
comments on the proposal from two distinct groups of commenters. One
group that included virtually all commenters representing industry
associations, businesses, and many States agreed with the Agency's
proposed decision to retain the level of the 24-hour PM2.5
standard. The other group of commenters included many medical groups,
numerous physicians and academic researchers, many public health
organizations, some State and local agencies, five State Attorneys
General, and a large number of individual commenters. These commenters
disagreed with the Agency's proposed decision and argued that EPA
should lower the level of the 24-hour standard to 30 or 25 [mu]g/m\3\.
Comments from these groups on the level of the 24-hour PM2.5
standard are addressed below and in the Response to Comments Document.
As noted above, of the public commenters who addressed the level of
the 24-hour PM2.5 standard, all industry commenters and most
State and local commenters supported the proposed decision to retain
the current level of 35 [mu]g/m\3\. In many cases, these groups agreed
with the rationale supporting the Administrator's proposed decision to
retain the current 24-hour PM2.5 standard, including her
emphasis on the annual standard as the generally controlling standard
with the 24-hour standard providing supplementary protection, and her
conclusion that multi-city, short-term exposure studies provide the
strongest data set for informing decisions on the appropriate 24-hour
standard level. Many of these commenters agreed with the
Administrator's view that the single-city, short-term studies provided
a much more limited data set (e.g., limited statistical power, limited
exposure data) and more equivocal results (e.g., mixed results within
the same study area), making them an unsuitable basis for setting the
level of the 24-hour standard.
While these commenters agreed with the EPA's proposed decision to
retain the current 24-hour PM2.5 standard, some did not
agree with the EPA's approach to considering the evidence from short-
term multi-city studies. For example, a commenter representing UARG
pointed out that the 98th percentile concentrations reported in the
proposal for multi-city studies reflect the averages of 98th percentile
concentrations across the cities included in those studies (UARG, 2012;
Attachment 1; p. 25). This commenter contended that such averaged 98th
percentile PM2.5 concentrations do not provide information
that can appropriately inform a decision on the adequacy of the public
health protection provided by the current or alternative 24-hour
standards.
While the EPA agrees that there is uncertainty in linking effects
reported in multi-city studies to specific air quality concentrations
(U.S. EPA, 2011a, section 2.3.4.1), the EPA disagrees with this
commenter's view that such uncertainty precludes the use of averaged
98th percentile PM2.5 concentrations to inform a decision on
the appropriateness of the protection provided by the 24-hour
PM2.5 standard. In particular, the EPA notes that averaged
98th percentile concentrations do provide information on the extent to
which study cities contributing to reported associations would likely
have met or violated the current 24-hour PM2.5 standard
during the study period. As evidence of this, the EPA notes the three
multi-city studies specifically highlighted by this commenter as having
averaged 98th percentile 24-hour PM2.5 concentrations below
35 [mu]g/m\3\ (Dominici et al., 2006a; Bell et al., 2008; Zanobetti and
Schwartz, 2009). Based on the 98th percentiles of 24-hour
PM2.5 concentrations in the individual cities evaluated in
these studies, the EPA notes that the majority of these study cities
would likely have met the current standard during the study periods
(Hassett-Sipple et al., 2010). Therefore, regardless of whether the
averaged 98th percentile concentrations or the 98th

[[Page 3156]]

percentile concentrations in each city are considered, these studies
provide evidence for associations between short-term PM2.5
and mortality or morbidity across a large number of U.S. cities, the
majority of which would likely have met the current 24-hour
PM2.5 standard during study periods. In their review of the
PM Policy Assessment, CASAC endorsed the conclusions drawn from
analyses of averaged 98th percentile 24-hour PM2.5
concentrations, and the EPA continues to conclude that this type of
information can appropriately inform the Administrator's decision on
the current 24-hour PM2.5 standard.\103\
---------------------------------------------------------------------------

\103\ This is not to say that the EPA's decision on whether to
revise the 24-hour PM2.5 standard should be based on or
only be informed by considerations of whether studies reported
associations with mortality or morbidity in areas with averaged 98th
percentile PM2.5 concentrations less than 35 mg/m\3\. As
discussed below, in reaching a decision in this final notice on the
most appropriate approach to strengthen the suite of
PM2.5 standards, the Administrator considers the degree
of public health protection provided by the combination of the
annual and 24-hour standards together.
---------------------------------------------------------------------------

Another group of commenters argued that the 24-hour standard level
should be lowered. Many of these commenters supported setting the level
of the 24-hour PM2.5 standard at either 25 or 30 [mu]g/m\3\.
In support of their position, the ALA et. al., AHA et al., five state
Attorneys General, and a number of additional groups pointed to 98th
percentile PM2.5 concentrations in locations of multi-city
and single-city epidemiological studies. For example, the ALA and
others pointed to multi-city studies by Dominici et al. (2006a),
Zanobetti and Schwartz (2009), Burnett et al. (2000), and Bell et al.
(2008) as providing evidence for associations with mortality and
morbidity in study locations with averaged (i.e., averaged across
cities) 98th percentile 24-hour PM2.5 concentrations below
35 [mu]g/m\3\. These commenters also pointed to several single-city and
panel studies reporting associations between short-term
PM2.5 and mortality or morbidity in locations with
relatively low 24-hour PM2.5 concentrations. Because some of
these multi- and single-city studies have reported associations with
health effects in locations with 98th percentile PM2.5
concentrations below 35 [mu]g/m\3\, commenters maintained that the
current 24-hour PM2.5 standard (i.e., with its level of 35
[mu]g/m\3\) does not provide an appropriate degree of protection in all
areas.
In further support of their position that the level of the current
24-hour standard should be lowered, these commenters pointed out the
variability across the U.S. in ratios of 24-hour to annual
PM2.5 concentrations. They noted that some locations,
including parts of the northwestern U.S., experience relatively low
annual PM2.5 concentrations but can experience relatively
high 24-hour concentrations at certain times of the year. In order to
provide protection against effects associated with short-term
PM2.5 exposures, especially in locations with high ratios of
24-hour to annual PM2.5 concentrations, these commenters
advocated setting a lower level for the 24-hour standard.
The EPA agrees with these commenters that it is appropriate to
maintain a 24-hour PM2.5 standard in order to supplement the
protection provided by the revised annual standard, particularly in
locations with relatively high ratios of 24-hour to annual
PM2.5 concentrations. However, in highlighting 98th
percentile PM2.5 concentrations in study locations without
also considering the impact of a revised annual standard on short-term
concentrations, these commenters ignore the fact that many areas would
be expected to experience decreasing short- and long-term
PM2.5 concentrations in response to a revised annual
standard.
In considering the specific multi-city studies highlighted by
public commenters who advocated a more stringent 24-hour standard, the
EPA notes that such studies have reported consistently positive and
statistically significant associations with short-term PM2.5
exposures in locations with averaged 98th percentile PM2.5
concentrations ranging from 45.8 to 34.2 [mu]g/m\3\ and long-term mean
PM2.5 concentrations ranging from 13.4 to 12.9 (Burnett and
Goldberg, 2003; Burnett et al., 2004; Dominici et al., 2006a; Bell et
al., 2008; Franklin et al., 2008; Zanobetti and Schwartz, 2009).\104\
The EPA notes that to the extent air quality distributions are reduced
to meet the current 24-hour standard with its level of 35 [mu]g/m\3\
and/or the revised annual standard with its level of 12 [mu]g/m\3\,
additional protection would be anticipated against the effects reported
in these short-term, multi-city studies. Put another way, to attain an
annual standard with a level below the long-term means in the locations
of these short-term studies (as EPA is adopting here), the overall air
quality distributions in the majority of study cities will necessarily
be reduced, resulting in lower daily PM2.5 ambient
concentrations. We therefore expect that the revised annual standard
will result in 98th percentile PM2.5 concentrations in these
cities that are lower than those measured in the studies, and that the
overall distributions of PM2.5 concentrations will be lower
than those reported to be associated with health effects. Thus, even
for effects reported in multi-city studies with averaged 98th
percentile concentrations below 35 [mu]g/m\3\, additional protection
from the risks associated with short-term exposures is anticipated from
the revised annual standard, without revising the 24-hour standard,
because long-term average PM2.5 concentrations in multi-city
study locations were above the level of the revised annual standard
(i.e., 12 [mu]g/m\3\).\105\ As discussed above, reducing the annual
standard is the most efficient way to reduce the risks from short-term
exposures identified in these studies, as the bulk of the risk comes
from the large number of days across the bulk of the air quality
distribution, not the relatively small number of days with peak
concentrations.
---------------------------------------------------------------------------

\104\ Commenters also highlighted associations with short-term
PM2.5 concentrations reported in sub-analyses restricted
to days with 24-hour concentrations at or below 35 [mu]g/m\3\
(Dominici, 2006b). These sub-analyses were not included in the
original publication by Dominici et al. (2006a). Authors provided
results of sub-analyses for the Administrator's consideration in a
letter to the docket following publication of the proposed rule in
January 2006 (personal communication with Dr. Francesca Dominici,
2006b). As noted in section III.A.3, these sub-analyses are part of
the basis for the conclusion that there is no evidence suggesting
that risks associated with long-term exposures are likely to be
disproportionately driven by peak 24-hour concentrations. Because
the sub-analyses did not present long-term average PM2.5
concentrations, it is not clear whether they reflected
PM2.5 air quality that would have been allowed by the
revised annual PM2.5 standard being established in this
rule.
\105\ It is also the case that additional protection is
anticipated in locations with 98th percentile 24-hour
PM2.5 concentrations above 35 [mu]g/m\3\, even if long-
term concentrations are below 12 [mu]g/m\3\. As noted in the
proposal and in the Policy Assessment (U.S. EPA, 2011a, Figure 2-
10), parts of the northwestern U.S. are more likely than other parts
of the country to violate the 24-hour standard and meet the revised
annual standard.
---------------------------------------------------------------------------

In considering the single-city studies highlighted by public
commenters who advocated a more stringent 24-hour standard, the EPA
first notes that, overall, these single-city studies reported mixed
results. Specifically, some studies reported positive and statistically
significant associations with PM2.5, some studies reported
positive but non-significant associations, and several studies reported
negative associations or a mix of positive and negative associations
with PM2.5. In light of these inconsistent results, the
proposal noted that the overall body of evidence from single-city
studies is mixed, particularly in locations with 98th percentiles of
24-hour concentrations below 35 [mu]g/m\3\. Therefore, although some
single-city

[[Page 3157]]

studies reported effects at appreciably lower PM2.5
concentrations than short-term multi-city studies, the uncertainties
and limitations associated with the single-city studies were noted to
be greater. In light of these greater uncertainties and limitations,
the Administrator concluded in the proposal that she had less
confidence in using these studies as a basis for setting the level of
the standard (77 FR 38943).
Given the considerations and conclusions noted above, in the
proposal the Administrator concluded that the short-term multi-city
studies provide the strongest evidence to inform decisions on the level
of the 24-hour standard. Further, she viewed single-city, short-term
exposure studies as a much more limited data set providing mixed
results, and she had less confidence in using these studies as a basis
for setting the level of a 24-hour standard (77 FR 38942). In
highlighting specific single-city studies, public health,
environmental, and State and local commenters appear to have
selectively focused on studies reporting associations with
PM2.5 and to have overlooked studies that reported more
equivocal results (e.g., Ostro et al., 2003; Rabinovitch et al., 2004;
Slaughter et al., 2005; Villeneuve et al., 2006) (U.S. EPA, 2011,
Figure 2-9). As such, these commenters have not presented new
information that causes the EPA to reconsider its decision to emphasize
multi-city studies over single-city studies when identifying the
appropriate level of the 24-hour PM2.5 standard.
In further considering the single-city studies highlighted by
public commenters, the EPA notes that some commenters advocating for a
lower level for the 24-hour PM2.5 standard also discussed
short-term studies that have been published since the close of the
Integrated Science Assessment. These recent studies were conducted in
single cities or in small panels of volunteers. As in prior NAAQS
reviews and as discussed above in more detail (section II.B.3), the EPA
is basing its decisions in this review on studies and related
information assessed in the Integrated Science Assessment. The studies
assessed in the Integrated Science Assessment, and the conclusions
based on those studies, have undergone extensive critical review by the
EPA, CASAC, and the public. The rigor of that review makes the studies
assessed in the Integrated Science Assessment, and the conclusions
based on those studies, the most reliable source of scientific
information on which to base decisions on the NAAQS.
However, as discussed above (section II.B.3), the EPA recognizes
that ``new studies'' may sometimes be of such significance that it is
appropriate to delay a decision on revision of a NAAQS and to
supplement the pertinent air quality criteria so the studies can be
taken into account. In the present case, the EPA's provisional
consideration of ``new studies'' concludes that, taken in context, the
``new'' information and findings do not materially change any of the
broad scientific conclusions made in the air quality criteria regarding
the health effects of PM2.5 (U.S. EPA, 2012b).
For this reason, reopening the air quality criteria review would
not be warranted, even if there were time to do so under the court
order governing the schedule for completing this review. Accordingly,
the EPA is basing its final decisions in this review on the studies and
related information included in the PM Integrated Science Assessment
(i.e., the air quality criteria) that has undergone CASAC and public
review. The EPA will consider the ``new studies'' in the next periodic
review of the PM NAAQS, which will provide an opportunity to fully
assess these studies through a more rigorous review process involving
the EPA, CASAC, and the public.
Some public health, medical, and environmental commenters also
criticized the EPA's interpretation of PM2.5 risk results.
These commenters presented risk estimates for combinations of annual
and 24-hour standards using more recent air quality data than that used
in the EPA's Risk Assessment (U.S. EPA, 2010a). Based on these
additional risk analyses, the ALA and other commenters contended that
public health benefits could continue to increase as annual and 24-hour
standard levels decrease below 13 [mu]g/m\3\ and 35 [mu]g/m\3\,
respectively.
The EPA agrees with commenters that important public health
benefits are expected as a result of revising the level of the annual
standard to 12 [mu]g/m\3\, as is done in this rule, rather than 13
[mu]g/m\3\. The Agency also acknowledges that estimated
PM2.5-associated health risks continue to decrease with
annual standard levels below 12 [mu]g/m\3\ and/or with 24-hour standard
levels below 35 [mu]g/m\3\. However, the EPA disagrees with the
commenters' views regarding the extent to which risk estimates support
setting standard levels below 12 [mu]g/m\3\ (annual standard) and 35
[mu]g/m\3\ (24-hour standard).\106\
---------------------------------------------------------------------------

\106\ This section focuses on the 24-hour standard. Section
III.E.4.c.i above also discusses these commenters' recommendations
within the context of the annual PM2.5 standard.
---------------------------------------------------------------------------

The CAA charges the Administrator with setting NAAQS that are
``requisite'' (i.e., neither more nor less stringent than necessary) to
protect public health with an adequate margin of safety. In setting
such standards the Administrator must weigh the available scientific
evidence and information, including associated uncertainties and
limitations. As described above, in reaching her proposed decisions on
the PM2.5 standards that would provide ``requisite''
protection, the Administrator carefully considered the available
scientific evidence and risk information, making public health policy
judgments that, in her view, neither overstated nor understated the
strengths and limitations of that evidence and information. In
contrast, as discussed more fully above, public health, medical, and
environmental commenters who recommended levels below 35 [mu]g/m\3\ for
the 24-hour PM2.5 standard have not provided new information
or analyses to suggest that such standard levels are appropriate, given
the uncertainties and limitations in the available health evidence,
particularly uncertainties in studies conducted in locations with 98th
percentile 24-hour PM2.5 concentrations below 35 [mu]g/m\3\
and long-term average concentrations below 12 [mu]g/m\3\.
d. Administrator's Final Conclusions on the Primary PM2.5
Standard Levels
In reaching her conclusions regarding appropriate standard levels,
the Administrator has considered the epidemiological and other
scientific evidence, estimates of risk reductions associated with just
meeting alternative annual and/or 24-hour standards, air quality
analyses, related limitations and uncertainties, the advice of CASAC,
and extensive public comments on the proposal. After careful
consideration of all of these, the Administrator has decided to revise
the level of the primary annual PM2.5 standard from 15.0
[mu]g/m\3\ to 12.0 [mu]g/m\3\ and to retain the level of the primary
24-hour standard at 35 [mu]g/m\3\.
As an initial matter, the Administrator agrees with the approach
supported by CASAC and discussed in the Policy Assessment as summarized
in sections III.A.3 and III.E.4.a above, of considering the annual and
24-hour standards together in determining the protection afforded
against mortality and morbidity effects associated with both long- and
short-term exposures to PM2.5. This approach is consistent
with the approach taken in the review

[[Page 3158]]

completed in 1997, in contrast to the approach used in the review
completed in 2006 where each standard was considered independently of
the other (i.e., only data from long-term exposure studies were used to
inform the level of the annual standard and only data from short-term
exposure studies were used to inform the level of the 24-hour
standard).\107\
---------------------------------------------------------------------------

\107\ See 71 FR 61148 and 61168, October 17, 2006.
---------------------------------------------------------------------------

Based on the evidence and quantitative risk assessment, the
Administrator concludes that it is appropriate to set an annual
standard that is generally controlling, which will lower the broad
distribution of 24-hour average concentrations in an area as well as
the annual average concentration, so as to provide protection from both
long- and short-term PM2.5 exposures. In conjunction with
this, it is appropriate to set a 24-hour standard focused on providing
supplemental protection, particularly for areas with high peak-to-mean
ratios of 24-hour concentrations, possibly associated with strong local
or seasonal sources, and for PM2.5-related effects that may
be associated with shorter-than daily exposure periods. The
Administrator concludes this approach will reduce aggregate risks
associated with both long- and short-term exposures more consistently
than a generally controlling 24-hour standard and is the most effective
and efficient way to reduce total PM2.5-related population
risk and to protect public health with an adequate margin of safety.
In selecting the level of the annual PM2.5 standard,
based on the characterization and assessment of the epidemiological and
other studies presented and assessed in the Integrated Science
Assessment and the Policy Assessment, the Administrator recognizes the
substantial increase in the number and diversity of studies available
in this review. This expanded body of evidence includes extended
analyses of the seminal studies of long-term PM2.5 exposures
(i.e., ACS and Harvard Six Cities studies) as well as important new
long-term exposure studies (as summarized in Figures 1 and 2).
Collectively, the Administrator notes that these studies, along with
evidence available in the last review, provide consistent and stronger
evidence than previously observed of an association between long-term
PM2.5 exposures and premature mortality in areas with lower
long-term ambient concentrations than previously observed, with the
strongest evidence related to cardiovascular-related mortality. The
Administrator also recognizes the availability of stronger evidence of
morbidity effects associated with long-term PM2.5 exposures,
including evidence of respiratory effects such as decreased lung
function growth, from the extended analyses for the Southern California
Children's Health Study and evidence of cardiovascular effects from the
WHI study. Furthermore, the Administrator recognizes new U.S. multi-
city studies that greatly expand and reinforce our understanding of
mortality and morbidity effects associated with short-term
PM2.5 exposures, providing stronger evidence of associations
in areas with ambient concentrations similar to those previously
observed in short-term exposure studies considered in the previous
review (as summarized in Figure 3).
The Administrator recognizes the strength of the scientific
evidence for evaluating health effects associated with fine particles,
noting that the newly available scientific evidence builds upon the
previous scientific data base to provide evidence of generally robust
associations and a basis for greater confidence in the reported
associations than in the last review. She notes the conclusion of the
Integrated Science Assessment that this body of evidence supports a
causal relationship between long- and short-term PM2.5
exposures and mortality and cardiovascular effects and a likely causal
relationship between long- and short-term PM2.5 exposures
and respiratory effects. In addition, the Administrator notes
additional, but more limited evidence, for a broader range of health
endpoints including evidence suggestive of a causal relationship for
developmental and reproductive effects as well as for carcinogenic
effects.
Based on information discussed and presented in the Integrated
Science Assessment, the Administrator recognizes that health effects
may occur over the full range of concentrations observed in the
epidemiological studies of both long-term and short-term exposures,
since no discernible population-level threshold for any such effects
can be identified based on the currently available evidence (U.S. EPA,
2009a, section 2.4.3). To inform her decisions on an appropriate level
for the annual standard that will protect public health with an
adequate margin of safety, in the absence of any discernible
population-level thresholds, the Administrator judges that it is
appropriate to consider the relative degree of confidence in the
magnitude and significance of the associations observed in
epidemiological studies across the range of long-term PM2.5
concentrations in such studies. Further, she recognizes, in taking note
of CASAC advice and the distributional statistics analysis discussed in
the Policy Assessment and in section III.E.4.a above, that there is
significantly greater confidence in the magnitude and significance of
observed associations for the part of the air quality distribution
corresponding to where the bulk of the health events evaluated in each
study have been observed, generally at and around the long-term mean
concentrations. Conversely, she also recognizes that there is
significantly diminished confidence in the magnitude and significance
of observed associations in the lower part of the air quality
distribution corresponding to where a relatively small proportion of
the health events were observed. Further, the Administrator recognizes
that the long-term mean concentrations, or any other specific point in
the air quality distribution of each study, do not represent a ``bright
line'' at and above which effects have been observed and below which
effects have not been observed.
In considering the long-term mean concentrations reported in
epidemiological studies, the Administrator recognizes that in selecting
a level of the annual standard that will protect public health with an
adequate margin of safety, it is not sufficient to focus on a
concentration generally somewhere within the range of long-term mean
concentrations from the key long-term and short-term exposure studies
that reported lower concentrations than had been observed in earlier
reviews. These key studies provide information for various types of
serious health endpoints (including mortality and morbidity effects),
different study populations (which may include at-risk populations such
as children and older adults), and different air quality distributions
that are specific to each study. A level somewhere within the range of
long-term mean concentrations of the full set of key studies would be
higher than the long-term mean of at least one of the studies being
considered and therefore would not provide a sufficient degree of
protection against the health effects observed in that study. Absent
some reasoned basis to place less weight on the evidence in the
epidemiological study with the lowest long-term mean concentration
among these key studies, this approach would not be consistent with the
requirement to set a standard that will protect public health with an

[[Page 3159]]

adequate margin of safety.\108\ Thus, the Administrator recognizes it
is important to protect against the serious effects observed in each of
these studies so as to protect public health with an adequate margin of
safety. In so doing, she looks to identify the study with the lowest
long-term mean concentration within the full set of key studies to help
inform her decision of the appropriate standard level which will
provide protection for the broad array of health outcomes observed in
all of the studies, including effects observed in at-risk populations.
---------------------------------------------------------------------------

\108\ See American Farm Bureau Federation v. EPA, 559 F. 3d 512,
525-26 (D.C. Cir. 2009).
---------------------------------------------------------------------------

Further, consistent with the general approach summarized in section
III.E.4.a above and supported by CASAC as discussed in section
III.E.4.b.ii above, the Administrator recognizes that it is appropriate
to consider a level for an annual standard that is not just at but
rather is somewhat below the long-term mean PM2.5
concentrations reported in each of the key long- and short-term
exposure studies. In so doing, she focuses especially on multi-city
studies that evaluated health endpoints for which the associations are
causal or likely causal (i.e., mortality and cardiovascular and
respiratory effects associated with both long- and short-term
PM2.5 exposures). As discussed above, the importance of
considering a level somewhat below the lowest long-term mean
concentrations in this set of key studies is to establish a standard
that would be protective against the observed effects in all of the
studies, and that takes into account the relative degree of confidence
in the magnitude and significance of observed associations across the
air quality distributions in these studies.
The Administrator recognizes that there is no clear way to identify
how much below the long-term mean concentrations of key studies to set
a standard that would provide requisite protection with an adequate
margin of safety. She therefore must use her judgment to weigh the
available scientific and technical information, and associated
uncertainties, to reach a final decision on the appropriate standard
level. In considering the information in Figures 1-4 for effects
classified as having evidence of a causal or likely causal relationship
with long- or short-term PM2.5 exposures, she observes a
cluster of short-term exposure studies with long-term mean
concentrations within a range of 13.4 [mu]g/m\3\ down to 12.8 [mu]g/
m\3\ (Dominici et al., 2006a; Burnett and Goldberg, 2003; Zanobetti and
Schwartz, 2009; Bell et al., 2008; Burnett et al., 2004). She also
observes a cluster of long-term exposure studies with long-term mean
concentrations within a range of 14.5 [mu]g/m\3\ to 13.6 [mu]g/m\3\
(Dockery et al., 1996; Lipfert et al., 2006a; Zeger et al., 2008;
McConnell et al., 2003; Goss et al., 2004; Eftim et al., 2008). For the
reasons discussed in response to public comments in section III.E.4.c
above, the Administrator is less influenced by the long-term mean
PM2.5 concentrations from the Miller et al. (2007) and
Krewski et al. (2009) studies with reported long-term mean
PM2.5 concentrations of 12.9 and 14.0 [mu]g/m\3\,
respectively. In each case, the most relevant exposure periods would
likely have had higher mean PM2.5 concentrations than those
reported in the studies.\109\ Thus, the Administrator considers the
long-term mean PM2.5 concentrations from these two studies
to be a highly uncertain basis for informing her selection of the
annual standard level.\110\
---------------------------------------------------------------------------

\109\ In the case of Miller et al. (2007), the mean
concentration is based on a single year of air quality data which
post-dated by two years the period for which the health events data
were collected. In the case of Krewski et al. (2009), the air
quality data were based on the last two years of the 18-year period
for which the health event data were collected.
\110\ Nonetheless, as noted above, the EPA notes that the
Krewski et al. (2009) and Miller et al. (2007) studies provide
strong evidence of mortality and cardiovascular-related effects
associated with long-term PM2.5 exposures to inform
causality determinations reached in the Integrated Science
Assessment (U.S. EPA, 2009a, sections 7.2.11 and 7.6).
---------------------------------------------------------------------------

To help guide her judgment of the appropriate level below the long-
term mean concentrations in the epidemiological studies at which to set
the standard, the Administrator considered additional information from
epidemiological studies concerning the broader distribution of
PM2.5 concentrations which correspond to the health events
observed in these studies (e.g., deaths, hospitalizations). The
Administrator observes that the development and use of this information
in considering standard levels is consistent with CASAC's advice, as
discussed in section III.E.4.b.ii above, to focus on understanding the
concentrations that were most influential in generating the health
effect estimates in individual studies (Samet, 2010d, p. 2).
In considering this additional population-level information, the
Administrator recognizes that, in general, the confidence in the
magnitude and significance of an association identified in a study is
strongest at and around the long-term mean concentration for the air
quality distribution, as this represents the part of the distribution
in which the data in any given study are generally most concentrated.
She also recognizes that the degree of confidence decreases as one
moves towards the lower part of the distribution. Consistent with the
approach used in the Policy Assessment, the Administrator believes that
the range from approximately the 25th to 10th percentiles is a
reasonable range for providing a general frame of reference as to the
part of the distribution in which her confidence in the associations
observed in epidemiological studies is appreciably lower. However, as
noted above, it is important to emphasize that there is no clear
dividing line or single percentile within a given distribution provided
by the scientific evidence that is most appropriate or `correct' to use
to characterize where the degree of confidence in the associations
warrants setting the annual standard level. The decision of the
appropriate standard level below the long-term mean concentrations of
the key studies, which in conjunction with the other elements of the
standard would protect public health with an adequate margin of safety,
is largely a public health policy judgment, taking into account all of
the evidence and its related uncertainties.
As discussed in section III.E.4.b, the Administrator takes note of
additional population-level data that were made available to the EPA by
study authors.\111\ In considering this information, the Administrator
particularly focuses on the analysis of the distributions of the health
event data for each area within these studies and the corresponding air
quality data for the two short-term exposure studies (Zanobetti and
Schwartz, 2009; Bell et al., 2008). These short-term exposure studies
evaluate the relationship between daily changes (one or more days) in
PM2.5 concentrations and daily changes in health events
(e.g., deaths, hospitalizations), such that the air quality
concentrations that comprise the most relevant exposure periods in
these

[[Page 3160]]

studies are contemporaneous with the health event data. In addition,
these studies considered more recent air quality data, representing
generally lower PM2.5 concentrations, in a large number of
study areas across the U.S. Thus, such studies provide the most useful
evidence for an analysis evaluating the distribution of health event
data and the corresponding long-term mean PM2.5
concentrations across the areas included in each multi-city study.
---------------------------------------------------------------------------

\111\ As summarized in section III.E.4.a, population-level data
were provided to the EPA for four studies. These four studies
represent some of the strongest evidence showing associations
between health effects and PM2.5 within the overall body
of scientific evidence and include three studies (Krewski et al.,
2009; Bell et al., 2008; and Zanobetti and Schwartz, 2009) that were
used as the basis for concentration-response functions in the
quantitative risk assessment (U.S. EPA, 2010a, section 3.3.3). The
Administrator recognizes that the additional population-level data
available for these four multi-city studies represents a more
limited data set compared to the set of long-term mean
PM2.5 concentrations which were available in the
published literature for all studies considered in the Integrated
Science Assessment.
---------------------------------------------------------------------------

The Administrator also considered the additional population-level
data that were made available to EPA for two long-term exposure studies
(Krewski et al., 2009; Miller et al., 2007). She recognizes that in
long-term exposure studies investigators follow a specific group of
study participants (i.e., cohort) over time and across urban study
areas, and evaluate how PM2.5 concentrations averaged over a
period of years are associated with specific health endpoints (e.g.,
deaths) across cities. As discussed in response to public comments in
section III.E.4.c, disentangling the effects observed in long-term
exposure studies associated with more recent air quality measurements
from effects that may have been associated with earlier, and most
likely higher, PM2.5 exposures introduces some uncertainty
with regard to understanding the appropriate exposure window associated
with the observed effects. This is in contrast to the short-term
exposure studies where the relevant exposure period is contemporaneous
to the period for which the health data were collected. In light of
these considerations, as noted above, the Administrator considers the
analysis of air quality concentrations that correspond to the
distribution of population-level data in these two studies to be a
highly uncertain basis for informing her selection of the annual
standard level.
Based on the above considerations, the Administrator views the
additional population-level data for the two short-term exposure
studies as appropriate to help inform her judgment of how much below
the long-term mean concentrations to set the level of the annual
standard. The Administrator notes that the long-term mean
PM2.5 concentrations corresponding with study areas
contributing to the 25th percentiles of the distribution of deaths and
cardiovascular-related hospitalizations in these two short-term
exposure studies were 12.5 [micro]g/m\3\ and 11.5 [micro]g/m\3\,
respectively, for Zanobetti and Schwartz (2009) and for Bell et al.
(2008), with the 10th percentiles being lower by approximately 2
[micro]g/m\3\ in each study.
The Administrator recognizes, as summarized in section III.B above
and discussed more fully in section III.B.2 of the proposal, that
important uncertainties remain in the evidence and information
considered in this review of the primary fine particle standards. These
uncertainties are generally related to understanding the relative
toxicity of the different components in the fine particle mixture, the
role of PM2.5 in the complex ambient mixture, exposure
measurement errors, and the nature and magnitude of estimated risks
related to increasingly lower ambient PM2.5 concentrations.
Furthermore, the Administrator notes that epidemiological studies have
reported heterogeneity in responses both within and between cities and
geographic regions across the U.S. She recognizes that this
heterogeneity may be attributed, in part, to differences in fine
particle composition in different regions and cities.\112\
---------------------------------------------------------------------------

\112\ Nonetheless, as explained in section III.E.1, the
currently available evidence is not sufficient to support replacing
or supplementing the PM2.5 indicator with any other
indicator defined in terms of a specific fine particle component or
group of components associated with any source categories of fine
particles. Furthermore, the evidence is not sufficient to support
eliminating any component or group of components associated with any
source categories of fine particles from the mix of fine particles
included in the PM2.5 indicator.
---------------------------------------------------------------------------

With regard to evidence for reproductive and developmental effects
identified as being suggestive of a causal relationship with long-term
PM2.5 exposures, the Administrator recognizes that there are
a number of limitations associated with this body of evidence
including: the limited number of studies evaluating such effects;
uncertainties related to identifying the relevant exposure time periods
of concern; and limited toxicological evidence providing little
information on the mode of action(s) or biological plausibility for an
association between long-term PM2.5 exposures and adverse
birth outcomes. Nonetheless, the Administrator believes that this more
limited body of evidence provides some support for considering that
serious effects may be occurring in a susceptible population at
concentrations lower than those associated with effects classified as
having a causal or likely causal relationship with long-term
PM2.5 exposures (i.e., mortality, cardiovascular, and
respiratory effects).
Overall, the Administrator believes that the available evidence
interpreted in light of the remaining uncertainties, as summarized
above and discussed more fully in the Integrated Science Assessment and
the Policy Assessment, provides increased confidence relative to
information available in the last review and provides a strong basis
for informing her final decisions in the current review. The
Administrator is mindful that considering what standards are requisite
to protect public health with an adequate margin of safety requires
public health policy judgments that neither overstate nor understate
the strength and limitations of the evidence or the appropriate
inferences to be drawn from the evidence. In considering how to
translate the available information into appropriate standard levels,
the Administrator weighs the available scientific information and
associated uncertainties and limitations. For the purpose of
determining what annual standard level is appropriate the Administrator
recognizes that there is no single factor or criterion that comprises
the ``correct'' approach to weighing the various types of available
evidence and information.
In considering this information, the Administrator notes the advice
of CASAC that ``there are significant public health consequences at the
current levels of the standards that justify consideration of lowering
the PM2.5 NAAQS further'' (Samet, 2010c, p. 12). In
addition, she recognizes that CASAC concluded, ``although there is
increasing uncertainty at lower levels, there is no evidence of a
threshold (i.e., a level below which there is no risk for adverse
effects)'' (Samet, 2010d, p.ii) and that the final decisions on
standard levels must reflect a judgment of the available scientific
information with respect to her interpretation of the CAA's requirement
to set primary standards that provide requisite protection to public
health with an adequate margin of safety (Samet, 2010d, p. 4). The
Administrator recognizes CASAC's advice that the currently available
scientific information provided support for considering an annual
standard level within a range of 13 to 11 [mu]g/m\3\ and a 24-hour
standard level within a range of 35 to 30 [mu]g/m\3\. In considering
how the annual and 24-hour standards work together to provide
appropriate public health protection, the Administrator observes that
CASAC did not express support for any specific levels or combinations
of standards within these ranges. She also notes that CASAC encouraged
the EPA staff to consider additional data from epidemiological studies
to help quantify the characterization of the PM2.5
concentrations that were most influential in generating the health

[[Page 3161]]

effect estimates in these studies (Samet, 2010d, p. 2).
In response to CASAC's advice, the Administrator recognizes that
the EPA staff acquired additional data from authors of key
epidemiological studies and analyzed these data to characterize the
distribution of PM2.5 concentrations in relation to health
events data to better understand the degree of confidence in the
associations observed in the studies as discussed above. The
Administrator recognizes that the final Policy Assessment included
consideration of these additional analyses in reaching final staff
conclusions with regard to the broadest range of alternative standard
levels supported by the science. She takes note that the final Policy
Assessment concluded that while alternative standard levels within the
range of 13 to 11 [mu]g/m\3\ were appropriate to consider, the evidence
most strongly supported consideration of an annual standard level in
the range of 12 to 11 [mu]g/m\3\. The Administrator is aware that, in
transmitting the final Policy Assessment to CASAC, the Agency notified
CASAC that the final staff conclusions reflected consideration of
CASAC's advice and that those staff conclusions were based, in part, on
the specific distributional analysis that CASAC had urged the EPA to
conduct (Wegman, 2011). Thus, CASAC had an opportunity to comment on
the final Policy Assessment, but chose not to provide any additional
comments or advice after receiving it.
In selecting the annual standard level, the Administrator has
considered many factors including the nature and severity of the health
effects involved, the strength of the overall body of scientific
evidence as considered in reaching causality determinations, the size
of the at-risk populations, and the estimated public health impacts.
She has also considered the kind and degree of the uncertainties that
remain in the available scientific information. She recognizes that the
association between PM2.5 and serious health effects is well
established, including at concentrations below those allowed by the
current standard. Further, she recognizes the CAA requirement that
requires primary standards to provide an adequate margin of safety was
intended to address uncertainties associated with inconclusive
scientific and technical information as well as to provide a reasonable
degree of protection against hazards that research has not yet
identified. In considering the currently available evidence, as
summarized and discussed more broadly above, the information on risk,
CASAC advice, the conclusions of the Policy Assessment, and public
comments on the proposal, the Administrator strongly believes that a
lower annual standard level is needed to protect public health with an
adequate margin of safety.
In reaching her final decision on the appropriate annual standard
level to set, the Administrator is mindful that the CAA does not
require that primary standards be set at a zero-risk level, but rather
at a level that reduces risk sufficiently so as to protect public
health, including the health of at-risk populations, with an adequate
margin of safety. On balance, the Administrator concludes that an
annual standard level of 12 [mu]g/m\3\ would be requisite to protect
the public health with an adequate margin of safety from effects
associated with long- and short-term PM2.5 exposures, while
still recognizing that uncertainties remain in the scientific
information.
In the Administrator's judgment, an annual standard of 12 [mu]g/
m\3\ appropriately reflects placing greatest weight on evidence of
effects for which the Integrated Science Assessment determined there is
a causal or likely causal relationship with long- and short-term
PM2.5 exposures. An annual standard level of 12 [mu]g/m\3\
is below the long-term mean PM2.5 concentrations reported in
each of the key multi-city, long- and short-term exposures studies
providing evidence of an array of serious health effects (e.g.,
premature mortality, increased hospitalization for cardiovascular and
respiratory effects). As noted above, the importance of considering a
level somewhat below the lowest long-term mean concentration in the
full set of studies considered is to set a standard that would provide
appropriate protection against the observed effects in all such
studies.
In reaching her decision, the Administrator has taken into account
that at and around the mean PM2.5 concentration in any given
study represents a part of the air quality distribution in which the
health event data in that study are generally most concentrated.
Furthermore, in identifying an appropriate annual standard level below
the long-term mean concentrations, she recognizes that there is no
evidence to support the existence of any discernible threshold, and,
therefore, she has a high degree of confidence that the observed
effects are associated with concentrations not just at but extending
somewhat below the long-term mean concentration. To further inform her
judgment in setting the annual standard level so as to protect public
health with an adequate margin of safety, the Administrator has placed
weight on additional population-level information available from a
subset of these epidemiological studies, consistent with CASAC advice.
In particular, she has drawn from two short-term exposure studies,
which provide the most relevant information for evaluating the
distribution of health events and corresponding long-term
PM2.5 concentrations. As explained above, this helps inform
her judgment as to the degree of confidence in the observed
associations in the epidemiological studies. In this regard, the
Administrator generally judges the region around the 25th percentile as
a reasonable part of the distribution to help guide her decision on the
appropriate standard level. Since this evidence comes primarily from
two studies, a relatively modest data set, the Administrator deems it
reasonable not to draw further inferences from air quality and health
event data in the lower part of the distribution for the purpose of
setting a standard level. The Administrator notes that the long-term
mean PM2.5 concentrations around the 25th percentile of the
distributions of deaths and cardiovascular-related hospitalizations
were approximately around 12 [mu]g/m\3\ in these two studies. The
Administrator views this information as helpful in guiding her
determination as to where her confidence in the magnitude and
significance of the associations is reduced to such a degree that a
standard set at a lower level would not be warranted to provide
requisite protection that is neither more nor less than needed to
provide an adequate margin of safety.
The Administrator also recognizes that a level of 12 [mu]g/m\3\
places some weight on studies which provide evidence of reproductive
and developmental effects (e.g., infant mortality, low birth weight).
These studies were identified in the Integrated Science Assessment as
having evidence suggestive of a causal relationship with long-term
PM2.5 concentrations. A level of 12 [mu]g/m\3\ is
approximately the same level as the lowest long-term mean concentration
reported in such studies (Figures 2 and 4; 11.9 [mu]g/m\3\ for Bell et
al., 2007).\113\ While the Administrator

[[Page 3162]]

acknowledges that this evidence is limited, she believes it is
appropriate to place some weight on these studies in order to set a
standard that provides protection with an adequate margin of safety,
including providing protection for at-risk populations, as required by
the CAA. Due to the limited nature of this evidence, she has determined
it is not necessary to set a standard below the lowest long-term mean
concentration in these studies.
---------------------------------------------------------------------------

\113\ With respect to cancer, mutagenic, and genotoxic effects,
the Administrator observes that the PM2.5 concentrations
reported in studies evaluating these effects generally included
ambient concentrations that are equal to or greater than ambient
concentrations observed in studies that reported mortality and
cardiovascular and respiratory effects (U.S. EPA, 2009a, section
7.5). Therefore, the Administrator concludes that in selecting
alternative standard levels that provide protection from mortality
and cardiovascular and respiratory effects, it is reasonable to
anticipate that protection will also be provided for carcinogenic
effects.
---------------------------------------------------------------------------

In reflecting on extensive public comments received on the proposal
as discussed in section III.E.4.c above, the Administrator recognizes
that some commenters have offered different evaluations of the evidence
and other information available in this review and would make different
judgments about the weight to place on the relative strengths and
limitations of the scientific information and about how such
information could be used in making public health policy decisions on
the annual standard level. One group of such commenters who supported a
higher annual standard level (e.g., above 13 [mu]g/m\3\) would place
greater weight on the remaining uncertainties in the evidence as a
basis for supporting a higher standard level than the Administrator
judges to be appropriate. Such an approach is based on these
commenters' judgment that the uncertainties remaining in the evidence
are too great to warrant setting an annual standard below the current
level. The Administrator does not agree.
As an initial matter, an annual standard level of 13 [mu]g/m\3\ or
higher would be above the long-term mean concentrations reported in two
well-conducted, multi-city short-term exposure studies reporting
positive and statistically significant associations of serious effects
(Burnett et al., 2004 and Bell et al., 2008). These important studies
are fully consistent with the pattern of evidence presented by the
large body of evidence in this review. As the Administrator recognized
in the proposal, and as advised by CASAC, the appropriate focus for
selecting the level of the annual PM2.5 standard is on
concentrations somewhat below the lowest long-term mean concentrations
from the set of key studies of both long-term and short-term
PM2.5 exposures considered by the EPA (i.e., as shown in
Figure 4). Thus, a standard level set at 13 [mu]g/m\3\ or higher would
clearly not provide protection for the effects observed in the full set
epidemiological studies and, therefore, this standard level could not
be judged to be requisite with an adequate margin of safety.\114\
---------------------------------------------------------------------------

\114\ The Administrator is mindful that, in reviewing the 2006
final PM NAAQS decisions, the D.C. Circuit Court of Appeals
concluded that the EPA failed to adequately explain why that annual
standard provided requisite protection from effects associated with
both long- and short-term exposures or from morbidity effects in
children and other at-risk populations when long-term means of
important short-term studies were below the level the Administrator
selected for the annual standard. See American Farm Bureau v. EPA.
559 F. 3d 512, 524-26. There is no reasonable basis to discount
these two studies for purposes of setting the level of the annual
standard.
---------------------------------------------------------------------------

In addition, as noted above, in recognizing that there is no
evidence to support the existence of a discernible threshold below
which an effect would not occur, the Administrator is mindful that
effects occur around and below the long-term mean concentrations
reported in both the short-term and long-term the epidemiological
studies. A standard level of 13 [mu]g/m\3\ or higher would not
appropriately take into account evidence from the two well-conducted,
multi-city, short-term exposure studies reporting serious effects with
long-term mean concentrations below 13 [mu]g/m\3\ noted above (Burnett
et al, 2004; Bell et al., 2008). Such a standard level would also not
appropriately take into account additional population-level data from a
limited number of epidemiological studies. This approach would ignore
CASAC's advice to consider such information in order to better
understand the concentrations over which there is a high degree of
confidence regarding the magnitude and significance of the associations
observed in individual epidemiological studies and where there is
appreciably less confidence.
Furthermore, a standard level of 13 [mu]g/m\3\ or higher would not
appropriately take into account the more limited evidence of effects in
some at-risk populations (e.g., low birth weight). In the
Administrator's view, a standard set at this level would not provide
protection with an adequate margin of safety, including providing
protection for at-risk populations. The Administrator is mindful that
the CAA requirement that primary standards provide an adequate margin
of safety, discussed in section II.A above, was intended to address
uncertainties associated with inconclusive scientific and technical
information available at the time of standard setting as well as to
provide a reasonable degree of protection against hazards that research
has not yet identified.
In light of the entire body of evidence as discussed above, the
Administrator judges that an annual standard level set above 12 [mu]g/
m\3\ would not be sufficient to protect public health with an adequate
margin of safety from the serious health effects associated with long-
and short-term exposure to PM2.5.
The Administrator also recognizes that a second group of commenters
supported a lower annual standard level (e.g., no higher than 11 [mu]g/
m\3\). Such a standard level would reflect placing essentially as much
weight on the relatively more limited data providing evidence
suggestive of a causal relationship for effects observed in some at-
risk populations (e.g., low birth weight) as on more certain evidence
of effects classified as having a causal or likely causal relationship
with PM2.5 exposures. In the Administrator's view, while it
is important to place some weight on such suggestive evidence, it would
not be appropriate to place as much weight on it as the commenters
would do.
An annual standard level of 11 [mu]g/m\3\ would also reflect these
commenters' judgment that it is appropriate to focus on a lower part of
the distributions of health event data from the small number of
epidemiological studies for which this information was made available
than the Administrator believes is warranted. In the Administrator's
view, using this type of information to set a standard level of 11
[mu]g/m\3\ or below would assume too high a degree of confidence in the
magnitude and significance of the associations observed in the lower
part of the distributions of health events observed in these studies.
Given the uncertainties in the evidence and the limited set of studies
for which the EPA has information on the distribution of health event
data and corresponding air quality data, the Administrator believes it
is not appropriate to focus on the lower part of the distributions of
health events data.
On balance, the Administrator finds that the available evidence
interpreted in light of the remaining uncertainties does not justify a
standard level set below 12 [mu]g/m\3\ as necessary to protect public
health with an adequate margin of safety.
After carefully considering the above considerations and the public
comments summarized in section III.E.4.c above, the Administrator has
decided to set the level of the primary annual PM2.5
standard at 12 [mu]g/m\3\. In her judgment, a standard set at this
level provides the requisite degree of public health protection,
including the health of at-risk populations, with an adequate margin of
safety and is neither more nor less stringent than necessary for this
purpose.

[[Page 3163]]

As discussed above, the Administrator concludes that an approach
that focuses on setting a generally controlling annual standard is the
most effective and efficient way to reduce total population risk
associated with both long- and short-term PM2.5 exposures.
Such an approach would result in more uniform protection across the
U.S. than the alternative of setting the levels of the 24-hour and
annual standard such that the 24-hour standard would generally be the
controlling standard in areas across the country (see section III.A.3).
The Administrator recognizes that potential air quality changes
associated with meeting an annual standard level of 12.[mu]g/m\3\ will
result in lowering risks associated with both long- and short-term
PM2.5 exposures by lowering the overall air quality
distribution. However, the Administrator recognizes that such an annual
standard alone would not be expected to offer sufficient protection
with an adequate margin of safety against the effects of short-term
PM2.5 exposures in all parts of the country. As a result, in
conjunction with an annual standard level of 12 [mu]g/m\3\, the
Administrator concludes that it is appropriate to continue to provide
supplemental protection by means of a 24-hour standard set at the
appropriate level, particularly for areas with high peak-to-mean ratios
possibly associated with strong local or seasonal sources and for areas
with PM2.5-related effects that may be associated with
shorter-than-daily exposure periods.
In selecting the level of a 24-hour standard meant to provide such
supplemental protection, the Administrator relies upon evidence and air
quality information from key short-term exposure studies. In
considering these studies, the Administrator notes that to the extent
air quality distributions in the study areas considered are reduced to
meet the current 24-hour standard (at a level of 35 [mu]g/m\3\) or to
meet the revised annual standard discussed above (at a level of 12
[mu]g/m\3\), additional protection would be anticipated against the
effects observed in these studies. In light of this, when selecting the
appropriate level for the 24-hour standard, the Administrator considers
both the 98th percentiles of 24-hour PM2.5 concentrations
and the long-term mean PM2.5 concentrations in the locations
of the short-term exposure studies. She notes that such consideration
of both short- and long-term PM2.5 concentrations can inform
her decision on the extent to which a given 24-hour standard, in
combination with the revised annual standard established in this rule,
would provide protection against the health effects reported in short-
term studies.
As discussed in section III.E.4.a above, the Administrator
concludes that multi-city short-term exposure studies provide the
strongest data set for informing her decisions on appropriate 24-hour
standard levels. With regard to the limited number of single-city
studies that reported positive and statistically significant
associations for a range of health endpoints related to short-term
PM2.5 concentrations in areas that would likely have met the
current suite of PM2.5 standards, the Administrator
recognizes that many of these studies had significant limitations
(e.g., limited statistical power, limited exposure data) or equivocal
results (mixed results within the same study area) that make them
unsuitable to form the basis for setting the level of a 24-hour
standard.
With regard to multi-city studies that evaluated effects associated
with short-term PM2.5 exposures, the Administrator observes
an overall pattern of positive and statistically significant
associations in studies with 98th percentile 24-hour values averaged
across study areas within the range of 45.8 to 34.2 [mu]g/m\3\ (Burnett
et al., 2004; Zanobetti and Schwartz, 2009; Bell et al., 2008; Dominici
et al., 2006a, Burnett and Goldberg, 2003; Franklin et al., 2008). The
Administrator notes that, to the extent air quality distributions are
reduced to reflect just meeting the current 24-hour standard,
additional protection would be provided for the effects observed in the
three multi-city studies with 98th percentile values greater than 35
[mu]g/m\3\ (Burnett et al., 2004; Burnett and Goldberg, 2003; Franklin
et al., 2008). In the three additional multi-city studies with 98th
percentile values below 35 [mu]g/m\3\, specifically 98th percentile
concentrations of 34.2, 34.3, and 34.8 [mu]g/m\3\, the Administrator
notes that these studies reported long-term mean PM2.5
concentrations of 12.9, 13.2, and 13.4 [mu]g/m\3\, respectively (Bell
et al., 2008; Zanobetti and Schwartz, 2009; Dominici et al., 2006a). In
revising the level of the annual standard to 12 [mu]g/m\3\, as
discussed above, the Administrator recognizes that additional
protection would be provided for the short-term effects observed in
these multi-city studies such that revision to the 24-hour standard
would not be warranted. That is, by lowering the level of the annual
standard to 12 [mu]g/m\3\, the 98th percentile of the distribution
would be lowered as well such that additional protection from effects
associated with short-term exposures would be afforded. Therefore, the
epidemiological evidence supports a conclusion that it is appropriate
to retain the level of the 24-hour standard at 35 [mu]g/m\3\, in
conjunction with a revised annual standard level of 12 [mu]g/m\3\.
In addition to considering the epidemiological evidence, the
Administrator also has taken into account air quality information based
on county-level 24-hour and annual design values to understand the
implications of revising the annual standard level from 15 to 12 [mu]g/
m\3\ in conjunction with retaining the 24-hour standard level at 35
[mu]g/m\3\. She has considered this information to evaluate the public
health protection provided by the two standards in combination and to
evaluate the most appropriate means of developing a suite of standards
providing requisite public health protection with an adequate margin of
safety.
In considering the air quality information, the Administrator
observes that a suite of PM2.5 standards that includes an
annual standard level of 12 [mu]g/m\3\ and a 24-hour standard level of
35 [mu]g/m\3\ would result in the annual standard as the generally
controlling standard in most regions across the country, except for
certain areas in the Northwest, where the annual mean PM2.5
concentrations have historically been low but where relatively high 24-
hour concentrations occur, often related to seasonal wood smoke
emissions (U.S. EPA, 2011a, pp. 2-89 to 2-91, Figure 2-10). In fact,
these are the type of areas for which the supplemental protection
afforded by the 24-hour standard is intended, such that the two
standards together provide the requisite degree of protection. The
Administrator concludes the current 24-hour standard at a level of 35
[mu]g/m\3\, in conjunction with a revised annual standard level of 12
[mu]g/m\3\, will provide appropriate protection from effects observed
in studies in such areas in which the long-term mean concentrations
were below 12 [mu]g/m\3\ and the 98th percentile 24-hour concentrations
were above 35 [mu]g/m\3\ (e.g., areas in the Northwest U.S.).
After carefully taking the public comments and above considerations
into account, the Administrator has decided to retain the current level
of the primary PM2.5 24-hour standard at 35 [mu]g/m\3\ in
conjunction with revising the annual standard level from 15.0 [mu]g/
m\3\ to 12.0 [mu]g/m\3\.\115\ In the Administrator's

[[Page 3164]]

judgment, this suite of primary PM2.5 standards and the
rationale supporting these levels appropriately reflects consideration
of the strength of the available evidence and other information and its
associated uncertainties as well as the advice of CASAC and
consideration of public comments. In the Administrator's judgment, this
suite of primary PM2.5 standards is sufficient but not more
protective than necessary to protect the public health, including at-
risk populations, with an adequate margin of safety from effects
associated with long- and short-term exposures to fine particles. This
suite of standards will provide significant protection from serious
health effects including premature mortality and cardiovascular and
respiratory morbidity effects that are causally or likely causally
related to long- and short-term PM2.5 exposures. These
standards will also provide an appropriate degree of protection against
other health effects for which there is more limited evidence of
effects and causality, such as reproductive and developmental effects.
This judgment by the Administrator appropriately considers the
requirement for a standard that is requisite to protect public health
but is neither more nor less stringent than necessary.\116\
---------------------------------------------------------------------------

\115\ As noted in section II.B.1, Table 1 and section III.E.4.a
above, the annual standard level is defined to one decimal place.
Throughout this section, the annual standard levels discussed have
been denoted as integer values (e.g., 12 [mu]g/m\3\) for simplicity.
\116\ The Administrator also judges that this suite of standards
addresses the issues raised by the D.C. Circuit's remand of the 2006
primary annual PM2.5 standard by appropriately revising
that standard.
---------------------------------------------------------------------------

D. Administrator's Final Decisions on Primary PM2.5 Standards

For the reasons discussed above, and taking into account the
information and assessments presented in the Integrated Science
Assessment, Risk Assessment, and Policy Assessment, the advice and
recommendations of CASAC, and public comments to date, the
Administrator revises the current suite of primary PM2.5
standards. Specifically, the Administrator revises: (1) The level of
the primary annual PM2.5 standard to 12.0 [mu]g/m\3\ and (2)
the form of the primary annual PM2.5 standard to one based
on the highest appropriate area-wide monitor in an area, with no option
for spatial averaging. In conjunction with revising the primary annual
PM2.5 standard to provide protection from effects associated
with long- and short-term PM2.5 exposures, the Administrator
retains the level of 35 [mu]g/m\3\ and the 98th percentile form of the
primary 24-hour PM2.5 standard to continue to provide
supplemental protection for areas with high peak PM2.5
concentrations. The Administrator is not revising the current
PM2.5 indicator or the annual and 24-hour averaging times
for the primary PM2.5 standards. The Administrator concludes
that this suite of standards would be requisite to protect public
health with an adequate margin of safety against health effects
potentially associated with long- and short-term PM2.5
exposures.

IV. Rationale for Final Decision on Primary PM10 Standard

This section presents the rationale for the Administrator's final
decision to retain the current 24-hour primary PM10 standard
in order to continue to provide public health protection against short-
term exposures to inhalable particles in the size range of 2.5 to 10
[mu]m (i.e., PM10-2.5 or thoracic coarse particles). These
are particles capable of reaching the most sensitive areas of the lung,
including the trachea, bronchi, and deep lungs. The current standard
uses PM10 as the indicator for thoracic coarse particles,
and thus is referred to as a PM10 standard.\117\
---------------------------------------------------------------------------

\117\ Throughout this section of the preamble, we are using the
terms ``thoracic coarse particles'', ``inhalable coarse particles'',
and ``PM10-2.5'' synonymously.
---------------------------------------------------------------------------

As discussed more fully in the proposal and below, this rationale
is based on a thorough review of the latest scientific evidence,
published through mid-2009 and assessed in the Integrated Science
Assessment (U.S. EPA, 2009a), evaluating human health effects
associated with long- and short-term exposures to thoracic coarse
particles. The Administrator's final decision also takes into account:
(1) The EPA staff analyses of air quality information and health
evidence and staff conclusions regarding the current and potential
alternative standards, as presented in the Policy Assessment for the PM
NAAQS (U.S. EPA, 2011a); (2) CASAC advice and recommendations, as
reflected in discussions at public meetings of drafts of the Integrated
Science Assessment and Policy Assessment, and in CASAC's letters to the
Administrator; (3) the multiple rounds of public comments received
during the development of the Integrated Science Assessment and Policy
Assessment, both in connection with CASAC meetings and separately; and
(4) public comments (including testimony at the public hearings)
received on the proposal.
In presenting the rationale for the final decision to retain the
current primary PM10 standard, this section discusses the
EPA's past reviews of the PM NAAQS and the general approach taken to
review the current standard (section IV.A), the health effects
associated with exposures to ambient PM10-2.5 (section
IV.B), the consideration of the current and potential alternative
standards in the Policy Assessment (section IV.C), CASAC
recommendations regarding the current and potential alternative
standards (section IV.D), the Administrator's proposed decision to
retain the current primary PM10 standard (section IV.E),
public comments received in response to the Administrator's proposed
decision (section IV.F), and the Administrator's final decision to
retain the current primary PM10 standard (section IV.G).

A. Background

The following sections discuss previous reviews of the PM NAAQS
(section IV.A.1), the litigation of the EPA's 2006 decision on the
PM10 standards (section IV.A.2), and the general approach
taken to review the primary PM10 standard in the current
review (section IV.A.3).
1. Previous Reviews of the PM NAAQS
a. Reviews Completed in 1987 and 1997
The PM NAAQS have always included some type of a primary standard
to protect against effects associated with exposures to thoracic coarse
particles. In 1987, when the EPA first revised the PM NAAQS, the EPA
changed the indicator for PM from TSP to focus on inhalable particles,
those which can penetrate into the trachea, bronchi, and deep lungs (52
FR 24634, July 1, 1987). In that review, the EPA changed the PM
indicator to PM10 based on evidence that the risk of adverse
health effects associated with particles with a nominal mean
aerodynamic diameter less than or equal to 10 [mu]m was significantly
greater than risks associated with larger particles (52 FR 24639, July
1, 1987).
In the 1997 review, in conjunction with establishing new fine
particle (i.e., PM2.5) standards (discussed above in
sections II.B.1 and III.A.1), the EPA concluded that continued
protection was warranted against potential effects associated with
thoracic coarse particles in the size range of 2.5 to 10 [mu]m. This
conclusion was based on particle dosimetry, toxicological information,
and on limited epidemiological evidence from studies that measured
PM10 in areas where the coarse fraction was likely to
dominate PM10 mass (62 FR 38677, July 18, 1997). The EPA
concluded there that a PM10 standard could provide requisite
protection against effects associated with particles

[[Page 3165]]

in the size range of 2.5 to 10 [mu]m.\118\ Although the EPA considered
a more narrowly defined indicator for thoracic coarse particles in that
review (i.e., PM10-2.5), the EPA concluded that it was more
appropriate, based on existing evidence, to continue to use
PM10 as the indicator. This decision was based, in part, on
the recognition that the only studies of clear quantitative relevance
to health effects most likely associated with thoracic coarse particles
used PM10. These were two studies conducted in areas where
the coarse fraction was the dominant fraction of PM10, and
which substantially exceeded the 24-hour PM10 standard (62
FR 38679). In addition, there were only very limited ambient air
quality data then available specifically for PM10-2.5, in
contrast to the extensive monitoring network already in place for
PM10. Therefore, the EPA considered it more administratively
feasible to use PM10 as an indicator. The EPA also stated
that the PM10 standards would work in conjunction with the
PM2.5 standards by regulating the portion of particulate
pollution not regulated by the then newly adopted PM2.5
standards.
---------------------------------------------------------------------------

\118\ With regard to the 24-hour PM10 standard, the
EPA retained the indicator, averaging time, and level (150 [mu]g/
m\3\), but revised the form (i.e., from one-expected-exceedance to
the 99th percentile).
---------------------------------------------------------------------------

In May 1998, a three-judge panel of the U.S. Court of Appeals for
the District of Columbia Circuit found ``ample support'' for the EPA's
decision to regulate coarse particle pollution, but vacated the 1997
PM10 standards, concluding that the EPA had failed to
adequately explain its choice of PM10 as the indicator for
thoracic coarse particles American Trucking Associations v. EPA, 175 F.
3d 1027, 1054-56 (D.C. Cir. 1999). In particular, the court held that
the EPA had not explained the use of an indicator under which the
allowable level of coarse particles varied according to the amount of
PM2.5 present, and which, moreover, potentially double
regulated PM2.5. The court also rejected considerations of
administrative feasibility as justification for use of PM10
as the indicator for thoracic coarse PM, since NAAQS (and their
elements) are to be based exclusively on health and welfare
considerations. Id. at 1054. Pursuant to the court's decision, the EPA
removed the vacated 1997 PM10 standards from the CFR (69 FR
45592, July 30, 2004) and deleted the regulatory provision (at 40 CFR
50.6(d)) that controlled the transition from the pre-existing 1987
PM10 standards to the 1997 PM10 standards (65 FR
80776, December 22, 2000). The pre-existing 1987 PM10
standards thus remained in place. Id. at 80777.
b. Review Completed in 2006
In the review of the PM NAAQS that concluded in 2006, the EPA
considered the growing, but still limited, body of evidence supporting
associations between health effects and thoracic coarse particles
measured as PM10-2.5.\119\ The new studies available in the
2006 review included epidemiological studies that reported associations
with health effects using direct measurements of PM10-2.5,
as well as dosimetric and toxicological studies. In considering this
growing body of PM10-2.5 evidence, as well as evidence from
studies that measured PM10 in locations where the majority
of PM10 was in the PM10-2.5 fraction (U.S. EPA,
2005, section 5.4.1), staff concluded that the level of protection
afforded by the existing 1987 PM10 standard remained
appropriate (U.S. EPA, 2005, p. 5-67) but recommended that the
indicator for the standard be revised. Specifically, staff recommended
replacing the PM10 indicator with an indicator of urban
thoracic coarse particles in the size range of 10-2.5 [mu]m (U.S. EPA,
2005, pp. 5-70 to 5-71). The agency proposed to retain a standard for a
subset of thoracic coarse particles, proposing a qualified
PM10-2.5 indicator to focus on the mix of thoracic coarse
particles generally present in urban environments. More specifically,
the proposed revised thoracic coarse particle standard would have
applied only to an ambient mix of PM10-2.5 dominated by
resuspended dust from high-density traffic on paved roads and/or by
industrial and construction sources. The proposed revised standard
would not have applied to any ambient mix of PM10-2.5
dominated by rural windblown dust and soils. In addition, agricultural
sources, mining sources, and other similar sources of crustal material
would not have been subject to control in meeting the standard (71 FR
2667 to 2668, January 17, 2006).
---------------------------------------------------------------------------

\119\ The PM Staff Paper (U.S. EPA, 2005) also presented results
of a quantitative assessment of health risks for
PM10-2.5. However, staff concluded that the nature and
magnitude of the uncertainties and concerns associated with this
risk assessment weighed against its use as a basis for recommending
specific levels for a thoracic coarse particle standard (U.S. EPA,
2005, p. 5-69).
---------------------------------------------------------------------------

The Agency received a large number of comments overwhelmingly and
persuasively opposed to the proposed qualified PM10-2.5
indicator (71 FR 61188 to 61197, October 17, 2006). After careful
consideration of the scientific evidence and the recommendations
contained in the 2005 Staff Paper, the advice and recommendations from
CASAC, and the public comments received regarding the appropriate
indicator for coarse particles, and after extensive evaluation of the
alternatives available to the Agency, the Administrator decided it
would not be appropriate to adopt the proposed qualified
PM10-2.5 indicator, or any qualified indicator. Underlying
this determination was the Administrator's decision that it was
requisite to provide protection from exposure to all thoracic coarse
PM, regardless of its origin. The Administrator thus rejected arguments
that there are no health effects from community-level exposures to
coarse PM in non-urban areas (71 FR 61189). The EPA concluded that
dosimetric, toxicological, occupational and epidemiological evidence
supported retention of a primary standard for short-term exposures that
included all thoracic coarse particles (i.e., particles of both urban
and non-urban origin), consistent with the Act's requirement that
primary NAAQS must be requisite to protect the public health and
provide an adequate margin of safety. At the same time, the Agency
concluded that the standard should target protection toward urban
areas, where the evidence of health effects from exposure to
PM10-2.5 was strongest (71 FR at 61193, 61197). The proposed
indicator was not suitable for that purpose. Not only did it
inappropriately provide no protection at all to many areas, but it
failed to identify many areas where the ambient particle mix was
dominated by coarse particles contaminated with urban/industrial types
of coarse particles for which evidence of health effects was strongest
(71 FR 61193).
The Agency ultimately concluded that the existing indicator,
PM10, was most consistent with the evidence. Although
PM10 includes both coarse and fine PM, the Agency concluded
that it remained an appropriate indicator for thoracic coarse particles
because, as discussed in the PM Staff Paper (U.S. EPA, 2005, p. 2-54,
Figures 2-23 and 2-24), fine particle levels are generally higher in
urban areas and, therefore, a PM10 standard set at a single
unvarying level will generally result in lower allowable concentrations
of thoracic coarse particles in urban areas than in non-urban areas (71
FR 61195-96). The EPA considered this to be an appropriate targeting of
protection given that the strongest evidence for effects associated
with thoracic coarse particles came from epidemiological studies
conducted in urban areas and that elevated fine particle concentrations
in urban areas could result in increased contamination of coarse
fraction particles by PM2.5,

[[Page 3166]]

potentially increasing the toxicity of thoracic coarse particles in
urban areas (id.). Given the evidence that the existing (i.e., 1987)
PM10 standard was established at a level and form which
afforded requisite protection with an adequate margin of safety, the
Agency retained the level and form of the 24-hour PM10
standard.\120\
---------------------------------------------------------------------------

\120\ Thus, the standard is met when a 24-hour average
PM10 concentration of 150 [mu]g/m\3\ is not exceeded more
than one day per year, on average over a three-year period. As noted
above, the 1987 PM10 standard was not adopted solely to
control thoracic coarse particles. However, when reviewing this
standard in the 2006 review, EPA determined that the level and form
of the standard being reviewed (i.e., the 1987 PM10
standard) provided requisite protection with an adequate margin of
safety from short-term exposures to thoracic coarse particles.
---------------------------------------------------------------------------

The Agency also revoked the annual PM10 standard, in
light of the conclusion in the PM Criteria Document (U.S. EPA, 2004, p.
9-79) that the available evidence does not suggest an association with
long-term exposure to PM10-2.5 and the conclusion in the
Staff Paper (U.S. EPA, 2005, p. 5-61) that there is no quantitative
evidence that directly supports retention of an annual standard. This
decision was consistent with CASAC advice and recommendations
(Henderson, 2005a,b).
In the same rulemaking, the EPA also included a new FRM for the
measurement of PM10-2.5 in the ambient air (71 FR 61212 to
61213, October 17, 2006). Although the standard for thoracic coarse
particles does not use a PM10-2.5 indicator, the new FRM for
PM10-2.5 was established to provide a basis for approving
FEMs and to promote the gathering of scientific data to support future
reviews of the PM NAAQS (71 FR 61202/3, October 17, 2006).\121\
---------------------------------------------------------------------------

\121\ As noted below, however, with this rule the EPA is
revoking the requirement for PM10-2.5 speciation at NCore
monitoring sites due to technical issues related to the development
of appropriate monitoring methods (section VIII.B.3.c). The
requirement for PM10-2.5 mass measurements at NCore sites
is being retained.
---------------------------------------------------------------------------

2. Litigation Related to the 2006 Primary PM10 Standards
A number of groups filed suit in response to the final decisions
made in the 2006 review. See American Farm Bureau Federation v. EPA,
559 F. 3d 512 (D.C. Cir. 2009). Among the petitions for review were
challenges from industry groups on the decision to retain the
PM10 indicator and the level of the PM10 standard
and from environmental and public health groups on the decision to
revoke the annual PM10 standard. The court upheld both the
decision to retain the 24-hour PM10 standard and the
decision to revoke the annual standard.
First, the court upheld the EPA's decision for a standard to
encompass all thoracic coarse PM, both of urban and non-urban origin.
The court rejected arguments that the evidence showed there are no
risks from exposure to non-urban coarse PM. The court further found
that the EPA had a reasonable basis not to set separate standards for
urban and non-urban coarse PM, namely the inability to reasonably
define what ambient mixes would be included under either `urban' or
`non-urban;' and the evidence in the record that supported the EPA's
appropriately cautious decision to provide ``some protection from
exposure to thoracic coarse particles * * * in all areas.'' 559 F. 3d
at 532-33. Specifically, the court stated,

Although the evidence of danger from coarse PM is, as EPA
recognizes, ``inconclusive,'' (71 FR 61193, October 17, 2006), the
agency need not wait for conclusive findings before regulating a
pollutant it reasonably believes may pose a significant risk to
public health. The evidence in the record supports the EPA's
cautious decision that ``some protection from exposure to thoracic
coarse particles is warranted in all areas.'' Id. As the court has
consistently reaffirmed, the CAA permits the Administrator to ``err
on the side of caution'' in setting NAAQS. 559 F. 3d at 533.

The court also upheld the EPA's decision to retain the level of the
standard at 150 [mu]g/m\3\ and to use PM10 as the indicator
for thoracic coarse particles. In upholding the level of the standard,
the court referred to the conclusion in the Staff Paper that there is
``little basis for concluding that the degree of protection afforded by
the current PM10 standards in urban areas is greater than
warranted, since potential mortality effects have been associated with
air quality levels not allowed by the current 24-hour standard, but
have not been associated with air quality levels that would generally
meet that standard, and morbidity effects have been associated with air
quality levels that exceeded the current 24-hour standard only a few
times.'' 559 F. 3d at 534. The court also rejected arguments that a
PM10 standard established at an unvarying level will result
in arbitrarily varying levels of protection given that the level of
coarse PM would vary based on the amount of fine PM present. The court
agreed that the variation in allowable coarse PM was in accord with the
strength of the evidence: Typically less coarse PM would be allowed in
urban areas (where levels of fine PM are typically higher), in accord
with the strongest evidence of health effects from coarse particles.
559 F. 3d at 535-36. In addition, such regulation would not
impermissibly double regulate fine particles, since any additional
control of fine particles (beyond that afforded by the primary
PM2.5 standard) would be for a different purpose: To prevent
contamination of coarse particles by fine particles. 559 F. 3d at 535,
536. These same explanations justified the choice of PM10 as
an indicator and provided the reasoned explanation for that choice
lacking in the record for the 1997 standard. 559 F. 3d at 536.
With regard to the challenge from environmental and public health
groups, the court upheld the EPA's decision to revoke the annual
PM10 standard. The court rejected the argument that the EPA
is required by law to have an annual PM10 standard, holding
that section 109(d)(1) of the Act allows the EPA to revoke a standard
no longer warranted by the current scientific understanding. 559 F. 3d
at 538. The court further held that the EPA's decision to revoke the
annual standard was supported by the science:

The EPA reasonably decided that an annual coarse PM standard is
not necessary because, as the Criteria Document and the Staff Paper
make clear, the latest scientific data do not indicate that long-
term exposure to coarse particles poses a health risk. The CASAC
also agreed that an annual coarse PM standard is unnecessary. 559 F.
3d at 538-39.
3. General Approach Used in the Current Review
The approach taken to considering the existing and potential
alternative primary PM10 standards in the current review
builds upon the approaches used in previous PM NAAQS reviews. This
approach is based most fundamentally on using information from
epidemiological studies and air quality analyses to inform the
identification of a range of policy options for consideration by the
Administrator. The Administrator considers the appropriateness of the
current and potential alternative standards, taking into account the
four elements of the NAAQS: Indicator, averaging time, form, and level.
Evidence-based approaches to using information from epidemiological
studies to inform decisions on PM standards are complicated by the
recognition that no population threshold, below which it can be
concluded with confidence that PM-related effects do not occur, can be
discerned from the available evidence (U.S. EPA, 2009a, sections 2.4.3
and 6.5.2.7).\122\ As a result, any approach to

[[Page 3167]]

reaching decisions on what standards are appropriate requires judgments
about how to translate the information available from the
epidemiological studies into a basis for appropriate standards, which
includes consideration of how to weigh the uncertainties in reported
associations across the distributions of PM concentrations in the
studies. The approach taken to informing these decisions in the current
review recognizes that the available health effects evidence reflects a
continuum consisting of ambient levels at which scientists generally
agree that health effects are likely to occur through lower levels at
which the likelihood and magnitude of the response become increasingly
uncertain. Such an approach is consistent with setting standards that
are neither more nor less stringent than necessary, recognizing that a
zero-risk standard is not required by the CAA.
---------------------------------------------------------------------------

\122\ Studies that have characterized the concentration-response
relationships for PM exposures have evaluated PM10, which
includes both coarse and fine particles, and PM2.5 (U.S.
EPA, 2009a, sections 2.4.3 and 6.5.2.7).
---------------------------------------------------------------------------

Because the purpose of the PM10 standard is to protect
against exposures to PM10-2.5, it is most appropriate to
focus on PM10-2.5 health studies when considering the degree
of public health protection provided by the current PM10
standard. Compared to health studies of PM10, studies that
evaluate associations with PM10-2.5 provide clearer evidence
for health effects following exposures to thoracic coarse particles. In
contrast, it is difficult to interpret PM10 studies within
the context of a standard meant to protect against exposures to
PM10-2.5 because PM10 is comprised of both fine
and coarse particles, even in locations with the highest concentrations
of PM10-2.5 (U.S. EPA, 2011a, Figure 3-4). Therefore, the
extent to which PM10 effect estimates reflect associations
with PM10-2.5 versus PM2.5 can be highly
uncertain. In light of this uncertainty, it is preferable to consider
PM10-2.5 studies when such studies are available. Given the
availability in this review of a number of studies that evaluated
associations with PM10-2.5, and given that the Integrated
Science Assessment weight-of-evidence conclusions for thoracic coarse
particles were based on studies of PM10-2.5, in this review
the EPA focuses primarily on studies that have specifically evaluated
PM10-2.5.\123\
---------------------------------------------------------------------------

\123\ It should also be noted that CASAC endorsed the approach
adopted in the Integrated Science Assessment, which draws weight-of-
evidence conclusions for PM2.5 and PM10-2.5,
but not for PM10 (Samet, 2009f).
---------------------------------------------------------------------------

As discussed in more detail in the Risk Assessment (U.S. EPA,
2010a, Appendix H), the EPA did not conduct a quantitative assessment
of health risks associated with PM10-2.5. The Risk
Assessment concluded that limitations in the monitoring network and in
the health studies that rely on that monitoring network, which would be
the basis for estimating PM10-2.5 health risks, would
introduce significant uncertainty into a PM10-2.5 risk
assessment such that the risk estimates generated would be of limited
value in informing review of the standard. Therefore, it was judged
that a quantitative assessment of PM10-2.5 risks is not
supportable at this time (U.S. EPA, 2010a, p. 2-6). This decision does
not indicate that health effects are not associated with exposure to
thoracic coarse particles. Rather, as noted above, it reflects the
conclusion that limitations in the available health studies and air
quality information would introduce significant uncertainty into a
quantitative assessment of PM10-2.5 risks such that the risk
estimates generated would be of limited value in informing review of
the standard.

B. Health Effects Related to Exposure to Thoracic Coarse Particles

This section briefly outlines the key information presented in
section IV.B of the proposal (77 FR 38947 to 38951, June 29, 2012), and
discussed more fully in the Integrated Science Assessment (U.S. EPA,
2009a, Chapters 2, 4, 5, 6, 7, and 8) and the Policy Assessment (U.S.
EPA, 2011a, Chapter 3), related to health effects associated with
thoracic coarse particle exposures. In looking across the new
scientific evidence available in this review, our overall understanding
of health effects associated with thoracic coarse particle exposures
has been expanded, though important uncertainties remain. Some
highlights of the key policy-relevant scientific evidence available in
this review include the following:

(1) A number of multi-city and single-city epidemiological
studies have evaluated associations between short-term
PM10-2.5 and mortality, cardiovascular effects (e.g.,
including hospital admissions and emergency department visits), and/
or respiratory effects. Despite differences in the approaches used
to estimate ambient PM10-2.5 concentrations, the majority
of these studies have reported positive, though often not
statistically significant, associations with short-term
PM10-2.5 concentrations. Most PM10-2.5 effect
estimates remained positive in co-pollutant models that included
either gaseous or particulate co-pollutants. In U.S. study locations
likely to have met the current PM10 standard during the
study period, a few PM10-2.5 effect estimates were
statistically significant and remained so in co-pollutant
models.\124\
---------------------------------------------------------------------------

\124\ The statistical significance of effect estimates provides
important information on their statistical precision. However, when
a group of studies report effect estimates that are similar in
direction and magnitude, such a pattern of results warrants
consideration of those studies even if not all reported
statistically significant associations in single- or co-pollutant
models (section III.D.2, above). In considering the
PM10-2.5 epidemiologic studies below, the Administrator
considers both the pattern of results across studies and the
statistical significance of those results.
---------------------------------------------------------------------------

(2) A small number of controlled human exposure studies have
reported alterations in heart rate variability or increased
pulmonary inflammation following short-term exposure to
PM10-2.5, providing some support for the associations
reported in epidemiological studies. Toxicological studies that have
examined the effects of PM10-2.5 have used intratracheal
instillation and, because these studies do not directly mirror any
real-world mode of exposure, they provide only limited evidence for
the biological plausibility of PM10-2.5-induced effects.
(3) Using a more formal framework for reaching causal
determinations than used in previous reviews, the Integrated Science
Assessment concluded that the existing evidence is ``suggestive'' of
a causal relationship between short-term PM10-2.5
exposures and mortality, cardiovascular effects, and respiratory
effects (U.S. EPA, 2009a, section 2.3.3).\125\ In contrast, the
Integrated Science Assessment concluded that available evidence is
``inadequate'' to infer a causal relationship between long-term
PM10-2.5 exposures and various health effects.
---------------------------------------------------------------------------

\125\ The causal framework draws upon the assessment and
integration of evidence from across epidemiological, controlled
human exposure, and toxicological studies, and the related
uncertainties that ultimately influence our understanding of the
evidence. This framework employs a five-level hierarchy that
classifies the overall weight-of-evidence using the following
categorizations: Causal relationship, likely to be causal
relationship, suggestive of a causal relationship, inadequate to
infer a causal relationship, and not likely to be a causal
relationship (U.S. EPA, 2009a, Table 1-3). In the case of a
``suggestive'' determination, ``the evidence is suggestive of a
causal relationship with relevant pollutant exposures, but is
limited because chance, bias and confounding cannot be ruled out.
For example, at least one high-quality epidemiologic study shows an
association with a given health outcome but the results of other
studies are inconsistent'' (U.S. EPA, 2009a, Table 1-3).
---------------------------------------------------------------------------

(4) There are several at-risk populations that may be especially
susceptible or vulnerable to PM-related effects, including effects
associated with exposures to coarse particles. These groups include
those with preexisting heart and lung diseases, specific genetic
differences, and lower socioeconomic status as well as the
lifestages of childhood and older adulthood. Evidence for PM-related
effects in these at-risk populations has expanded and is stronger
than previously observed. There is emerging, though still limited,
evidence for additional potentially at-risk populations, such as
those with diabetes, people who are obese, pregnant women, and the
developing fetus.
(5) The Integrated Science Assessment concludes that currently
available evidence is insufficient to draw distinctions in particle
toxicity based on composition and notes that recent studies have
reported that PM (both PM2.5 and PM10-2.5)
from a variety of sources,

[[Page 3168]]

including sources likely to be present in urban and non-urban
locations, is associated with adverse health effects.

Although new PM10-2.5 scientific studies have become
available since the last review and have expanded our understanding of
the association between PM10-2.5 and adverse health effects
(see above and U.S. EPA, 2009a, Chapter 6), important uncertainties
remain. These uncertainties, and their implications for interpreting
the scientific evidence, include the following:

(1) The potential for confounding by co-occurring pollutants,
especially PM2.5, has been addressed with co-pollutant
models in only a relatively small number of PM10-2.5
epidemiological studies (U.S. EPA, 2009a, section 2.3.3). This is a
particularly important limitation given the relatively small body of
experimental evidence (i.e., controlled human exposure and animal
toxicological studies) available to support the associations between
PM10-2.5 and adverse health effects. The net impact of
such limitations is to increase uncertainty in characterizations of
the extent to which PM10-2.5 itself, rather than one or
more co-occurring pollutants, is responsible for the mortality and
morbidity effects reported in epidemiological studies.
(2) There is greater spatial variability in PM10-2.5
concentrations than PM2.5 concentrations, resulting in
increased exposure error for PM10-2.5 (U.S. EPA, 2009a,
p. 2-8). Available measurements do not provide sufficient
information to adequately characterize the spatial distribution of
PM10-2.5 concentrations (U.S. EPA, 2009a, section
3.5.1.1). The net effect of these uncertainties on
PM10-2.5 epidemiological studies is to bias the results
of such studies toward the null hypothesis. That is, as noted in the
Integrated Science Assessment, these limitations in estimates of
ambient PM10-2.5 concentrations ``would tend to increase
uncertainty and make it more difficult to detect effects of
PM10-2.5 in epidemiologic studies'' (U.S. EPA, 2009a, p.
2-21).
(3) Only a relatively small number of PM10-2.5
monitoring sites are currently operating and such sites have been in
operation for a relatively short period of time, limiting the
spatial and temporal coverage for routine measurement of
PM10-2.5 concentrations. Given these limitations in
routine monitoring, epidemiological studies have employed different
approaches for estimating PM10-2.5 concentrations. Given
the relatively small number of PM10-2.5 monitoring sites,
the relatively large spatial variability in ambient
PM10-2.5 concentrations (see above), the use of different
approaches to estimating ambient PM10-2.5 concentrations
across epidemiological studies, and the limitations inherent in such
estimates, the distributions of thoracic coarse particle
concentrations over which reported health outcomes occur remain
highly uncertain (U.S. EPA, 2009a, sections 2.2.3, 2.3.3, 2.3.4, and
3.5.1.1).
(4) There is relatively little information on the chemical and
biological composition of PM10-2.5 and the effects
associated with the various components (U.S. EPA, 2009a, section
2.3.4). Without more information on the chemical speciation of
PM10-2.5, the apparent variability in associations with
health effects across locations is difficult to characterize (U.S.
EPA, 2009a, section 6.5.2.3).
(5) One of the implications of the uncertainties and limitations
discussed above is that the Risk Assessment concluded it would not
be appropriate to conduct a quantitative assessment of health risks
associated with PM10-2.5. The lack of a quantitative
PM10-2.5 risk assessment in the current review adds to
the uncertainty in any conclusions about the extent to which
revision of the current PM10 standard would be expected
to improve the protection of public health, beyond the protection
provided by the current standard.\126\
---------------------------------------------------------------------------

\126\ As noted above, the EPA's decision not to conduct a
quantitative risk assessment reflects uncertainty regarding the
value of such an assessment, but does not indicate that health
effects are not associated with exposure to thoracic coarse
particles.
---------------------------------------------------------------------------

C. Consideration of the Current and Potential Alternative Standards in
the Policy Assessment

The following sections discuss the Policy Assessment's
consideration of the current and potential alternative standards to
protect against exposures to thoracic coarse particles (U.S. EPA,
2011a, chapter 3). Section IV.C.1 discusses the consideration of the
current standard while section IV.C.2 discusses the consideration of
potential alternative standards in terms of the basic elements of a
standard: Indicator, averaging time, form, and level.
1. Consideration of the Current Standard in the Policy Assessment
As discussed above the 24-hour PM10 standard is meant to
protect the public health against exposures to thoracic coarse
particles (i.e., PM10-2.5). In considering the adequacy of
the current PM10 standard, the Policy Assessment considered
the health effects evidence linking short-term PM10-2.5
exposures with mortality and morbidity (U.S. EPA, 2009a, chapters 2 and
6), the ambient PM10 concentrations in PM10-2.5
study locations (U.S. EPA, 2011a, section 3.2.1), the uncertainties and
limitations associated with this health evidence (U.S. EPA, 2011a,
section 3.2.1), and the consideration of these uncertainties and
limitations as part of the weight of evidence conclusions in the
Integrated Science Assessment (U.S. EPA, 2009a).
In considering the health evidence, air quality information, and
associated uncertainties as they relate to the current PM10
standard, the Policy Assessment noted that a decision on the adequacy
of the public health protection provided by that standard is a public
health policy judgment in which the Administrator weighs the evidence
and information, as well as its uncertainties. Therefore, depending on
the emphasis placed on different aspects of the evidence, information,
and uncertainties, consideration of different conclusions on the
adequacy of the current standard could be supported. For example, the
Policy Assessment noted that one approach to considering the evidence,
information, and its associated uncertainties would be to place
emphasis on the following (U.S. EPA, 2011a, section 3.2.3):

(1) While most of PM10-2.5 effect estimates reported
for mortality and morbidity were positive, many were not
statistically significant, even in single-pollutant models. This
includes effect estimates reported in study locations with
PM10 concentrations above those allowed by the current
24-hour PM10 standard.
(2) The number of epidemiological studies that have employed co-
pollutant models to address the potential for confounding,
particularly by PM2.5, remains limited. Therefore, the
extent to which PM10-2.5 itself, rather than one or more
co-pollutants, contributes to reported health effects remains
uncertain.
(3) Only a limited number of experimental studies provide
support for the associations reported in epidemiological studies,
resulting in further uncertainty regarding the plausibility of a
causal link between PM10-2.5 and mortality and morbidity.
(4) Limitations in PM10-2.5 monitoring and the
different approaches used to estimate PM10-2.5
concentrations across epidemiological studies result in uncertainty
as to the ambient PM10-2.5 concentrations at which the
reported effects occur.
(5) The chemical and biological composition of
PM10-2.5, and the effects associated with the various
components, remains uncertain. Without more information on the
chemical speciation of PM10-2.5, the apparent variability
in associations across locations is difficult to interpret.
(6) In considering the available evidence and its associated
uncertainties, the Integrated Science Assessment concluded that the
evidence is ``suggestive'' of a causal relationship between short-
term PM10-2.5 exposures and mortality, cardiovascular
effects, and respiratory effects. These weight-of-evidence
conclusions contrast with those for the relationships between
PM2.5 exposures and adverse health effects, which were
judged in the Integrated Science Assessment to be either ``causal''
or ``likely causal'' for mortality, cardiovascular effects, and
respiratory effects.

The Policy Assessment concluded that, to the extent a decision on
the adequacy of the current 24-hour PM10 standard were to
place emphasis on the considerations noted above, it could be judged
that, although it remains appropriate to maintain a standard to protect
against short-term exposures to

[[Page 3169]]

thoracic coarse particles, the available evidence suggests that the
current 24-hour PM10 standard appropriately protects public
health and provides an adequate margin of safety against effects that
have been associated with PM10-2.5 exposures. Although such
an approach to considering the adequacy of the current standard would
recognize the positive, and in some cases statistically significant,
associations between all types of PM10-2.5 and mortality and
morbidity, it would place relatively greater emphasis on the
limitations and uncertainties noted above, which tend to complicate the
interpretation of that evidence.
In addition, the Policy Assessment noted the judgment that, given
the uncertainties and limitations in the PM10-2.5 health
evidence and air quality information, it would not have been
appropriate to conduct a quantitative assessment of health risks
associated with PM10-2.5 (U.S. EPA, 2011a, p. 3-6; U.S. EPA,
2010a, pp. 2-6 to 2-7, Appendix H). As discussed above, the lack of a
quantitative PM10-2.5 risk assessment adds to the
uncertainty associated with any characterization of potential public
health improvements that would be realized with a revised standard.
The Policy Assessment also noted an alternative approach to
considering the evidence and its uncertainties would place emphasis on
the following (U.S. EPA, 2011a, section 3.2.3):

(1) Several multi-city epidemiological studies conducted in the
U.S., Canada, and Europe, as well as a number of single-city
studies, have reported generally positive, and in some cases
statistically significant, associations between short-term
PM10-2.5 concentrations and adverse health endpoints
including mortality and cardiovascular-related and respiratory-
related hospital admissions and emergency department visits.
(2) Both single-city and multi-city analyses, using different
approaches to estimate ambient PM10-2.5 concentrations,
have reported positive PM10-2.5 effect estimates in
locations that would likely have met the current 24-hour
PM10 standard. In a few cases, these PM10-2.5
effect estimates were statistically significant.
(3) While limited in number, studies that have evaluated co-
pollutant models have generally reported that PM10-2.5
effect estimates remain positive, and in a few cases statistically
significant, when these models include gaseous pollutants or fine
particles.
(4) Support for the plausibility of the associations reported in
epidemiological studies is provided by a small number of controlled
human exposure studies reporting that short-term (i.e., 2-hour)
exposures to PM10-2.5 decrease heart rate variability and
increase markers of pulmonary inflammation.

This approach to considering the health evidence, air quality
information, and the associated uncertainties would place substantial
weight on the generally positive PM10-2.5 effect estimates
that have been reported for mortality and morbidity, even those effect
estimates that are not statistically significant. The Policy Assessment
concluded that this could be judged appropriate given that consistent
results have been reported across multiple studies using different
approaches to estimate ambient PM10-2.5 concentrations and
that exposure measurement error, which is likely to be larger for
PM10-2.5 than for PM2.5, tends to bias the
results of epidemiological studies toward the null hypothesis, making
it less likely that associations will be detected. Such an approach
would place less weight on the uncertainties and limitations in the
evidence that resulted in the Integrated Science Assessment conclusions
that the evidence is only suggestive of a causal relationship.
Given all of the above, the Policy Assessment concluded that it
would be appropriate to consider either retaining or revising the
current 24-hour PM10 standard, depending on the approach
taken to considering the available evidence, air quality information,
and the uncertainties and limitations associated with that evidence and
information (U.S. EPA, 2011a, section 3.2.3).

2. Consideration of Potential Alternative Standards in the Policy
Assessment

Given the conclusion that it would be appropriate to consider
either retaining or revising the current PM10 standard, the
Policy Assessment also considered what potential alternative standards,
if any, could be supported by the available scientific evidence in
order to increase public health protection against exposures to
PM10-2.5. The Policy Assessment considered such potential
alternative standards defined in terms of the elements of a standard
(i.e., indicator, averaging time, form, and level). Key conclusions
from the Policy Assessment regarding indicator, averaging time, and
form included the following:

(1) A PM10 indicator would continue to appropriately
target protection against thoracic coarse particle exposures to
those locations where the evidence is strongest for associations
with adverse health effects (i.e., urban areas).
(2) The available evidence supports the importance of
maintaining a standard that protects against short-term exposures to
all thoracic coarse particles. Given that the majority of this
evidence is based on 24-hour average thoracic coarse particle
concentrations, consideration of a 24-hour averaging time remains
appropriate.
(3) Given the limited body of evidence supporting
PM10-2.5-related effects following long-term exposures,
which resulted in the Integrated Science Assessment conclusion that
the available evidence is ``inadequate'' to infer a causal
relationship between long-term PM10-2.5 exposures and a
variety of health effects, consideration of an annual thoracic
coarse particle standard is not supported at this time.
(4) To the extent it is judged appropriate to revise the current
24-hour PM10 standard, it would be appropriate to
consider revising the form to the 3-year average of the 98th
percentile of the annual distribution of 24-hour PM10
concentrations.

In considering the available evidence and air quality information
within the context of identifying potential alternative standard levels
for consideration (assuming a decision were made that it is appropriate
to amend the standard), the Policy Assessment first noted that a
standard level as high as about 85 [mu]g/m\3\, for a 24-hour
PM10 standard with a 98th percentile form, could be
supported. Based on considering air quality concentrations in study
locations, the Policy Assessment noted that such a standard level would
be expected to maintain PM10 and PM10-2.5
concentrations below those present in U.S. locations of single-city
studies where PM10-2.5 effect estimates have been reported
to be positive and statistically significant and below those present in
some locations where single-city studies reported PM10-2.5
effect estimates that were positive, but not statistically significant.
These include some locations likely to have met the current
PM10 standard during the study periods (U.S. EPA, 2011a,
section 3.3.4).
The Policy Assessment also noted that, based on analysis of the
number of people living in counties that could violate the current and
potential alternative PM10 standards, a 24-hour
PM10 standard with a 98th percentile form and a level
between 75 and 80 [mu]g/m\3\ would provide a level of public health
protection that is generally equivalent, across the U.S., to that
provided by the current standard. Given this, the Policy Assessment
concluded that it would be appropriate to consider standard levels in
the range of approximately 75 to 80 [mu]g/m\3\ (with a 98th percentile
form), to the extent population counts were emphasized in comparing the
public health protection provided by the current and potential
alternative standards and to the extent it was judged appropriate to
set a revised standard providing at least the level of public health
protection that is provided by the current standard, based on such
population counts (U.S. EPA, 2011a, section 3.3.4).

[[Page 3170]]

The Policy Assessment also concluded that alternative approaches to
considering the evidence could lead to consideration of standard levels
below 75 [mu]g/m\3\ for a standard with a 98th percentile form. For
example, a number of single-city epidemiological studies have reported
positive, though not statistically significant, PM10-2.5
effect estimates in locations with 98th percentile PM10
concentrations below 75 [mu]g/m\3\. Given that exposure error is
particularly important for PM10-2.5 epidemiological studies
and can bias the results of these studies toward the null hypothesis
(see section IV.B above), the Policy Assessment noted that it could be
judged appropriate to place more weight on positive associations
reported in these epidemiological studies, even when those associations
are not statistically significant. In addition, the Policy Assessment
noted that multi-city averages of 98th percentile PM10
concentrations in the locations evaluated by U.S. multi-city studies of
thoracic coarse particles (Zanobetti and Schwartz, 2009; Peng et al.,
2008) were near or below 75 ppb. Despite uncertainties in the extent to
which effects reported in multi-city studies are associated with the
short-term air quality in any particular location, the Policy
Assessment noted that emphasis could be placed on these multi-city
averaged concentrations. The Policy Assessment concluded that, to the
extent more weight is placed on single-city studies reporting positive,
but not statistically significant, PM10-2.5 effect estimates
and on multi-city studies, it could be appropriate to consider standard
levels as low as 65 [mu]g/m\3\ with a 98th percentile form (U.S. EPA,
2011a, section 3.3.4).
In considering potential alternative standard levels below 65
[mu]g/m\3\, the Policy Assessment noted that the overall body of
PM10-2.5 health evidence is relatively uncertain, with
somewhat stronger support in U.S. studies for associations with
PM10-2.5 in locations with 98th percentile PM10
concentrations above 85 [mu]g/m\3\ than in locations with 98th
percentile PM10 concentrations below 65 [mu]g/m\3\. In light
of the limitations in the evidence for a relationship between
PM10-2.5 and adverse health effects in locations with
relatively low PM10 concentrations, along with the overall
uncertainties in the body of PM10-2.5 health evidence as
described above and in the Integrated Science Assessment, the Policy
Assessment concluded that consideration of standard levels below 65
[mu]g/m\3\ was not appropriate (U.S. EPA, 2011a, section 3.3.4).

D. CASAC Advice

Following their review of the first and second draft Policy
Assessments, CASAC provided advice and recommendations regarding the
current and potential alternative standards for thoracic coarse
particles (Samet, 2010c,d). With regard to the existing PM10
standard, CASAC concluded that ``the current data, while limited, is
sufficient to call into question the level of protection afforded the
American people by the current standard'' (Samet, 2010d, p. 7). In
drawing this conclusion, CASAC noted the positive associations in
multi-city and single-city studies, including in locations with
PM10 concentrations below those allowed by the current
standard. In addition, CASAC gave ``significant weight to studies that
have generally reported that PM10-2.5 effect estimates
remain positive when evaluated in co-pollutant models'' and concluded
that ``controlled human exposure PM10-2.5 studies showing
decreases in heart rate variability and increases in markers of
pulmonary inflammation are deemed adequate to support the plausibility
of the associations reported in epidemiologic studies'' (Samet, 2010d,
p. 7).\127\ Given all of the above conclusions CASAC recommended that
``the primary standard for PM10 should be revised'' (Samet,
2010d, p. ii and p. 7). In discussing potential revisions, while CASAC
noted that the scientific evidence supports adoption of a standard at
least as stringent as the current standard, they recommended revising
the current standard in order to increase public health protection. In
considering potential alternative standards, CASAC drew conclusions and
made recommendations in terms of the major elements of a standard:
indicator, averaging time, form, and level.
---------------------------------------------------------------------------

\127\ Nonetheless, CASAC endorsed the Integrated Science
Assessment weight of evidence conclusions for PM10-2.5
(i.e., that the evidence is only ``suggestive'' of a causal
relationship between short-term exposures and mortality, respiratory
effects, and cardiovascular effects) (Samet, 2009e; Samet, 2009f).
---------------------------------------------------------------------------

The CASAC agreed with the EPA staff's conclusions that the
available evidence supports consideration in the current review of
retaining the current PM10 indicator and the current 24-hour
averaging time (Samet, 2010c, Samet, 2010d). Specifically, with regard
to indicator, CASAC concluded that ``[w]hile it would be preferable to
use an indicator that reflects the coarse PM directly linked to health
risks (PM10-2.5), CASAC recognizes that there is not yet
sufficient data to permit a change in the indicator from
PM10 to one that directly measures thoracic coarse
particles'' (Samet, 2010d, p. ii). In addition, CASAC ``vigorously
recommends the implementation of plans for the deployment of a network
of PM10-2.5 sampling systems so that future epidemiological
studies will be able to more thoroughly explore the use of
PM10-2.5 as a more appropriate indicator for thoracic coarse
particles'' (Samet, 2010d, p. 7).
The CASAC also agreed that the evidence supports consideration of a
potential alternative form. Specifically, CASAC ``felt strongly that it
is appropriate to change the statistical form of the PM10
standard to a 98th percentile'' (Samet, 2010d, p.7). In reaching this
conclusion, CASAC noted that ``[p]ublished work has shown that the
percentile form has greater power to identify non-attainment and a
smaller probability of misclassification relative to the expected
exceedance form of the standard'' (Samet, 2010d. p. 7).
With regard to standard level, in conjunction with a 98th
percentile form, CASAC concluded that ``alternative standard levels of
85 and 65 [mu]g/m\3\ (based on consideration of 98th percentile
PM10 concentration) could be justified'' (Samet, 2010d,
p.8). However, in considering the evidence and uncertainties, CASAC
recommended a standard level from the lower part of the range discussed
in the Policy Assessment, recommending a level ``somewhere in the range
of 75 to 65 [mu]g/m\3\'' (Samet, 2010d, p. ii).
In making this recommendation, CASAC noted that the number of
people living in counties with air quality not meeting the current
standard is approximately equal to the number living in counties that
would not meet a 98th percentile standard with a level between 75 and
80 [mu]g/m\3\. CASAC used this information as the basis for their
conclusion that a 98th percentile standard between 75 and 80 [mu]g/m\3\
would be ``comparable to the degree of protection afforded to the
current PM10 standard'' (Samet, 2010d, p. ii). Given this
conclusion regarding the comparability of the current and potential
alternative standards, as well as their conclusion on the public health
protection provided by the current standard (i.e., that available
evidence is sufficient to call it into question), CASAC recommended a
level within a range of 75 to 65 [mu]g/m\3\ in order to increase public
health protection, relative to that provided by the current standard
(Samet 2010d, p. ii).

[[Page 3171]]

E. Administrator's Proposed Conclusions Concerning the Adequacy of the
Current Primary PM10 Standard

In considering the evidence and information as they relate to the
adequacy of the current 24-hour PM10 standard, the
Administrator first noted in the proposal that this standard is meant
to protect the public health against effects associated with short-term
exposures to PM10-2.5. In the last review, it was judged
appropriate to maintain such a standard given the ``growing body of
evidence suggesting causal associations between short-term exposure to
thoracic coarse particles and morbidity effects, such as respiratory
symptoms and hospital admissions for respiratory diseases, and possibly
mortality'' (71 FR 61185, October 17, 2006). Given the continued
expansion in the body of scientific evidence linking short-term
PM10-2.5 to health outcomes such as premature death and
hospital visits, discussed in detail in the Integrated Science
Assessment (U.S. EPA, 2009a, Chapter 6) and summarized in the proposal,
the Administrator provisionally concluded that the available evidence
continued to support the appropriateness of maintaining a standard to
protect the public health against effects associated with short-term
(e.g., 24-hour) exposures to all PM10-2.5. In drawing
provisional conclusions in the proposal as to whether the current
PM10 standard remains requisite (i.e., neither more nor less
stringent than necessary) to protect public health with an adequate
margin of safety against such exposures, the Administrator considered
the following:

(1) The extent to which it is appropriate to maintain a standard
that provides some measure of protection against all
PM10-2.5, regardless of composition or source of origin;
(2) The extent to which it is appropriate to retain a
PM10 indicator for a standard meant to protect against
exposures to ambient PM10-2.5; and
(3) The extent to which the current PM10 standard
provides an appropriate degree of public health protection.

With regard to the first point, the proposal noted the conclusion
from the last review that dosimetric, toxicological, occupational, and
epidemiological evidence supported retention of a primary standard to
provide some measure of protection against short-term exposures to all
thoracic coarse particles, regardless of their source of origin or
location, consistent with the Act's requirement that primary NAAQS
provide requisite protection with an adequate margin of safety (71 FR
61197). In that review, the EPA concluded that PM from a number of
source types, including motor vehicle emissions, coal combustion, oil
burning, and vegetative burning, are associated with health effects
(U.S. EPA, 2004). This information formed part of the basis for the
D.C. Circuit's holding that it was appropriate for the thoracic coarse
particle standard to provide ``some protection from exposure to
thoracic coarse particles * * * in all areas'' (American Farm Bureau
Federation v. EPA, 559 F. 3d at 532-33).
In considering this issue in the proposal, the Administrator judged
that the expanded body of scientific evidence in this review provides
even more support for a standard that protects against exposures to all
thoracic coarse particles, regardless of their location or source of
origin. Specifically, the Administrator noted that epidemiological
studies have reported positive associations between PM10-2.5
and mortality or morbidity in a large number of cities across North
America, Europe, and Asia, encompassing a variety of environments where
PM10-2.5 sources and composition are expected to vary
widely. See 77 FR 38959. In considering this evidence, the Integrated
Science Assessment concluded that ``many constituents of PM can be
linked with differing health effects'' (U.S. EPA, 2009a, p. 2-26).
While PM10-2.5 in most of these study areas is of largely
urban origin, the Administrator noted that some recent studies have
also linked mortality and morbidity with relatively high ambient
concentrations of thoracic coarse particles of non-urban crustal
origin. In considering these studies, she noted the Integrated Science
Assessment's conclusion that ``PM (both PM2.5 and
PM10-2.5) from crustal, soil or road dust sources or PM
tracers linked to these sources are associated with cardiovascular
effects'' (U.S. EPA, 2009a, p. 2-26).
In light of this body of available evidence reporting
PM10-2.5-associated health effects across different
locations with a variety of sources, as well as the Integrated Science
Assessment's conclusions regarding the links between adverse health
effects and PM sources and composition, the Administrator provisionally
concluded in the proposal that it is appropriate to maintain a standard
that provides some measure of protection against exposures to all
thoracic coarse particles, regardless of their location, source of
origin, or composition (77 FR 38959-60).
With regard to the second point, in considering the appropriateness
of a PM10 indicator for a standard meant to provide such
public health protection, the Administrator noted that the rationale
used in the last review to support the unqualified PM10
indicator (see above) remains relevant in the current review.
Specifically, as an initial consideration, she noted that
PM10 mass includes both coarse PM (PM10-2.5) and
fine PM (PM2.5). As a result, the concentration of
PM10-2.5 allowed by a PM10 standard set at a
single level declines as the concentration of PM2.5
increases. At the same time, the Administrator noted that
PM2.5 concentrations tend to be higher in urban areas than
in rural areas (U.S. EPA, 2005, p. 2-54, and Figures 2-23 and 2-24)
and, therefore, a PM10 standard will generally allow lower
PM10-2.5 concentrations in urban areas than in rural areas.
77 FR 38960.
In considering the appropriateness of this variation in allowable
PM10-2.5 concentrations, the Administrator considered the
relative strength of the evidence for health effects associated with
PM10-2.5 of urban origin versus non-urban origin. She
specifically noted that, as described above and similar to the
scientific evidence available in the last review, the large majority of
the available evidence for thoracic coarse particle health effects
comes from studies conducted in locations with sources more typical of
urban and industrial areas than of rural areas. Although as just noted,
associations with adverse health effects have been reported in some
study locations where PM10-2.5 is largely non-urban in
origin (i.e., in dust storm studies), particle concentrations in these
study areas are typically much higher than reported in study locations
where the PM10-2.5 is of urban origin. Therefore, the
Administrator noted that the strongest evidence for a link between
PM10-2.5 and adverse health impacts, particularly for such a
link at relatively low particle concentrations, comes from studies
where exposure is to PM10-2.5 of urban or industrial origin.
77 FR 38960.
The Administrator also noted that chemical constituents present at
higher levels in urban or industrial areas, including byproducts of
incomplete combustion (e.g. polycyclic aromatic hydrocarbons) emitted
as PM2.5 from motor vehicles as well as metals and other
contaminants emitted from anthropogenic sources, can contaminate
PM10-2.5 (U.S. EPA, 2004, p. 8-344; 71 FR 2665). While the
Administrator acknowledged the uncertainty expressed in the Integrated
Science Assessment regarding the extent to which, based on available
evidence, particle composition can be linked to health outcomes, she
also considered the possibility that PM10-2.5 contaminants
typical of urban or industrial areas could increase the

[[Page 3172]]

toxicity of thoracic coarse particles in urban locations (77 FR 38960).
Given that the large majority of the evidence for
PM10-2.5 toxicity, particularly at relatively low particle
concentrations, comes from study locations where thoracic coarse
particles are of urban origin, and given the possibility that
PM10-2.5 contaminants in urban areas could increase particle
toxicity, the Administrator provisionally concluded in the proposal
that it remains appropriate to maintain a standard that targets public
health protection to urban locations. Specifically, she concluded at
proposal that it is appropriate to maintain a standard that allows
lower ambient concentrations of PM10-2.5 in urban areas,
where the evidence is strongest that thoracic coarse particles are
linked to mortality and morbidity, and higher concentrations in non-
urban areas, where the public health concerns are less certain. Id.
Given all of the above considerations and conclusions, the
Administrator judged that the available evidence supported retaining a
PM10 indicator for a standard that is meant to protect
against exposure to thoracic coarse particles. In reaching this initial
judgment, she noted that, to the extent a PM10 indicator
results in lower allowable concentrations of thoracic coarse particles
in some areas compared to others, lower concentrations will be allowed
in those locations (i.e., urban or industrial areas) where the science
has shown the strongest evidence of adverse health effects associated
with exposure to thoracic coarse particles and where we have the most
concern regarding PM10-2.5 toxicity. Therefore, the
Administrator provisionally concluded that the varying amounts of
coarse particles that are allowed in urban vs. non-urban areas under
the 24-hour PM10 standard, based on the varying levels of
PM2.5 present, appropriately reflect the differences in the
strength of evidence regarding coarse particle effects in urban and
non-urban areas (77 FR 38960).
In reaching this provisional conclusion, the Administrator also
noted that, in their review of the second draft Policy Assessment,
CASAC concluded that ``[w]hile it would be preferable to use an
indicator that reflects the coarse PM directly linked to health risks
(PM10-2.5), CASAC recognizes that there is not yet
sufficient data to permit a change in the indicator from
PM10 to one that directly measures thoracic coarse
particles'' (Samet, 2010d, p. ii). In addition, CASAC ``vigorously
recommends the implementation of plans for the deployment of a network
of PM10-2.5 sampling systems so that future epidemiological
studies will be able to more thoroughly explore the use of
PM10-2.5 as a more appropriate indicator for thoracic coarse
particles'' (Samet, 2010d, p. 7). Given this recommendation, the
Administrator further judged that, although current evidence is not
sufficient to identify a standard based on an alternative indicator
that would be requisite to protect public health with an adequate
margin of safety across the United States, consideration of alternative
indicators (e.g., PM10-2.5) in future reviews is desirable
and could be informed by additional research, as described in the
Policy Assessment (U.S. EPA, 2011a, section 3.5).
With regard to the third point, in evaluating the degree of public
health protection provided by the current PM10 standard, the
Administrator noted that the Policy Assessment discussed two different
approaches to considering the scientific evidence and air quality
information (U.S. EPA, 2011a, section 3.2.3). These different
approaches, which are described above (section IV.C.1), lead to
different conclusions regarding the appropriateness of the degree of
public health protection provided by the current PM10
standard. The Administrator further noted that the primary difference
between the two approaches lies in the extent to which weight is placed
on the following (U.S. EPA, 2011a, section 3.2.3):

(1) The PM10-2.5 weight-of-evidence classifications
presented in the Integrated Science Assessment concluding that the
existing evidence is suggestive of a causal relationship between
short-term PM10-2.5 exposures and mortality,
cardiovascular effects, and respiratory effects (a classification
supported by CASAC);
(2) Individual PM10-2.5 epidemiological studies
reporting associations in locations that meet the current
PM10 standard, including associations that are not
statistically significant;
(3) The limited number of PM10-2.5 epidemiological
studies that have evaluated co-pollutant models;
(4) The limited number of PM10-2.5 controlled human
exposure studies;
(5) Uncertainties in the PM10-2.5 air quality
concentrations reported in epidemiological studies, given
limitations in PM10-2.5 monitoring data and the different
approaches used across studies to estimate ambient
PM10-2.5 concentrations; and
(6) Uncertainties and limitations in the evidence that tend to
call into question the presence of a causal relationship between
PM10-2.5 exposures and mortality/morbidity.

In evaluating the different possible approaches to considering the
public health protection provided by the current PM10
standard, the Administrator first noted that when the available
PM10-2.5 scientific evidence and its associated
uncertainties are considered, the Integrated Science Assessment
concluded that the evidence is suggestive of a causal relationship
between short-term PM10-2.5 exposures and mortality,
cardiovascular effects, and respiratory effects. As discussed in
section IV.B.1 above and in more detail in the Integrated Science
Assessment (U.S. EPA, 2009a, section 1.5), a suggestive determination
is made when the ``[e]vidence is suggestive of a causal relationship
with relevant pollutant exposures, but is limited because chance, bias
and confounding cannot be ruled out.'' In contrast, the Administrator
noted that she proposed to strengthen the annual fine particle standard
based on a body of scientific evidence judged sufficient to conclude
that a causal relationship exists (i.e., mortality, cardiovascular
effects) or is likely to exist (i.e., respiratory effects) (section
III.B). 77 FR 38961. The suggestive judgment for PM10-2.5
reflects the greater degree of uncertainty associated with this body of
evidence, as discussed above (sections IV.B and IV.C) and summarized
below.
In the proposal (77 FR 38961), the Administrator noted that the
important uncertainties and limitations associated with the scientific
evidence and air quality information raise questions as to whether
public health benefits would be achieved by revising the existing
PM10 standard. Such uncertainties and limitations include
the following:

(1) While PM10-2.5 effect estimates reported for
mortality and morbidity were generally positive, most were not
statistically significant, even in single-pollutant models. This
includes effect estimates reported in some study locations with
PM10 concentrations above those allowed by the current
24-hour PM10 standard.
(2) The number of epidemiological studies that have employed co-
pollutant models to address the potential for confounding,
particularly by PM2.5, remains limited. Therefore, the
extent to which PM10-2.5 itself, rather than one or more
co-pollutants, contributes to reported health effects is less
certain.
(3) Only a limited number of experimental studies (i.e.,
controlled human exposure and animal toxicological) provide support
for the associations reported in epidemiological studies, resulting
in further uncertainty regarding the plausibility of the
associations between PM10-2.5 and mortality and morbidity
reported in epidemiological studies.
(4) Limitations in PM10-2.5 monitoring data and the
different approaches used by epidemiological study researchers to
estimate PM10-2.5 concentrations across epidemiological
studies result in uncertainty in the ambient PM10-2.5
concentrations at which the reported effects occur, increasing
uncertainty in estimates of the extent to

[[Page 3173]]

which changes in ambient PM10-2.5 concentrations would
likely impact public health.
(5) The lack of a quantitative PM10-2.5 risk
assessment further contributes to uncertainty regarding the extent
to which any revisions to the current PM10 standard would
be expected to improve the protection of public health, beyond the
protection provided by the current standard (see section III.B.5
above).
(6) The chemical and biological composition of
PM10-2.5, and the effects associated with the various
components, remains uncertain. Without more information on the
chemical speciation of PM10-2.5, the apparent variability
in associations across locations is difficult to interpret.

In considering these uncertainties and limitations, the
Administrator noted in particular the considerable degree of
uncertainty in the extent to which health effects reported in
epidemiological studies are due to PM10-2.5 itself, as
opposed to one or more co-occurring pollutants. As discussed above,
this uncertainty reflects the fact that there are a relatively small
number of PM10-2.5 studies that have utilized co-pollutant
models, particularly co-pollutant models that have included
PM2.5, and a very limited body of controlled human exposure
evidence supporting the biological plausibility of a causal
relationship between PM10-2.5 and mortality and morbidity at
ambient concentrations. The Administrator noted that these important
limitations in the overall body of health evidence introduce
uncertainty into the interpretation of individual epidemiological
studies, particularly those studies reporting associations with
PM10-2.5 that are not statistically significant. Given this,
the Administrator reached the provisional conclusion in the proposal
that it is appropriate to place relatively little weight on
epidemiological studies reporting associations with PM10-2.5
that are not statistically significant in single-pollutant and/or co-
pollutant models. Id.
With regard to this provisional conclusion, the Administrator noted
that, for single-city mortality studies conducted in the United States
where ambient PM10 concentration data were available for
comparison to the current standard, positive and statistically
significant PM10-2.5 effect estimates were only reported in
study locations that would likely have violated the current
PM10 standard during the study period (U.S. EPA, 2011a,
Figure 3-2). In U.S. study locations that would likely have met the
current standard, PM10-2.5 effect estimates for mortality
were positive, but not statistically significant (U.S. EPA, 2011a,
Figure 3-2). In considering U.S. study loc`ations where single-city
morbidity studies were conducted, and which would likely have met the
current PM10 standard during the study period, the
Administrator noted that PM10-2.5 effect estimates were both
positive and negative, with most not statistically significant (U.S.
EPA, 2011a, Figure 3-3).
In addition, in considering single-city analyses for the locations
evaluated in a large U.S. multi-city mortality study (Zanobetti and
Schwartz, 2009), the Administrator noted that associations in most of
the study locations were not statistically significant and that this
was the only study to estimate ambient PM10-2.5
concentrations as the difference between county-wide PM10
and PM2.5 mass. As discussed in the Policy Assessment and in
the proposal, it is not clear how computed PM10-2.5
measurements, such as those used by Zanobetti and Schwartz (2009),
compare with the PM10-2.5 concentrations obtained in other
studies either by direct measurement or by calculating the difference
using co-located samplers (U.S. EPA, 2009a, section 6.5.2.3). For these
reasons, in the proposal the Administrator noted that ``there is
considerable uncertainty in interpreting the associations in these
single-city analyses'' (77 FR 38961-62).
The Administrator acknowledged that an approach to considering the
available scientific evidence and air quality information that
emphasizes the above considerations differs from the approach taken by
CASAC. Specifically, in its review of the draft Policy Assessment CASAC
placed a substantial amount of weight on individual studies,
particularly those reporting positive health effects associations for
PM10-2.5 in locations that met the current PM10
standard during the study period. In emphasizing these studies, as well
as the limited number of supporting studies that have evaluated co-
pollutant models and the small number of supporting experimental
studies, CASAC concluded that ``the current data, while limited, is
sufficient to call into question the level of protection afforded the
American people by the current standard'' (Samet, 2010d, p. 7) and
recommended revising the current PM10 standard (Samet,
2010d).
The Administrator carefully considered CASAC's advice and
recommendations. She noted that in making its recommendation on the
current PM10 standard, CASAC did not discuss its approach to
considering the important uncertainties and limitations in the health
evidence, and did not discuss how these uncertainties and limitations
were reflected in its recommendation. Nor did CASAC discuss
uncertainties in the reported concentrations of PM10-2.5 in
the epidemiological studies, or how reported concentrations in the
various studies relate to one another when differing measurement
methodologies are used. As discussed above, such uncertainties and
limitations contributed to the conclusions in the Integrated Science
Assessment that the PM10-2.5 evidence is only suggestive of
a causal relationship, a conclusion that CASAC endorsed (Samet,
2009e,f). Given the importance of these uncertainties and limitations
to the interpretation of the evidence, as reflected in the weight of
evidence conclusions in the Integrated Science Assessment and as
discussed above, the Administrator judged it appropriate to consider
and account for them when drawing conclusions about the potential
implications of individual PM10-2.5 health studies for the
current standard.
In light of the above approach to considering the scientific
evidence, air quality information, and associated uncertainties, the
Administrator reached the following provisional conclusions in the
proposal:

(1) When viewed as a whole the available evidence and
information suggests that the degree of public health protection
provided against short-term exposures to PM10-2.5 does
not need to be increased beyond that provided by the current
PM10 standard. This provisional conclusion noted the
important uncertainties and limitations associated with the overall
body of health evidence and air quality information for
PM10-2.5, as discussed above and as reflected in the
Integrated Science Assessment weight-of-evidence conclusions; that
PM10-2.5 effect estimates for the most serious health
effect, mortality, were not statistically significant in U.S.
locations that met the current PM10 standard and where
coarse particle concentrations were either directly measured or
estimated based on co-located samplers; and that PM10-2.5
effect estimates for morbidity endpoints were both positive and
negative in locations that met the current standard, with most not
statistically significant.
(2) The degree of public health protection provided by the
current standard is not greater than warranted. This provisional
conclusion noted that positive and statistically significant
associations with mortality were reported in single-city U.S. study
locations likely to have violated the current PM10
standard.\128\
---------------------------------------------------------------------------

\128\ There are similarities with the conclusions drawn by the
Administrator in the last review. There, the Administrator concluded
that there was no basis for concluding that the degree of protection
afforded by the current PM10 standards in urban areas is
greater than warranted, since potential mortality effects have been
associated with air quality levels not allowed by the current 24-
hour standard, but have not been associated with air quality levels
that would generally meet that standard, and morbidity effects have
been associated with air quality levels that exceeded the current
24-hour standard only a few times (71 FR 61202). In addition, the
Administrator concluded that there was a high degree of uncertainty
in the relevant population exposures implied by the morbidity
studies suggesting that there is little basis for concluding that a
greater degree of protection is warranted. Id. The D.C. Circuit in
American Farm Bureau Federation v EPA explicitly endorsed this
reasoning. 559 F. 3d at 534.

[[Page 3174]]

---------------------------------------------------------------------------

In reaching these provisional conclusions, the Administrator noted
that the Policy Assessment also discussed the potential for a revised
PM10 standard (i.e., with a revised form and level) to be
``generally equivalent'' to the current standard, but to better target
public health protection to locations where there is greater concern
regarding PM10-2.5-associated health effects (U.S. EPA,
2011a, sections 3.3.3 and 3.3.4). In considering such a potential
revised standard, the Policy Assessment discussed the large amount of
variability in PM10 air quality correlations across
monitoring locations and over time (U.S. EPA, 2011a, Figure 3-7) and
the regional variability in the relative degree of public health
protection that could be provided by the current and potential
alternative standards (U.S. EPA, 2011a, Table 3-2). In light of this
variability, the Administrator noted the Policy Assessment conclusion
that no single revised PM10 standard (i.e., with a revised
form and level) would provide public health protection equivalent to
that provided by the current standard, consistently over time and
across locations (U.S. EPA, 2011a, section 3.3.4). That is, a revised
standard, even one that is meant to be ``generally equivalent'' to the
current PM10 standard, could increase protection in some
locations while decreasing protection in others (77 FR 38962).
In considering the appropriateness of revising the current
PM10 standard in this way, the Administrator noted the
following:

(1) Positive PM10-2.5 effect estimates for mortality
were not statistically significant in U.S. locations that met the
current PM10 standard and where coarse particle
concentrations were either directly measured or estimated based on
co-located samplers, while positive and statistically significant
associations with mortality were reported in locations likely to
have violated the current PM10 standard.
(2) Effect estimates for morbidity endpoints in locations that
met the current standard were both positive and negative, with most
not statistically significant.
(3) Important uncertainties and limitations associated with the
overall body of health evidence and air quality information for
PM10-2.5, as discussed above and as reflected in the
Integrated Science Assessment weight-of-evidence conclusions, call
into question the extent to which the type of quantified and refined
targeting of public health protection envisioned under a revised
standard could be reliably accomplished.

Given all of the above considerations, the Administrator noted that
there is a large amount of uncertainty in the extent to which public
health would be improved by changing the locations to which the
PM10 standard targets protection. Therefore, she reached the
provisional conclusion that the current PM10 standard should
not be revised in order to change that targeting of protection.
In considering all of the above, including the scientific evidence,
the air quality information, the associated uncertainties, and CASAC's
advice, the Administrator reached the provisional conclusion that the
current 24-hour PM10 standard is requisite (i.e., neither
more protective nor less protective than necessary) to protect public
health with an adequate margin of safety against effects that have been
associated with PM10-2.5. In light of this provisional
conclusion, the Administrator proposed to retain the current
PM10 standard in order to protect against health effects
associated with short-term exposures to PM10-2.5 (77 FR
38963).
The Administrator recognized that her proposed conclusions and
decision to retain the current PM10 standard differed from
CASAC's recommendations, stemming from the differences in how the
Administrator and CASAC considered and accounted for the evidence and
its limitations and uncertainties. In light of CASAC's views and
recommendation to revise the current PM10 standard, the
Administrator welcomed the public's views on these different approaches
to considering and accounting for the evidence and its limitations and
uncertainties, as well as on the appropriateness of revising the
primary PM10 standard, including revising the form and level
of the standard. In doing so, the Administrator solicited comment on
all aspects of the proposed decision, including her rationale for
reaching the provisional conclusion that the current PM10
standard is requisite to protect public health with an adequate margin
of safety and the provisional conclusion that it is not appropriate to
revise the current PM10 standard by setting a ``generally
equivalent'' standard with the goal of better targeting public health
protection.

F. Public Comments on the Administrator's Proposed Decision To Retain
the Primary PM10 Standard

This section discusses the major public comments received on the
Administrator's proposed decision to retain the primary PM10 standard.
Additional comments are addressed in the Response to Comments Document
(U.S. EPA, 2012a).
Many public commenters agreed with the Administrator's proposed
decision to retain the current 24-hour primary PM10
standard. Among those expressing a position on this proposed decision,
industry groups and most State and Local commenters endorsed the
Administrator's proposed rationale for retaining the current primary
PM10 standard, including her consideration of the available
scientific evidence and associated uncertainties and her consideration
of CASAC recommendations.
Although industry commenters generally agreed with the
Administrator's proposed decision to retain the current primary
PM10 standard, some also contended that the current standard
is ``excessively precautionary'' (NMA and NCBA, 2012, p. 4) and a few
expressed support for a less stringent standard for coarse particles
that are comprised largely of crustal material. For example, the Coarse
Particulate Matter Coalition (CPMC) (2012) and several other industry
commenters recommended that the final decision allow application of a
98th percentile form for the current standard (i.e. with its level of
150 [mu]g/m\3\) in cases where coarse particles consist primarily of
crustal material. Such an approach would allow more yearly exceedances
of the existing standard level than are allowed with the current one-
expected-exceedance form. These industry commenters contended that a
98th percentile form applied in this way would provide appropriate
regulatory relief for areas where the evidence for coarse particle-
related health effects is relatively uncertain.
In reaching her conclusion that the current primary PM10
standard is requisite to protect public health with an adequate margin
of safety, the Administrator considered the degree of public health
protection provided by the current standard as a whole, including all
elements of that standard (i.e., indicator, averaging time, form,
level). As discussed above and in the following section, this
conclusion reflects the Administrator's judgments that (1) the current
standard appropriately provides some measure of protection against
exposures to all thoracic coarse

[[Page 3175]]

particles, regardless of their location, source of origin, or
composition and (2) the current standard appropriately allows lower
ambient concentrations of PM10-2.5 in urban areas, where the
evidence is strongest that thoracic coarse particles are linked to
mortality and morbidity, and higher concentrations in non-urban areas,
where the public health concerns are less certain.
Because the considerations that led to these judgments, and to the
conclusion that the current primary PM10 standard is
requisite to protect public health, took into account the degree of
public health protection provided by the standard as a whole, it would
not be appropriate to consider revising one element of the standard
(e.g., the form, as suggested by commenters in this case) without
considering the extent to which the other elements of the standard
should also be revised. The change in form requested by industry
commenters, without also lowering the level of the standard, would
markedly reduce the public health protection provided against exposures
to thoracic coarse particles.\129\ However, industry commenters have
not presented new evidence or analyses to support their conclusion that
an appropriate degree of public health protection could be achieved by
allowing the use of an alternative form (i.e., 98th percentile) for
some coarse particles, while retaining the other elements of the
current standard. Nor have these commenters presented new evidence or
analyses challenging the basis for the conclusion in the proposal that
the varying amounts of coarse particles allowed in urban versus non-
urban areas under the current 24-hour PM10 standard, based
on the varying levels of PM2.5 present, appropriately
reflect the differences in the strength of evidence regarding coarse
particle effects in urban and non-urban areas. In light of this, EPA
does not believe that a reduction in public health protection, such as
that requested by industry commenters, is warranted.
---------------------------------------------------------------------------

\129\ Based on regression analyses presented in the PA (U.S.
EPA, 2011a, Figures 3-7 and 3-8), PM10 one-expected-
exceedance concentration-equivalent design values were between
approximately 175 and 300 [mu]g/m\3\ at monitoring locations
recording 3-year averages of 98th percentile 24-hour PM10
concentrations around 150 [mu]g/m\3\ (i.e., the level of the current
standard). This suggests that, depending on the location, a 24-hour
PM10 standard with a 98th percentile form in conjunction
with the current level (i.e., as recommended by these commenters)
could be ``generally equivalent'' to a 24-hour PM10
standard with a one-expected-exceedance form and a level as high as
approximately 300 [mu]g/m\3\. Based on this analysis, a 24-hour
PM10 standard with a 98th percentile form and a level of
150 [mu]g/m\3\ would be markedly less health protective than the
current standard.
---------------------------------------------------------------------------

In further considering these comments, it is to be remembered that
epidemiologic studies have not demonstrated that coarse particles of
non-urban origin do not cause health effects, and commenters have not
provided additional evidence on this point. While there are fewer
studies of non-urban coarse particles than of urban coarse particles,
several studies have reported positive and statistically significant
associations between coarse particles of crustal, non-urban origin and
mortality or morbidity (Ostro et al., 2003; Bell et al., 2008; Chan et
al., 2008; Middleton et al., 2008; Perez et al., 2008). These studies
formed part of the basis for the PM Integrated Science Assessment
conclusion that ``recent studies have suggested that PM (both
PM2.5 and PM10-2.5) from crustal, soil or road
dust sources or PM tracers linked to these sources are associated with
cardiovascular effects'' (U.S. EPA, 2009a, p. 2-26). Moreover, crustal
coarse particles may be contaminated with toxic trace elements and
other components from previously deposited fine PM from ubiquitous
sources such as mobile source engine exhaust, as well as by toxic
metals from smelters or other industrial activities, animal waste, or
pesticides (U.S. EPA, 2004, p. 8-344). In the proposal, the
Administrator acknowledged the potential for this type of contamination
to increase the toxicity of coarse particles of crustal, non-urban
origin (77 FR 38960; see also 71 FR 61190).
In suggesting a change in the form of the current standard,
industry commenters also did not address the manifold difficulties
noted above, and in the last review, associated with developing an
indicator that could reliably identify ambient mixes dominated by
particular types of sources of coarse particles. See above and 71 FR
61193. Yet such an indicator would be a prerequisite of the type of
standard these commenters request.
For all of the reasons discussed above, the EPA does not agree with
industry commenters who recommended allowing the application of a 98th
percentile form for the current standard in cases where coarse
particles consist primarily of crustal material.
Some industry commenters contended that the uncertainties and
limitations that precluded a quantitative risk assessment also preclude
revising the PM10 standard. Although the EPA agrees that
there are important uncertainties and limitations in the extent to
which the quantitative relationships between ambient
PM10-2.5 and health outcomes can be characterized in risk
models, the Agency does not agree that such limitations alone preclude
the option of revising a NAAQS. As noted above, the lack of a
quantitative PM10-2.5 risk assessment in the current review
adds uncertainty to conclusions about the extent to which revision of
the current PM10 standard would be expected to improve the
protection of public health, beyond the protection provided by the
current standard. However, the EPA does not agree that such
uncertainties necessarily preclude revision of a NAAQS. Indeed, with
respect to thoracic coarse particles, the DC Circuit noted that
``[a]lthough the evidence of danger from coarse PM is, as the EPA
recognizes, `inconclusive', the agency need not wait for conclusive
findings before regulating a pollutant it reasonably believes may pose
a significant risk to public health.'' 559 F. 3d at 533. Thus, the
Administrator's conclusion that the current 24-hour PM10
standard provides requisite protection of public health relies on her
consideration of the broad body of evidence, rather than solely on the
uncertainties that led to the decision not to conduct a quantitative
assessment of PM10-2.5 health risks.
Commenters representing a number of environmental groups and
medical organizations disagreed with the Administrator's proposal to
retain the current primary PM10 standard. These commenters
generally requested that the EPA revise the PM10 standard to
increase public health protection, consistent with the recommendations
from CASAC.
As discussed above and in the proposal, in reaching provisional
conclusions in the proposal regarding the current standard, the
Administrator carefully considered CASAC's advice and recommendations.
She specifically noted that in making its recommendation on the current
PM10 standard, CASAC did not discuss its approach to
considering the important uncertainties and limitations in the health
evidence, and did not discuss how these uncertainties and limitations
were reflected in its recommendations. Such uncertainties and
limitations contributed to the conclusions in the Integrated Science
Assessment that the PM10-2.5 evidence is only suggestive of
a causal relationship, a conclusion that CASAC endorsed (Samet,
2009e,f). These commenters also did not address the important
uncertainties in the epidemiologic studies on which their comments are
based. Given the importance of these uncertainties and limitations to
the interpretation of the

[[Page 3176]]

evidence, as reflected in the weight of evidence conclusions in the
Integrated Science Assessment and as discussed in the proposal, the
Administrator judges that it is appropriate to consider and account for
them when drawing conclusions about the implications of individual
PM10-2.5 health studies for the current standard. Commenters
have not provided new information that would change the Administrator's
views on the evidence and uncertainties.
In recommending that the PM10 standard be revised, some
commenters supported their conclusions by referencing studies that
evaluated PM10, rather than PM10-2.5. These
commenters contended that ``[t]he most relevant studies to the setting
of a PM10 standard are the thousands of studies that have
reported adverse effects associated with PM10 pollution''
(ALA et al., 2012).
As discussed in the Policy Assessment, the proposal, and above,
since the establishment of the primary PM2.5 standards, the
purpose of the primary PM10 standard has been to protect
against health effects associated with exposures to
PM10-2.5. PM10 is the indicator, not the target
pollutant. With regard to the appropriateness of considering
PM10 health studies for the purpose of reaching conclusions
on a standard meant to protect against exposures to
PM10-2.5, the proposal noted that PM10 includes
both fine and coarse particles, even in locations with the highest
concentrations of PM10-2.5. Therefore, the extent to which
PM10 effect estimates reflect associations with
PM10-2.5 versus PM2.5 can be highly uncertain and
it is often unclear how PM10 health studies should be
interpreted when considering a standard meant to protect against
exposures to PM10-2.5. Given this uncertainty and the
availability of a number of PM10-2.5 health studies in this
review, the Integrated Science Assessment considered
PM10-2.5 studies, but not PM10 studies, when
drawing weight-of-evidence conclusions regarding the coarse
fraction.\130\ In light of the uncertainty in ascribing
PM10-related health effects to the coarse or fine fractions,
indicating that the best evidence for effects associated with exposures
to PM10-2.5 comes from studies evaluating
PM10-2.5 itself, and given CASAC's support for the approach
adopted in the Integrated Science Assessment, which draws weight-of-
evidence conclusions for PM2.5 and PM10-2.5 but
not for PM10 (Samet, 2009f), the EPA continues to conclude
that it is appropriate to focus on PM10-2.5 health studies
when considering the degree of public health protection provided by the
current primary PM10 standard, a standard intended
exclusively to provide protection against exposures to
PM10-2.5.
---------------------------------------------------------------------------

\130\ Although EPA relied in the 1997 review on evidence from
PM10 studies, EPA did so out of necessity (i.e., there
were as yet no reliable studies measuring PM10-2.5). In
the 2006 review, EPA placed primary reliance on epidemiologic
studies measuring or estimating PM10-2.5, although there
were comparatively few such studies. In this review, a larger body
of PM10-2.5 studies are available. EPA regards these
studies as the evidence to be given principal weight in reviewing
the adequacy of the PM10 standard.
---------------------------------------------------------------------------

G. Administrator's Final Decision on the Primary PM10 Standard

In reaching a final decision on the primary PM10
standard, the Administrator takes into account the available scientific
evidence, and the assessment of that evidence, in the Integrated
Science Assessment; the analyses and staff conclusions presented in the
Policy Assessment; the advice and recommendations of CASAC; and public
comments on the proposal. In particular, as in the proposal, the
Administrator places emphasis on her consideration of the following
issues:

Each of these issues is discussed below.
With regard to the first issue, as in the proposal the
Administrator judges that the expanded body of scientific evidence
available in this review provides ample support for a standard that
protects against exposures to all thoracic coarse particles, regardless
of their location or source of origin. There was already ample evidence
for this position in the previous review,\131\ and that evidence has
since increased. Specifically, the Administrator notes that
epidemiological studies have reported positive associations between
PM10-2.5 and mortality or morbidity in a large number of
cities across North America, Europe, and Asia, encompassing a variety
of environments where PM10-2.5 sources and composition are
expected to vary widely. In considering this evidence, the Integrated
Science Assessment concludes that ``many constituents of PM can be
linked with differing health effects'' (U.S. EPA, 2009a, p. 2-26).
Although PM10-2.5 in most of these study areas is of largely
urban origin, the Administrator notes that some recent studies have
also linked mortality and morbidity with relatively high ambient
concentrations of particles of non-urban crustal origin. In considering
these studies, she notes the Integrated Science Assessment's conclusion
that ``PM (both PM2.5 and PM10-2.5) from crustal,
soil or road dust sources or PM tracers linked to these sources are
associated with cardiovascular effects'' (U.S. EPA, 2009a, p. 2-26).
The Administrator likewise notes CASAC's emphatic advice that a
standard remains needed for all types of thoracic coarse PM.\132\ In
light of this body of available evidence reporting PM10-2.5-
associated health effects across different locations with a variety of
sources, the Integrated Science Assessment's conclusions regarding the
links between adverse health effects and PM sources and composition,
and CASAC's advice, the Administrator concludes in the current review
that it is appropriate to maintain a standard that provides some
measure of protection against exposures to all thoracic coarse
particles, regardless of their location, source of origin, or
composition.
---------------------------------------------------------------------------

\131\ The D.C. Circuit agreed. See 559 F. 3d at 532-33.
\132\ Indeed, CASAC recommended making the standard for all
types of thoracic coarse PM more stringent (Samet, 2010d).
---------------------------------------------------------------------------

With regard to the second issue, in considering the appropriateness
of a PM10 indicator for a standard meant to provide such
public health protection, the Administrator notes that the rationale
used in the last review to support the unqualified PM10
indicator remains relevant in the current review. Specifically, as an
initial consideration, she notes that PM10 mass includes
both coarse PM (PM10-2.5) and fine PM (PM2.5). As
a result, the concentration of PM10-2.5 allowed by a
PM10 standard set at a single level declines as the
concentration of PM2.5 increases. At the same time, the
Administrator notes that PM2.5 concentrations tend to be
higher in urban areas than rural areas (U.S. EPA, 2005, p. 2-54, and
Figures 2-23 and 2-24) and, therefore, a PM10 standard will
generally allow lower PM10-2.5 concentrations in urban areas
than in rural areas.
In considering the appropriateness of this variation in allowable
PM10-2.5 concentrations, the Administrator considers the
relative strength of the evidence for health effects associated with
PM10-2.5 of urban origin versus non-urban origin. She
specifically notes that, as discussed in the proposal, the large
majority of the available evidence for

[[Page 3177]]

thoracic coarse particle health effects comes from studies conducted in
locations with sources more typical of urban and industrial areas than
rural areas. While associations with adverse health effects have been
reported in some study locations where PM10-2.5 is largely
non-urban in origin (i.e., in dust storm studies), particle
concentrations in these study areas are typically much higher than
reported in study locations where the PM is of urban origin. Therefore,
the Administrator notes that the strongest evidence for a link between
PM10-2.5 and adverse health impacts, particularly for such a
link at relatively low particle concentrations, comes from studies of
urban or industrial PM10-2.5.
The Administrator also notes that chemical constituents present at
higher levels in urban or industrial areas, including byproducts of
incomplete combustion (e.g. polycyclic aromatic hydrocarbons) emitted
as PM2.5 from motor vehicles as well as metals and other
contaminants emitted from anthropogenic sources, can contaminate
PM10-2.5 (U.S. EPA, 2004, p. 8-344; 71 FR 2665, January 17,
2006). While the Administrator acknowledges the uncertainty expressed
in the Integrated Science Assessment regarding the extent to which
particle composition can be linked to health outcomes based on
available evidence, she also considers the possibility that
PM10-2.5 contaminants typical of urban or industrial areas
could increase the toxicity of thoracic coarse particles in urban
locations.
Given that the large majority of the evidence for
PM10-2.5 toxicity, particularly at relatively low particle
concentrations, comes from study locations where thoracic coarse
particles are of urban origin, and given the possibility that
PM10-2.5 contaminants in urban areas could increase particle
toxicity, the Administrator concludes that it remains appropriate to
maintain a standard that provides some protection in all areas but
targets public health protection to urban locations. Specifically, she
concludes that it is appropriate to maintain a standard that allows
lower ambient concentrations of PM10-2.5 in urban areas,
where the evidence is strongest that thoracic coarse particles are
linked to mortality and morbidity, and higher concentrations in non-
urban areas, where the public health concerns are less certain.
Given all of the above considerations and conclusions, the
Administrator judges that the available evidence supports retaining a
PM10 indicator for a standard that is meant to protect
against exposures to thoracic coarse particles. In reaching this
judgment, she notes that, to the extent a PM10 indicator
results in lower allowable concentrations of thoracic coarse particles
in some areas compared to others, lower concentrations will be allowed
in those locations (i.e., urban or industrial areas) where the science
has shown the strongest evidence of adverse health effects associated
with exposure to thoracic coarse particles and where we have the most
concern regarding PM10-2.5 toxicity. Therefore, the
Administrator concludes that the varying amounts of coarse particles
that are allowed in urban vs. non-urban areas under the 24-hour
PM10 standard, based on the varying levels of
PM2.5 present, appropriately reflect the differences in the
strength of evidence regarding coarse particle effects in urban and
non-urban areas.^{133 134}
---------------------------------------------------------------------------

\133\ As discussed in the proposal, the Administrator recognizes
that this relationship is qualitative. That is, the varying coarse
particle concentrations allowed under the PM10 standard
do not precisely correspond to the variable toxicity of thoracic
coarse particles in different areas (insofar as that variability is
understood). Although currently available information does not allow
any more precise adjustment for relative toxicity, the Administrator
believes the standard will generally ensure that the coarse particle
levels allowed will be lower in urban areas and higher in non-urban
areas. Addressing this qualitative relationship, the DC Circuit held
that ``[i]t is true that the EPA relies on a qualitative analysis to
describe the protection the coarse PM NAAQS will provide. But the
fact that the EPA's analysis is qualitative rather than quantitative
does not undermine its validity as an acceptable rationale for the
EPA's decision.'' 559 F. 3d at 535.
\134\ The D.C. Circuit agreed with similar conclusions in the
last review and held that this rationale reasonably supported use of
an unqualified PM10 indicator for thoracic coarse
particles. American Farm Bureau Federation v. EPA, 559 F. 3d at 535-
36.
---------------------------------------------------------------------------

In reaching this conclusion, the Administrator also notes that, in
their review of the second draft Policy Assessment, CASAC concluded
that ``[w]hile it would be preferable to use an indicator that reflects
the coarse PM directly linked to health risks (PM10-2.5),
CASAC recognizes that there is not yet sufficient data to permit a
change in the indicator from PM10 to one that directly
measures thoracic coarse particles'' (Samet, 2010d, p. ii). Thus,
consistent the considerations presented above and with CASAC advice,
the Administrator concludes that it is appropriate to retain
PM10 as the indicator for thoracic coarse particles.\135\
---------------------------------------------------------------------------

\135\ In addition, CASAC ``vigorously recommends the
implementation of plans for the deployment of a network of
PM10-2.5 sampling systems so that future epidemiological
studies will be able to more thoroughly explore the use of
PM10-2.5 as a more appropriate indicator for thoracic
coarse particles'' (Samet, 2010d, p. 7). Consideration of
alternative indicators (e.g., PM10-2.5) in future reviews
could be informed by additional research, as described in the Policy
Assessment (U.S. EPA, 2011a, section 3.5).
---------------------------------------------------------------------------

With regard to the third issue, in evaluating the degree of public
health protection provided by the current PM10 standard, the
Administrator first notes that when the available PM10-2.5
scientific evidence and its associated uncertainties were considered,
the Integrated Science Assessment concluded that the evidence is
suggestive of a causal relationship between short-term
PM10-2.5 exposures and mortality, cardiovascular effects,
and respiratory effects. As discussed above and in more detail in the
Integrated Science Assessment (U.S. EPA, 2009a, section 1.5), a
suggestive determination is made when the ``[e]vidence is suggestive of
a causal relationship with relevant pollutant exposures, but is limited
because chance, bias and confounding cannot be ruled out.'' In
contrast, the Administrator notes that she is strengthening the annual
fine particle standard based on a body of scientific evidence judged
sufficient to conclude that a causal relationship exists (i.e.,
mortality, cardiovascular effects) or is likely to exist (i.e.,
respiratory effects). The suggestive judgment for PM10-2.5
reflects the greater degree of uncertainty associated with this body of
evidence, as discussed above and in more detail in the proposal, and as
summarized below.
The Administrator notes that the important uncertainties and
limitations associated with the scientific evidence and air quality
information raise questions as to whether public health benefits would
be achieved by revising the existing PM10 standard. Such
uncertainties and limitations include the following:

(1) While PM10-2.5 effect estimates reported for
mortality and morbidity were generally positive, most were not
statistically significant, even in single-pollutant models. This
includes effect estimates reported in some study locations with
PM10 concentrations above those allowed by the current
24-hour PM10 standard.
(2) The number of epidemiological studies that have employed co-
pollutant models to address the potential for confounding,
particularly by PM2.5, remains limited. Therefore, the
extent to which PM10-2.5 itself, rather than one or more
co-pollutants, contributes to reported health effects remains
uncertain.
(3) Only a limited number of experimental studies provide
support for the associations reported in epidemiological studies,
resulting in further uncertainty regarding the plausibility of the
associations between PM10-2.5 and mortality and morbidity
reported in epidemiological studies.

[[Page 3178]]

(4) Limitations in PM10-2.5 monitoring data and the
different approaches used to estimate PM10-2.5
concentrations across epidemiological studies result in uncertainty
in the ambient PM10-2.5 concentrations at which the
reported effects occur, increasing uncertainty in estimates of the
extent to which changes in ambient PM10-2.5
concentrations would likely impact public health.
(5) The lack of a quantitative PM10-2.5 risk
assessment further contributes to uncertainty regarding the extent
to which any revisions to the current PM10 standard would
be expected to improve the protection of public health, beyond the
protection provided by the current standard (see section III.B.5
above).
(6) The chemical and biological composition of
PM10-2.5, and the effects associated with the various
components, remains uncertain. Without more information on the
chemical speciation of PM10-2.5, the apparent variability
in associations across locations is difficult to characterize.

In considering these uncertainties and limitations, the
Administrator notes in particular the considerable degree of
uncertainty in the extent to which health effects reported in
epidemiological studies are due to PM10-2.5 itself, as
opposed to one or more co-occurring pollutants. As discussed above,
this uncertainty reflects the fact that there are a relatively small
number of PM10-2.5 studies that have evaluated co-pollutant
models, particularly co-pollutant models that have included
PM2.5, and a very limited body of controlled human exposure
evidence supporting the plausibility of a causal relationship between
PM10-2.5 and mortality and morbidity at ambient
concentrations. The Administrator notes that these important
limitations in the overall body of health evidence introduce
uncertainty into the interpretation of individual epidemiological
studies, particularly those studies reporting associations with
PM10-2.5 that are not statistically significant. Given this,
the Administrator reaches the conclusion that it is appropriate to
place relatively little weight on epidemiological studies reporting
associations with PM10-2.5 that are not statistically
significant in single-pollutant and/or co-pollutant models.\136\
---------------------------------------------------------------------------

\136\ The Administrator acknowledges that this approach to
interpreting the evidence differs in emphasis from the approach she
has adopted for the evidence relating to PM2.5. As
discussed above in section III.E.4, for fine particles the
Administrator has considered not only whether study results are
statistically significant (or remain so after application of co-
pollutant models), but she also places emphasis on the overall
pattern of results across the epidemiological literature. This
includes giving some credence to studies that reported statistically
non-significant associations. This difference in emphasis stems from
the much stronger overall body of evidence available for fine
particles, compared to coarse particles. As discussed above, when
the available PM2.5 scientific evidence and its
associated uncertainties were considered, the Integrated Science
Assessment concluded that the evidence was sufficient to conclude
that causal relationships exist with mortality and cardiovascular
effects, and that a causal relationship is likely to exist with
respiratory effects. In contrast, the Integrated Science Assessment
concluded that the evidence is suggestive of a causal relationship
between short-term PM10-2.5 exposures and mortality,
cardiovascular effects, and respiratory effects. A suggestive
determination is made when the ``[e]vidence is suggestive of a
causal relationship with relevant pollutant exposures, but is
limited because chance, bias and confounding cannot be ruled out''
(U.S. EPA, 2009a, section 1.5). The suggestive judgment for
PM10-2.5 reflects the greater degree of uncertainty
associated with this body of evidence.
---------------------------------------------------------------------------

With regard to this conclusion, the Administrator notes that, for
single-city mortality studies conducted in the United States where
ambient PM10 concentration data were available for
comparison to the current standard, positive and statistically
significant PM10-2.5 effect estimates were only reported in
study locations that would likely have violated the current
PM10 standard during the study period (U.S. EPA, 2011a,
Figure 3-2). In U.S. study locations that would likely have met the
current standard, PM10-2.5 effect estimates for mortality
were positive, but not statistically significant (U.S. EPA, 2011a,
Figure 3-2). In considering U.S. study locations where single-city
morbidity studies were conducted, and which would likely have met the
current PM10 standard during the study period, the
Administrator notes that PM10-2.5 effect estimates were both
positive and negative, with most not statistically significant (U.S.
EPA, 2011a, Figure 3-3).
In addition, in considering single-city analyses for the locations
evaluated in a large U.S. multi-city mortality study (Zanobetti and
Schwartz, 2009), the Administrator notes that associations in most of
the study locations were not statistically significant and that this
was the only study to estimate ambient PM10-2.5
concentrations as the difference between county-wide PM10
and PM2.5 mass. As discussed in the proposal, the
Administrator notes that it is not clear how computed
PM10-2.5 measurements, such as those used by Zanobetti and
Schwartz (2009), compare with the PM10-2.5 concentrations
obtained in other studies either by direct measurement by calculating
the difference using co-located samplers (U.S. EPA, 2009a, section
6.5.2.3). For these reasons, as in the proposal, the Administrator
notes that there is considerable uncertainty in interpreting the
associations, and especially the concentrations at which such
associations may have occurred, in these single-city analyses.
The Administrator acknowledges that an approach to considering the
available scientific evidence and air quality information that
emphasizes the above considerations differs from the approach taken by
CASAC. Specifically, CASAC placed a substantial amount of weight on
individual studies, particularly those reporting positive health
effects associations in locations that met the current PM10
standard during the study period. In emphasizing these studies, as well
as the limited number of supporting studies that have evaluated co-
pollutant models and the small number of supporting experimental
studies, CASAC concluded that ``the current data, while limited, is
sufficient to call into question the level of protection afforded the
American people by the current standard'' (Samet, 2010d, p. 7) and
recommended revising the current PM10 standard (Samet,
2010d).
The Administrator has carefully considered CASAC's advice and
recommendations. She notes that in making its recommendation on the
current PM10 standard, CASAC did not discuss its approach to
considering the important uncertainties and limitations in the health
evidence, and did not discuss how these uncertainties and limitations
are reflected in its recommendation. As discussed above, such
uncertainties and limitations contributed to the conclusions in the
Integrated Science Assessment that the PM10-2.5 evidence is
only suggestive of a causal relationship, a conclusion that CASAC
endorsed (Samet, 2009e,f). Given the importance of these uncertainties
and limitations to the interpretation of the evidence, as reflected in
the weight of evidence conclusions in the Integrated Science Assessment
and as discussed above, the Administrator judges that it is appropriate
to consider and account for them when drawing conclusions about the
potential implications of individual PM10-2.5 health studies
for the current standard.
In light of the above approach to considering the scientific
evidence, air quality information, and associated uncertainties, the
Administrator reaches the following conclusions:

(1) When viewed as a whole the available evidence and
information suggests that the degree of public health protection
provided against short-term exposures to PM10-2.5 should
be maintained but does not need to be increased beyond that provided
by the current PM10 standard. This conclusion emphasizes
the important uncertainties and limitations associated with the
overall body

[[Page 3179]]

of health evidence and air quality information for
PM10-2.5, as discussed above and as reflected in the
Integrated Science Assessment weight-of-evidence conclusions; that
PM10-2.5 effect estimates for the most serious health
effect, mortality, were not statistically significant in U.S.
locations that met the current PM10 standard and where
coarse particle concentrations were either directly measured or
estimated based on co-located samplers; and that PM10-2.5
effect estimates for morbidity endpoints were both positive and
negative in locations that met the current standard, with most not
statistically significant.\137\
---------------------------------------------------------------------------

\137\ This is not to say that the EPA could not adopt or revise
a standard for a pollutant for which the evidence is suggestive of a
causal relationship. Indeed, with respect to thoracic coarse
particles itself, the DC Circuit noted that ``[a]lthough the
evidence of danger from coarse PM is, as the EPA recognizes,
`inconclusive', the agency need not wait for conclusive findings
before regulating a pollutant it reasonably believes may pose a
significant risk to public health.'' American Farm Bureau Federation
v EPA 559 F. 3d at 533. As explained in the text above, it is the
Administrator's judgment that significant uncertainties presented by
the evidence and information before her in this review, both as to
causality and as to concentrations at which effects may be
occurring, best support a decision to retain rather than revise the
current primary 24-hour PM10 standard.
---------------------------------------------------------------------------

(2) The degree of public health protection provided by the
current standard is not greater than warranted. This conclusion
notes that positive and statistically significant associations with
mortality were reported in single-city U.S. study locations likely
to have violated the current PM10 standard.\138\
---------------------------------------------------------------------------

\138\ There are similarities with the conclusions drawn by the
Administrator in the last review. There, the Administrator concluded
that there was no basis for concluding that the degree of protection
afforded by the current PM10 standards in urban areas is
greater than warranted, since potential mortality effects have been
associated with air quality levels not allowed by the current 24-
hour standard, but have not been associated with air quality levels
that would generally meet that standard, and morbidity effects have
been associated with air quality levels that exceeded the current
24-hour standard only a few times. 71 FR 61202. In addition, the
Administrator concluded that there was a high degree of uncertainty
in the relevant population exposures implied by the morbidity
studies suggesting that there is little basis for concluding that a
greater degree of protection is warranted. Id. The D.C. Circuit in
American Farm Bureau Federation v EPA explicitly endorsed this
reasoning. 559 F. 3d at 534.

In reaching these conclusions, the Administrator notes that the
Policy Assessment also discussed the potential for a revised
PM10 standard (i.e., with a revised form and level) to be
``generally equivalent'' to the current standard, but to better target
public health protection to locations where there is greater concern
regarding PM10-2.5-associated health effects (U.S. EPA,
2011a, sections 3.3.3 and 3.3.4).\139\ In considering such a potential
revised standard, the Policy Assessment discusses the large amount of
variability in PM10 air quality correlations across
monitoring locations and over time (U.S. EPA, 2011a, Figure 3-7) and
the regional variability in the relative degree of public health
protection that could be provided by the current and potential
alternative standards (U.S. EPA, 2011a, Table 3-2). In light of this
variability, the Administrator notes the Policy Assessment conclusion
that no single revised PM10 standard (i.e., with a revised
form and level) would provide public health protection equivalent to
that provided by the current standard, consistently over time and
across locations (U.S. EPA, 2011a, section 3.3.4). That is, a revised
standard, even one that is meant to be ``generally equivalent'' to the
current PM10 standard, could increase protection in some
locations while decreasing protection in other locations.
---------------------------------------------------------------------------

\139\ As discussed in detail above (section IV.C.2.d) and in the
Policy Assessment (U.S. EPA, 2011a, sections 3.3.3 and 3.3.4), a
revised standard that is generally equivalent to the current
PM10 standard could provide a degree of public health
protection that is similar to the degree of protection provided by
the current standard, across the United States as a whole. However,
compared to the current PM10 standard, such a generally
equivalent standard would change the degree of public health
protection provided in some specific areas, providing increased
protection in some locations and decreased protection in other
locations.
---------------------------------------------------------------------------

In considering the appropriateness of revising the current
PM10 standard in this way, the Administrator notes the
following:

(1) As discussed above, positive PM10-2.5 effect
estimates for mortality were not statistically significant in U.S.
locations that met the current PM10 standard and where
coarse particle concentrations were either directly measured or
estimated based on co-located samplers, while positive and
statistically significant associations with mortality were reported
in locations likely to have violated the current PM10
standard.
(2) Also as discussed above, effect estimates for morbidity
endpoints in locations that met the current standard were both
positive and negative, with most not statistically significant.
(3) Important uncertainties and limitations associated with the
overall body of health evidence and air quality information for
PM10-2.5, as discussed above and as reflected in the
Integrated Science Assessment weight-of-evidence conclusions, call
into question the extent to which the type of quantified and refined
targeting of public health protection envisioned under a revised
standard could be reliably accomplished.

Given all of the above considerations, the Administrator notes that
there is a large amount of uncertainty in the extent to which public
health would be improved by changing the locations to which the
PM10 standard targets protection. Therefore, she reaches the
conclusion that the current PM10 standard should not be
revised in order to change that targeting of protection.
In considering all of the above, including the scientific evidence,
the air quality information, the associated uncertainties, CASAC's
advice, and public comments received on the proposed rule, the
Administrator reaches the conclusion in the current review that the
existing 24-hour PM10 standard, with its one-expected
exceedance form and a level of 150 [mu]g/m\3\, is requisite (i.e.,
neither more protective nor less protective than necessary) to protect
public health with an adequate margin of safety against effects that
have been associated with PM10-2.5. In light of this
conclusion, with this rule the Administrator retains the current
PM10 standard.

V. Communication of Public Health Information

Sections 319(a)(1) and (3) of the CAA require the EPA to establish
a uniform air quality index for reporting of air quality. These
sections specifically direct the Administrator to ``promulgate
regulations establishing an air quality monitoring system throughout
the United States which utilizes uniform air quality monitoring
criteria and methodology and measures such air quality according to a
uniform air quality index'' and ``provides for daily analysis and
reporting of air quality based upon such uniform air quality index * *
*'' In 1979, the EPA established requirements for index reporting (44
FR 27598, May 10, 1979). The requirement for State and local agencies
to report the AQI appears in 40 CFR 58.50, and the specific
requirements (e.g., what to report, how to report, reporting frequency,
calculations) are in appendix G to 40 CFR part 58.
Information on the public health implications of ambient
concentrations of criteria pollutants is currently made available
primarily by AQI reporting through EPA's AIRNow Web site.\140\ The
current AQI has been in use since its inception in 1999.\141\ It
provides accurate, timely, and easily understandable information about
daily levels of pollution (40 CFR 58.50). The AQI establishes a
nationally uniform system of indexing pollution levels for ozone,
carbon monoxide, nitrogen

[[Page 3180]]

dioxide, PM, and sulfur dioxide. The AQI is also recognized
internationally as a proven tool to effectively communicate air quality
information to the public.
---------------------------------------------------------------------------

\140\ See http://www.airnow.gov/.
\141\ In 1976, the EPA established a nationally uniform air
quality index, then called the Pollutant Standard Index (PSI), for
use by State and local agencies on a voluntary basis (41 FR 37660,
September 7, 1976). In August 1999, the EPA adopted revisions to
this air quality index (64 FR 42530, August 4, 1999) and renamed the
index the AQI.
---------------------------------------------------------------------------

The AQI converts pollutant concentrations in a community's air to a
number on a scale from 0 to 500. Reported AQI values enable the public
to know whether air pollution levels in a particular location are
characterized as good (0-50), moderate (51-100), unhealthy for
sensitive groups (101- 150), unhealthy (151-200), very unhealthy (201-
300), or hazardous (301-500). The AQI index value of 100 typically
corresponds to the level of the short-term (e.g., daily or hourly
standard) NAAQS for each pollutant. Below an index value of 100, an
intermediate value of 50 was defined either as the level of the annual
standard if an annual standard has been established (e.g.,
PM2.5, nitrogen dioxide), or as a concentration equal to
one-half the value of the short-term standard used to define an index
value of 100 (e.g., carbon monoxide). An AQI value greater than 100
means that a pollutant is in one of the unhealthy categories (i.e.,
unhealthy for sensitive groups, unhealthy, very unhealthy, or
hazardous) on a given day. An AQI value at or below 100 means that a
pollutant concentration is in one of the satisfactory categories (i.e.,
moderate or good). The underlying health information that supports the
NAAQS review also supports the selection of the AQI ``breakpoints''--
the ambient concentrations that delineate the various AQI categories
for each pollutant.
Historically, state and local agencies have primarily used the AQI
to provide general information to the public about air quality and its
relationship to public health. For more than a decade, many states and
local agencies, as well as the EPA and other Federal agencies, have
been developing new and innovative programs and initiatives to provide
more information to the public in a more timely way. These initiatives,
including air quality forecasting, real-time data reporting through the
AirNow Web site, and state and local air quality action day programs,
can serve to provide useful, up-to-date, and timely information to the
public about air pollution and its effects. Such information will help
individuals take actions to avoid or to reduce exposures to ambient
pollution at levels of concern to them. Thus, these programs have
significantly broadened the ways in which state and local agencies can
meet the nationally uniform AQI reporting requirements and contribute
to state and local efforts to provide community health protection.
With respect to an AQI value of 50, the historical approach is to
set it at the same level of the annual primary standard, if there is
one. This is consistent with the previous AQI sub-index for
PM2.5, in which the AQI value of 50 was set at 15 [micro]g/
m\3\ in 1999, consistent with the level of the annual PM2.5
standard at that time. In recognition of the proposed change to the
annual PM2.5 standard summarized in section III.F of the
proposal, the EPA proposed a conforming change to the PM2.5
sub-index of the AQI to be consistent with the proposed change to the
annual standard. As discussed below, no state or local agencies, or
their organizations (e.g., NACAA), that commented on the proposed
changes to the AQI disagreed with our proposed approach. Based on these
comments, the EPA continues to see no basis for deviating from this
approach in this review. Thus, the EPA is taking final action to set an
AQI value of 50 at 12.0 [mu]g/m\3\, 24-hour average, consistent with
the final decision on the annual PM2.5 standard level
(section III.F).
With respect to an AQI value of 100, which is the basis for
advisories to individuals in sensitive groups, in the proposal we
described two general approaches that could be used to select the
associated PM2.5 level. By far the most common approach,
which has been used with all of the other sub-indices, is to set an AQI
value of 100 at the same level as the short-term standard. In the
proposal, the EPA recognized that some state and local air quality
agencies have expressed a strong preference that the Agency set an AQI
value of 100 equal to any short-term standard (77 FR 38964). These
agencies typically express the view that this linkage is useful for the
purpose of communicating with the public about the standard, as well as
providing consistent messages about the health impacts associated with
daily air quality. The EPA proposed to use this approach to set the AQI
value of 100 at 35 [mu]g/m\3\, 24-hour average, consistent with the
proposed decision to retain the current 24-hour PM2.5
standard. Id.
An alternative approach discussed in the proposal (77 FR 38964),
was to directly evaluate the health effects evidence to select the
level for an AQI value of 100. This was the approach used in the 1999
rulemaking to set the AQI value of 100 at a level of 40 [mu]g/m\3\, 24-
hour average,\142\ when the 24-hour standard level was 65 [mu]g/m\3\.
This alternative approach was used in the case of the PM2.5
sub-index, because the annual and 24-hour PM2.5 standards
set in 1997 were designed to work together, and the intended degree of
health protection against short-term risks was not defined by the 24-
hour standard alone, but rather by the combination of the two standards
working in concert. Indeed, at that time, the 24-hour standard was set
to provide supplemental protection relative to the principal protection
provided by the annual standard. In the proposal, the EPA solicited
comment on this alternative approach in recognition that, as proposed,
the 24-hour PM2.5 standard is intended to continue to
provide supplemental protection against effects associated with short-
term exposures of PM2.5 by working in conjunction with the
annual standard to reduce 24-hour exposures to PM2.5. The
EPA recognized that in the past, some state and local air quality
agencies have expressed support for this alternative approach. Using
this alternative approach could have resulted in consideration of a
lower level for an AQI value of 100, based on the discussion of the
health information pertaining to the level of the 24-hour standard in
section III.E.4 of the proposal. The EPA encouraged state and local air
quality agencies to comment on both the approach and the level at which
to set an AQI value of 100 together with any supporting rationale. Of
the state or local agencies, or their organizations (e.g., NACAA), that
commented on the proposed changes to the AQI, only one organization,
NESCAUM, expressed some support for this approach. In its comments,
NESCAUM expressed support for a 24-hour standard set at 30 [mu]g/m\3\,
24-hour average. NESCAUM also expressed the view that EPA should
carefully consider how to set the breakpoint for an AQI value of 100.
NESCAUM expressed the view that if the EPA were to keep the 24-hour
PM2.5 standard at 35 [mu]g/m\3\, the annual standard would
be controlling, and a 24-hour breakpoint at that level (35 [mu]g/m\3\)
would not be very effective for the purposes of public health
messaging. However, other agencies, such as Georgia Department of
Natural Resources (Georgia DNR), expressed the view that linkage
between the short-term standard and the AQI of 100 is useful for the
purpose of communicating with the public about the standard as well as
providing consistent messages about the health

[[Page 3181]]

impacts associated with the daily air quality. Based on these comments,
the EPA sees no basis for deviating from the approach proposed in this
review. Thus, the EPA is taking final action to set an AQI value of 100
at 35 [mu]g/m\3\, 24-hour average, consistent with the final decision
on the 24-hour PM2.5 standard level (section III.F).
---------------------------------------------------------------------------

\142\ Currently, we are cautioning members of sensitive groups
at the AQI value of 100 at 35 [mu]g/m\3\, 24-hour average,
consistent with more recent guidance from the EPA with regard to the
development of State emergency episode contingency plans (Harnett,
2009, Attachment B).
---------------------------------------------------------------------------

With respect to an AQI value of 150, this level is based upon the
same health effects information that informs the selection of the level
of the 24-hour standard and the AQI value of 100. The AQI value of 150
was set in the 1999 rulemaking at a level of 65 [mu]g/m\3\, 24-hour
average. In considering what level to propose for an AQI value of 150,
we stated the view that the health effects evidence indicates that the
level of 55 [mu]g/m\3\, 24-hour average, is appropriate to use \143\ in
conjunction with an AQI value of 100 set at the level of 35 [mu]g/m\3\.
The Agency's approach to selecting the levels at which to set the AQI
values of 100 and 150 inherently recognizes that the epidemiological
evidence upon which these decisions are based provides no evidence of
discernible thresholds, below which effects do not occur in either
sensitive groups or in the general population, at which to set these
two breakpoints. Therefore, the EPA concluded the use of a proportional
adjustment would be appropriate. Commenters did not comment on this
proposed approach to revising the AQI value of 150; thus, the EPA is
taking final action to set an AQI value of 150 at 55 [mu]g/m\3\, 24-
hour average.
---------------------------------------------------------------------------

\143\ We note that this level is consistent with the level
recommended in the more recent EPA guidance (Harnett, 2009,
Attachment B), which is in use by many State and local agencies.
---------------------------------------------------------------------------

Based on the air quality and health considerations discussed in
section V of the proposal, the EPA concluded that it was appropriate to
propose to retain the current level of 500 [mu]g/m\3\, 24-hour average,
for the AQI value of 500. In addition, the EPA solicited comment on
alternative levels and approaches to setting a level for the AQI value
of 500, as well as supporting information and rationales for such
alternative levels. The EPA also solicited any additional information,
data, research or analyses that may be useful to inform a final
decision on the appropriate level to set the AQI value of 500.
Receiving no information with which to inform alternative approaches to
setting an AQI value of 500, the EPA is taking final action to retain
the current level of 500 [mu]g/m\3\, 24-hour average, for the AQI value
of 500.
For the intermediate breakpoints in the AQI between the values of
150 and 500, the EPA proposed PM2.5 concentrations that
generally reflected a linear relationship between increasing index
values and increasing PM2.5 values (77 FR 38965). The
available scientific evidence of health effects related to population
exposures to PM2.5 concentrations between the level of the
24-hour standard and an AQI value of 500 suggested a continuum of
effects in this range, with increasing PM2.5 concentrations
being associated with increasingly larger numbers of people likely to
experience such effects. The generally linear relationship between AQI
values and PM2.5 concentrations in this range is consistent
with the health evidence. This also is consistent with the Agency's
practice of setting breakpoints in symmetrical fashion where health
effects information does not suggest particular levels.
Table 2 below summarizes the finalized breakpoints for the
PM2.5 sub-index.\144\ Table 2 shows the intermediate
breakpoints for AQI values of 200, 300 and 400 based on a linear
interpolation between the proposed levels for AQI values of 150 and
500. If a different level were to be set for an AQI value of 150 or
500, intermediate levels would be calculated based on a linear
relationship between the selected levels for AQI values of 150 and 500.
---------------------------------------------------------------------------

\144\ As discussed in section VII.C below, the EPA is also
updating the data handling procedures for reporting the AQI and
corresponding updates for other AQI-sub-indices presented in Table 2
of appendix G of 40 CFR part 58.

Table 2--Breakpoints for PM2.5 Sub-Index
----------------------------------------------------------------------------------------------------------------
Proposed breakpoints
AQI category Index values ([mu]g/m\3\, 24-hour
average)
----------------------------------------------------------------------------------------------------------------
Good................................................................ 0-50 0.0-(12.0)
Moderate............................................................ 51-100 (12.1)-35.4
Unhealthy for Sensitive Groups...................................... 101-150 35.5-55.4
Unhealthy........................................................... 151-200 55.5-150.4
Very Unhealthy...................................................... 201-300 150.5-250.4
Hazardous........................................................... 301-400 250.5-350.4
401-500 350.5-500.4
----------------------------------------------------------------------------------------------------------------

In retaining the 500 level for the AQI as described above, we note
that the EPA is not establishing a Significant Harm Level (SHL) for
PM2.5. The SHL is an important part of air pollution
Emergency Episode Plans, which are required for certain areas by CAA
section 110(a)(2)(G) and associated regulations at 40 CFR 51.150, under
the Prevention of Air Pollution Emergency Episodes program. The Agency
believes that air quality responses established through an Emergency
Episode Plan should be developed through a collaborative process
working with State and Tribal air quality, forestry and agricultural
agencies, Federal land management agencies, private land managers and
the public. Therefore, if in future rulemaking the EPA proposes
revisions to the Prevention of Air Pollution Emergency Episodes
program, the proposal will include a SHL for PM2.5 that is
developed in collaboration with these organizations. As discussed in
the 1999 Air Quality Index Reporting Rule (64 FR 42530), if a future
rulemaking results in a SHL that is different from the 500 value of the
AQI for PM2.5, the AQI will be revised accordingly.
The EPA also received more general comments on AQI reporting,
comments that did not pertain to setting specific breakpoints. One set
of commenters (e.g., API and UARG), expressed the view that changes to
the AQI are not appropriate. They noted that air quality is getting
better, and in fact is better than when EPA established the AQI. These
commenters stated that the proposed changes to the annual standard and
the AQI would mean that the public would hear less often that air
quality is good, and thereby would receive apparently inconsistent or
misleading messages that air quality is

[[Page 3182]]

worse. The AQI has been revised several times in conjunction with
revisions to the standards. State and local air quality agencies and
organizations are proficient at communicating with the public about the
reasons for changes to the AQI. Therefore, the EPA strongly disagrees
with these commenters that the public will receive inconsistent or
misleading messages. Recognizing the importance of the AQI as a
communication tool that allows the public to take exposure reduction
measures when air quality may pose health risks, the EPA agrees with
state and local air quality agencies and organizations that favored
revising the AQI at the same time as the primary standard.
A few state and local air quality agencies and organizations
recommended against using near-roadway PM2.5 monitors for
AQI reporting. In support of this comment, they expressed the following
views, that near-roadway monitors are source-oriented, represent micro-
scale conditions, and the agencies don't have experience using them for
AQI reporting. The EPA disagrees with the comment in that these
monitors will be sited at existing near-road stations sited to be
representative of area-wide PM2.5 concentrations indicative
of general population exposure. Accordingly, data from these near-road
monitors should be included in the AQI since they provide information
about PM2.5 levels that millions of people, who work, live
and go to school near busy roadways, are exposed to. The stations are
representative of somewhat elevated concentrations in near-road
environments, but since these stations represent many such locations
throughout a metropolitan area, they are appropriate for characterizing
exposure in typical portions of major urban areas. The EPA is committed
to helping air quality agencies develop appropriate ways to report
PM2.5 levels from these monitors using the AQI.

VI. Rationale for Final Decisions on the Secondary PM Standards

This section presents the Administrator's final decisions regarding
the need to revise the current suite of secondary PM2.5 and
PM10 standards to address visibility impairment and other
welfare effects considered in this review. Specifically, this section
describes the Administrator's final decision to retain the current
suite of secondary PM standards to address PM-related visibility
impairment as well as other PM-related welfare effects, including
ecological effects, effects on materials, and climate impacts. This
suite of standards includes an annual PM2.5 standard of 15
[mu]g/m\3\, a 24-hour PM2.5 standard of 35 [mu]g/m\3\, and a
24-hour PM10 standard of 150 [mu]g/m\3\. The Administrator
is revising only the form of the secondary annual PM2.5
standard to remove the option for spatial averaging consistent with
this change to the primary annual PM2.5 standard. Contrary
to what was proposed, the Administrator has decided not to establish a
distinct standard to address PM-related visibility impairment. The
rationale for this decision is presented below.
The Administrator's final decisions on the secondary standards are
based on a thorough review of the latest scientific information
published through mid-2009 on welfare effects associated with fine and
coarse particles in the ambient air, as presented in the Integrated
Science Assessment. The final decisions also take into account: (1)
Staff assessments of the most policy-relevant information presented and
assessed in the Integrated Science Assessment and staff analyses of air
quality and visibility effects presented in the Visibility Assessment
and the Policy Assessment, upon which staff conclusions regarding
appropriate considerations in this review are based; (2) CASAC advice
and recommendations, as reflected in discussions of drafts of the
Integrated Science Assessment, Visibility Assessment, and Policy
Assessment at public meetings, in separate written comments, and in
CASAC's letters to the Administrator; (3) the multiple rounds of public
comments received during the development of these documents, both in
connection with CASAC meetings and separately; and (4) public comments
received on the proposal.
In particular, this section presents background information on the
EPA's previous and current reviews of the secondary PM standards
(section VI.A), a summary of the proposed decisions regarding the
secondary PM standards (section VI.B), a discussion of significant
public comments received on those proposed decisions (section VI.C),
and the Administrator's final decisions on the secondary PM standards
(section VI.D).

A. Background

The current suite of secondary PM standards is identical to the
suite of primary PM standards set in 2006, including 24-hour and annual
PM2.5 standards and a 24-hour PM10 standard. The
current secondary PM2.5 standards are intended to provide
protection from PM-related visibility impairment, whereas the entire
suite of secondary PM standards is intended to provide protection from
other PM-related effects on public welfare, including effects on
sensitive ecosystems, materials damage and soiling, and climatic and
radiative processes.
The approach used for reviewing the current suite of secondary PM
standards built upon and broadened the approaches used in previous PM
NAAQS reviews. The following discussion focuses particularly on the
current secondary PM2.5 standards related to visibility
impairment and provides a summary of the approaches used to review and
establish secondary PM2.5 standards in the last two reviews
(section VI.A.1); judicial review of the 2006 standards that resulted
in the remand of the secondary annual and 24-hour PM2.5
NAAQS to the EPA (section VI.A.2); and the approach used in this review
for evaluating the secondary PM2.5 standards (section
VI.A.3).
1. Approaches Used in Previous Reviews
The original secondary PM2.5 standards were established
in 1997, and a revision to the 24-hour standard was made in 2006. The
approaches used in making final decisions on secondary standards in
those reviews, as well as the current review, utilized different ways
to consider the underlying body of scientific evidence. They also
reflected an evolution in EPA's understanding of the nature of the
effect on public welfare from PM-related visibility impairment, from an
approach that focused only on Federal Class I area visibility impacts
to a more multifaceted approach that also considered PM-related impacts
on visibility in non-Federal Class I areas, such as in urban areas.
This evolution occurred in conjunction with the expansion of available
PM data and information from visibility-related studies of public
perception, valuation, and personal comfort and well-being.
In 1997, the EPA revised the PM NAAQS in part by establishing new
identical primary and secondary PM2.5 standards. In revising
the secondary standards, the EPA recognized that PM produces adverse
effects on visibility and that impairment of visibility was being
experienced throughout the U.S., in multi-state regions, urban areas,
and remote mandatory Federal Class I areas alike. However, in
considering an appropriate level for a secondary standard to address
adverse effects of PM2.5 on visibility, the EPA concluded
that the determination of a single national level was complicated by
important regional differences influenced by factors such as

[[Page 3183]]

background and current levels of PM2.5, composition of
PM2.5, and average relative humidity. Variations in these
factors across regions could thus result in situations where attaining
an appropriately protective concentration of fine particles in one
region might or might not provide adequate protection in a different
region. The EPA also determined that there was insufficient information
at that time to establish a level for a national secondary standard
that would represent a threshold above which visibility conditions
would always be adverse and below which visibility conditions would
always be acceptable.
Based on an assessment of the potential visibility improvements
that would result from reaching attainment with the new primary
standards for PM2.5, the EPA concluded that attainment of
the annual and 24-hour PM2.5 primary standards would lead to
visibility improvements in the eastern U.S. at both urban and regional
scales, but little or no change in the western U.S., except in and near
certain urban areas.
The EPA also considered the potential effectiveness of a regional
haze program, required by sections 169A and 169B of the CAA \145\ to
address those effects of PM on visibility that would not be addressed
through attainment of the primary PM2.5 standards. The
regional haze program would be designed to address the widespread,
regionally uniform type of haze caused by a multitude of sources. The
structure and requirements of sections 169A and 169B of the CAA provide
for visibility protection programs that can be more responsive to the
factors contributing to regional differences in visibility than can
programs addressing the kinds of nationally applicable secondary NAAQS
considered in the 1997 review. The regional haze visibility goal is
more protective than a secondary NAAQS since the goal is to eliminate
any anthropogenic impairment rather than to provide a level of
protection from visibility impairment that is requisite to protect the
public welfare. Thus, an important factor considered in the 1997 review
was whether a regional haze program, in conjunction with secondary
standards set identical to the suite of PM2.5 primary
standards, would provide appropriate protection for visibility in non-
Federal Class I areas. The EPA concluded that the two programs and
associated control strategies should provide such protection due to the
regional approaches needed to manage emissions of pollutants that
impair visibility in many of these areas.
---------------------------------------------------------------------------

\145\ In 1977, Congress established as a national goal ``the
prevention of any future, and the remedying of any existing,
impairment of visibility in mandatory Federal Class I areas which
impairment results from manmade air pollution,'' section 169A(a)(1)
of the CAA. The EPA is required by section 169A(a)(4) of the CAA to
promulgate regulations to ensure that ``reasonable progress'' is
achieved toward meeting the national goal.
---------------------------------------------------------------------------

For these reasons, in 1997 the EPA concluded that a national
regional haze program, combined with a nationally applicable level of
protection achieved through secondary PM2.5 standards set
identical to the primary PM2.5 standards, would be more
effective for addressing regional variations in the adverse effects of
PM2.5 on visibility than would be national secondary
standards for PM with levels lower than the primary PM2.5
standards. The EPA further recognized that people living in certain
urban areas may place a high value on unique scenic resources in or
near these areas and as a result might experience visibility problems
attributable to sources that would not necessarily be addressed by the
combined effects of a regional haze program and PM2.5
secondary standards. The EPA concluded that in such cases, state or
local regulatory approaches, such as past action in Colorado to
establish a local visibility standard for the City of Denver, would be
more appropriate and effective in addressing these special situations
because of the localized and unique characteristics of the problems
involved. Visibility in an urban area located near a mandatory Federal
Class I area could also be improved through state implementation of the
then-current visibility regulations, by which emission limitations can
be imposed on a source or group of sources found to be contributing to
``reasonably attributable'' impairment in the mandatory Federal Class I
area.
Based on these considerations, in 1997 the EPA set secondary
PM2.5 standards identical to the primary PM2.5
standards, that would work in conjunction with the Regional Haze
Program to be established under sections 169A and 169B of the CAA, as
the most appropriate and effective means of addressing the public
welfare effects associated with visibility impairment. Together, the
two programs and associated control strategies were expected to provide
appropriate protection against PM-related visibility impairment and
enable all regions of the country to make reasonable progress toward
the national visibility goal.
In 2006, the EPA revised the suite of secondary PM2.5
standards to address visibility impairment by making the suite of
secondary standards identical to the revised suite of primary
PM2.5 standards. The EPA's decision regarding the need to
revise the suite of secondary PM2.5 standards reflected a
number of new developments that had occurred and sources of information
that had become available following the 1997 review. First, the EPA
promulgated a Regional Haze Program in 1999 (65 FR 35713, July 1, 1999)
which required states to establish goals for improving visibility in
Federal Class I areas and to adopt control strategies to achieve these
goals. Second, extensive new information from visibility and fine
particle monitoring networks had become available, allowing for updated
characterizations of visibility trends and PM concentrations in urban
areas, as well as Federal Class I areas. These new data allowed the EPA
to better characterize visibility impairment in urban areas and the
relationship between visibility and PM2.5 concentrations.
Finally, additional studies in the U.S. and abroad provided the basis
for the establishment of standards and programs to address specific
visibility concerns in a number of local areas. These studies (Denver,
Phoenix, and British Columbia) utilized photographic representations of
visibility impairment and produced reasonably consistent results in
terms of the visual ranges found to be generally acceptable by study
participants. The EPA considered the information generated by these
studies useful in characterizing the nature of particle-induced haze
and for informing judgments about the acceptability of various levels
of visual air quality in urban areas across the U.S. Based largely on
this information, the Administrator concluded that it was appropriate
to revise the secondary PM2.5 standards to provide increased
protection from visibility impairment principally in urban areas, in
conjunction with the regional haze program for protection of visual air
quality in Federal Class I areas.
In so doing, the Administrator recognized that PM-related
visibility impairment is principally related to fine particle
concentrations and that perception of visibility impairment is most
directly related to short-term, nearly instantaneous levels of visual
air quality. Thus, in considering whether the then-current suite of
secondary standards would provide the appropriate degree of protection,
he concluded that it was appropriate to focus on just the 24-hour
secondary PM2.5 standard to provide requisite protection.
The Administrator then considered whether PM2.5 mass
remained the appropriate indicator for a secondary

[[Page 3184]]

standard to protect visibility, primarily in urban areas. The
Administrator noted that PM-related visibility impairment is
principally related to fine particle levels. Hygroscopic components of
fine particles, in particular sulfates and nitrates, contribute
disproportionately to visibility impairment under high humidity
conditions. Particles in the coarse mode generally contribute only
marginally to visibility impairment in urban areas. With the
substantial addition to the air quality and visibility data made
possible by the national urban PM2.5 monitoring networks, an
analysis conducted for the 2006 review found that, in urban areas,
visibility levels showed far less difference between eastern and
western regions on a 24-hour or shorter time basis than implied by the
largely non-urban data available in the 1997 review. In analyzing how
well PM2.5 concentrations correlated with visibility in
urban locations across the U.S., the 2005 Staff Paper concluded that
clear correlations existed between 24-hour average PM2.5
concentrations and calculated (i.e., reconstructed) light extinction,
which is directly related to visual range (U.S. EPA, 2005, p. 7-6).
These correlations were similar in the eastern and western regions of
the U.S. These correlations were less influenced by relative humidity
and more consistent across regions when PM2.5 concentrations
were averaged over shorter, daylight time periods (e.g., 4 to 8 hours)
when relative humidity in eastern urban areas was generally lower and
thus more similar to relative humidity in western urban areas. The 2005
Staff Paper noted that a standard set at any specific PM2.5
concentration would necessarily result in visual ranges that vary
somewhat in urban areas across the country, reflecting the variability
in the correlations between PM2.5 concentrations and light
extinction. The 2005 Staff Paper concluded that it was appropriate to
use PM2.5 as an indicator for standards to address
visibility impairment in urban areas, especially when the indicator is
defined for a relatively short period (e.g., 4 to 8 hours) of daylight
hours (U.S. EPA, 2005, p. 7-6). Based on their review of the Staff
Paper, most CASAC Panel members also endorsed such a PM2.5
indicator for a secondary standard to address visibility impairment
(Henderson, 2005a, p. 9). Based on the above considerations, the
Administrator concluded that PM2.5 should be retained as the
indicator for fine particles as part of a secondary standard to address
visibility protection, in conjunction with averaging times from 4 to 24
hours.
In considering what level of protection against PM-related
visibility impairment would be appropriate, the Administrator took into
account the results of the public perception and attitude surveys
regarding the acceptability of various degrees of visibility impairment
in the U.S. and Canada, state and local visibility standards within the
U.S., and visual inspection of photographic representations of several
urban areas across the U.S. In the Administrator's judgment, these
sources provided useful but still quite limited information on the
range of levels appropriate for consideration in setting a national
visibility standard primarily for urban areas, given the generally
subjective nature of the public welfare effect involved. Based on
photographic representations of varying levels of visual air quality,
public perception studies, and local and state visibility standards,
the 2005 Staff Paper had concluded that 30 to 20 [mu]g/m\3\
PM2.5 represented a reasonable range for a national
visibility standard primarily for urban areas, based on a sub-daily
averaging time (U.S. EPA, 2005, p. 7-13). The upper end of this range
was below the levels at which illustrative scenic views are
significantly obscured, and the lower end was around the level at which
visual air quality generally appeared to be good based on observation
of the illustrative views. This concentration range generally
corresponded to median visual ranges in urban areas within regions
across the U.S. of approximately 25 to 35 km, a range that was bounded
above by the visual range targets selected in specific areas where
state or local agencies placed particular emphasis on protecting visual
air quality. In considering a reasonable range of forms for a
PM2.5 standard within this range of levels, the 2005 Staff
Paper had concluded that a concentration-based percentile form was
appropriate, and that the upper end of the range of concentration
percentiles for consideration should be consistent with the 98th
percentile used for the primary standard and that the lower end of the
range should be the 92nd percentile, which represented the mean of the
distribution of the 20 percent most impaired days, as targeted in the
regional haze program (U.S. EPA, 2005 pp. 7-11 to 7-13). While
recognizing that it was difficult to select any specific level and form
based on then-currently available information (Henderson, 2005a, p. 9),
the CASAC Panel was generally in agreement with the ranges of levels
and forms presented in the 2005 Staff Paper.
The Administrator also considered the level of protection that
would be afforded by the proposed suite of primary PM2.5
standards (71 FR 2681, January 17, 2006), on the basis that although
significantly more information was available than in the 1997 review
concerning the relationship between fine PM levels and visibility
across the country, there was still little available information for
use in making the relatively subjective value judgment needed in
selecting the appropriate degree of protection to be afforded by such a
standard. In so doing, the Administrator compared the extent to which
the proposed suite of primary standards would require areas across the
country to improve visual air quality with the extent of increased
protection likely to be afforded by a standard based on a sub-daily
averaging time. Based on such an analysis, the Administrator observed
that the predicted percent of counties with monitors not likely to meet
the proposed suite of primary PM2.5 standards was actually
somewhat greater than the predicted percent of counties with monitors
not likely to meet a sub-daily secondary standard with an averaging
time of 4 daylight hours, a level toward the upper end of the range
recommended in the 2005 Staff Paper, and a form within the recommended
range. Based on this comparison, the Administrator tentatively
concluded that revising the secondary 24-hour PM2.5 standard
to be identical to the proposed revised primary PM2.5
standard (and retaining the then-current annual secondary
PM2.5 standard) was a reasonable policy approach to
addressing visibility protection primarily in urban areas. In proposing
this approach, the Administrator also solicited comment on a sub-daily
(4- to 8-hour averaging time) secondary PM2.5 standard (71
FR 2675 to 2781, January 17, 2006).
In commenting on the proposed decision, the CASAC requested that a
sub-daily standard to protect visibility ``be favorably reconsidered''
(Henderson, 2006a, p.6). The CASAC noted three cautions regarding the
proposed reliance on a secondary PM2.5 standard identical to
the proposed 24-hour primary PM2.5 standard: (1)
PM2.5 mass measurement is a better indicator of visibility
impairment during daylight hours, when relative humidity is generally
low; the sub-daily standard more clearly matches the nature of
visibility impairment, whose adverse effects are most evident during
the daylight hours; using a 24-hour PM2.5 standard as a
proxy introduces error and

[[Page 3185]]

uncertainty in protecting visibility; and sub-daily standards are used
for other NAAQS and should be the focus for visibility; (2) CASAC and
its monitoring subcommittees had repeatedly commended EPA's initiatives
promoting the introduction of continuous and near-continuous PM
monitoring and recognized that an expanded deployment of continuous
PM2.5 monitors would be consistent with setting a sub-daily
standard to protect visibility; and (3) the analysis showing a
similarity between percentages of counties not likely to meet what the
CASAC Panel considered to be a lenient 4- to 8-hour secondary standard
and a secondary standard identical to the proposed 24-hour primary
standard was a numerical coincidence that was not indicative of any
fundamental relationship between visibility and health. The CASAC Panel
further stated that ``visual air quality is substantially impaired at
PM2.5 concentrations of 35 [mu]g/m\3\'' and that ``[i]t is
not reasonable to have the visibility standard tied to the health
standard, which may change in ways that make it even less appropriate
for visibility concerns'' (Henderson, 2006a, pp. 5 to 6).
In reaching a final decision, the Administrator focused on the
relative protection provided by the proposed primary standards based on
the above-mentioned similarities in percentages of counties meeting
alternative standards and on the limitations in the information
available concerning studies of public perception and attitudes
regarding the acceptability of various degrees of visibility impairment
in urban areas, as well as on the subjective nature of the judgment
required. In so doing, the Administrator concluded that caution was
warranted in establishing a distinct secondary standard for visibility
impairment and that the available information did not warrant adopting
a secondary standard that would provide either more or less protection
against visibility impairment in urban areas than would be provided by
secondary standards set equal to the proposed primary PM2.5
standards.
2. Remand of 2006 Secondary PM2.5 Standards
As noted above in section II.B.2 above, several parties filed
petitions for review challenging EPA's decision to set the secondary
NAAQS for fine PM identical to the primary NAAQS. On judicial review,
the D.C. Circuit remanded to the EPA for reconsideration the secondary
NAAQS for fine PM because the Agency's decision was unreasonable and
contrary to the requirements of section 109(b)(2). American Farm Bureau
Federation v. EPA, 559 F. 3d 512 (D.C. Cir., 2009).
The petitioners argued that the EPA's decision lacked a reasoned
basis. First, they asserted that the EPA never determined what level of
visibility was ``requisite to protect the public welfare.'' They argued
that the EPA unreasonably rejected the target level of protection
recommended by its staff, while failing to provide a target level of
its own. The court agreed, stating that ``the EPA's failure to identify
such a level when deciding where to set the level of air quality
required by the revised secondary fine PM NAAQS is contrary to the
statute and therefore unlawful. Furthermore, the failure to set any
target level of visibility protection deprived the EPA's decision-
making of a reasoned basis.'' 559 F. 3d at 530.
Second, the petitioners challenged EPA's method of comparing the
protection expected from potential standards. They contended that the
EPA relied on a meaningless numerical comparison, ignored the effect of
humidity on the usefulness of a standard using a daily averaging time,
and unreasonably concluded that the primary standards would achieve a
level of visibility roughly equivalent to the level the EPA staff and
CASAC deemed ``requisite to protect the public welfare.'' The court
found that the EPA's equivalency analysis based on the percentages of
counties exceeding alternative standards ``failed on its own terms.''
The same table showing the percentages of counties exceeding
alternative secondary standards, used for comparison to the percentages
of counties exceeding alternative primary standards to show
equivalency, also included six other alternative secondary standards
within the recommended CASAC range that would be more ``protective''
under EPA's definition than the adopted primary standards. Two-thirds
of the potential secondary standards within the CASAC's recommended
range would be substantially more protective than the adopted primary
standards. The court found that the EPA failed to explain why it looked
only at one of the few potential secondary standards that would be less
protective, and only slightly less so, than the primary standards. More
fundamentally, however, the court found that the EPA's equivalency
analysis based on percentages of counties demonstrated nothing about
the relative protection offered by the different standards, and that
the tables offered no valid information about the relative visibility
protection provided by the standards. 559 F. 3d at 530-31.
Finally, the Staff Paper had made clear that a visibility standard
using PM2.5 mass as the indicator in conjunction with a
daily averaging time would be confounded by regional differences in
humidity. The court noted that the EPA acknowledged this problem, yet
did not address this issue in concluding that the primary standards
would be sufficiently protective of visibility. 559 F. 3d at 530.
Therefore, the court granted the petition for review and remanded for
reconsideration the secondary PM2.5 NAAQS.
3. General Approach Used in the Policy Assessment for the Current
Review
The approach used in this review broadened the general approaches
used in the last two PM NAAQS reviews by utilizing, to the extent
available, enhanced tools, methods, and data to more comprehensively
characterize visibility impacts. As such, the EPA took into account
considerations based on both the scientific evidence (``evidence-
based'') and a quantitative analysis of PM-related impacts on
visibility (``impact-based'') to inform conclusions related to the
adequacy of the current secondary PM2.5 standards and
alternative standards that were appropriate for consideration in this
review. As in past reviews, the EPA also considered that the secondary
NAAQS should address PM-related visibility impairment in conjunction
with the Regional Haze Program, such that the secondary NAAQS would
focus on protection from visibility impairment principally in urban
areas in conjunction with the Regional Haze Program that is focused on
improving visibility in Federal Class I areas. The EPA again recognized
that such an approach remains the most appropriate and effective means
of addressing the public welfare effects associated with visibility
impairment in areas across the country.
The Policy Assessment drew from the qualitative evaluation of all
studies discussed in the Integrated Science Assessment (U.S. EPA,
2009a). Specifically, the Policy Assessment considered the extensive
new air quality and source apportionment information available from the
regional planning organizations, long-standing evidence of PM effects
on visibility, and limited public preference study information from
four urban areas (U.S. EPA, 2009a, chapter 9), as well as the
integration of evidence across disciplines (U.S. EPA, 2009a, chapter
2). In addition, limited information that had become available
regarding the characterization of public preferences in urban areas
provided

[[Page 3186]]

some new perspectives on the usefulness of this information in
informing the selection of target levels of urban visibility
protection. On these bases, the Policy Assessment again focused
assessments on visibility conditions in urban areas.
The conclusions in the Policy Assessment reflected EPA staff's
understanding of both evidence-based and impact-based considerations to
inform two overarching questions related to (1) the adequacy of the
current suite of PM2.5 standards and (2) what potential
alternative standards, if any, should be considered in this review to
provide appropriate protection from PM-related visibility impairment.
In addressing these broad questions, the discussions in the Policy
Assessment were organized around a series of more specific questions
reflecting different aspects of each overarching question (U.S. EPA,
2011a, Figure 4-1). When evaluating the visibility protection afforded
by the current or any alternative standards considered, the Policy
Assessment took into account the four basic elements of the NAAQS:
indicator, averaging time, level, and form.

B. Proposed Decisions on Secondary PM Standards

At the time of proposal, the Administrator proposed to revise the
suite of secondary PM standards by adding a distinct standard for
PM2.5 to address PM-related visibility impairment, focused
primarily on visibility in urban areas. This proposed standard was to
be defined in terms of a PM2.5 visibility index, which would
use measured PM2.5 mass concentration, in combination with
speciation and relative humidity data, to calculate PM2.5
light extinction, translated into the deciview (dv) scale; a 24-hour
averaging time; a 90th percentile form, averaged over 3 years; and a
level of 28-30 dv. To address other non-visibility welfare effects, the
Administrator proposed to retain the current suite of secondary PM
standards generally, while revising only the form of the secondary
annual PM2.5 standard to remove the option for spatial
averaging consistent with this proposed change to the primary annual
PM2.5 standard. Each of these proposed decisions is
described in more detail in the proposal and below.
1. PM-Related Visibility Impairment
As discussed in Section VI.B of the proposal, the Administrator's
proposed decision regarding a distinct secondary standard to provide
protection from visibility impairment reflected careful consideration
of the following: (1) The latest scientific information on visibility
effects associated with PM as described in the Integrated Science
Assessment (U.S. EPA, 2009a); (2) insights gained from assessments of
correlations between ambient PM2.5 and visibility impairment
prepared by EPA staff in the Visibility Assessment (U.S. EPA, 2010b);
and (3) specific conclusions regarding the need for revisions to the
current standards (i.e., indicator, averaging time, form, and level)
that, taken together, would be requisite to protect the public welfare
from adverse effects on visual air quality. This section summarizes key
information from the proposal regarding the nature of visibility
impairment, including the relationship between ambient PM and
visibility, temporal variations in light extinction, periods during the
day of interest for assessing visibility conditions, and exposure
durations of interest (section VI.B.1.a); limited public perceptions
and attitudes about visibility impairment and the impacts of visibility
impairment on public welfare (section VI.B.1.b); CASAC advice regarding
the need for, and design of, secondary standards to protect visibility
(section VI.B.1.c); and the Administrator's proposed conclusions
regarding setting a distinct standard to address visibility impairment
(section VI.B.1.d).
a. Nature of PM-Related Visibility Impairment
As noted at the time of proposal, the fundamental science
characterizing the contribution of PM, especially fine particles, to
visibility impairment is well understood. This science provides the
basis for the Integrated Science Assessment designation of the
relationship between PM and visibility impairment as causal. New
research available in this review, discussed in chapter 9 of the
Integrated Science Assessment, continues to support and refine EPA's
understanding of the effect of PM on visibility and the source
contributions to that effect in rural and remote locations. This
research provides new insights regarding the regional source
contributions to urban visibility impairment and better
characterization of the increment in PM concentrations and visibility
impairment that occur in many cities (i.e., the urban excess) relative
to conditions in the surrounding rural areas (i.e., regional
background). Ongoing urban PM2.5 speciated and aggregated
mass monitoring has produced new information that has allowed for
updated characterization of current visibility levels in urban areas.
i. Relationship Between Ambient PM and Visibility
Visibility impairment is caused by the scattering and absorption of
light by suspended particles and gases in the atmosphere. When PM is
present in the air, its contribution to light extinction typically
greatly exceeds that of gases. The combined effect of light scattering
and absorption by both particles and gases is characterized as light
extinction, i.e., the fraction of light that is scattered or absorbed
in the atmosphere. Light extinction can be quantified by a light
extinction coefficient with units of 1/distance, which is often
expressed as 1/(1 million meters) or inverse megameters (abbreviated
Mm^-1) or in terms of an alternative scale known as the
deciview scale, defined by the following equation: \146\
---------------------------------------------------------------------------

\146\ As used in the Regional Haze Program, the term
bext refers to light extinction due to PM2.5,
PM10-2.5, and ``clean'' atmospheric gases. In the Policy
Assessment, in focusing on light extinction due to PM2.5,
the deciview values include only the effects of PM2.5 and
the gases. The ``Rayleigh'' term associated with clean atmospheric
gases is represented by the constant value of 10 Mm^-\1\.
Omission of the Rayleigh term would create the possibility of
negative deciview values when the PM2.5 concentration is
very low.
---------------------------------------------------------------------------

Deciview (dv) = 10 ln (bext/ 10 Mm^-1)

The deciview scale is frequently used in the scientific literature on
visibility, as well as in the Regional Haze Program. In particular, the
deciview scale is used in the public perception studies that were
considered in the past and current reviews to inform judgments about an
appropriate degree of protection to be provided by a secondary NAAQS.
The amount of light extinction contributed by PM depends on the
particle concentration as well as on the particle size distribution and
composition and also on the relative humidity. As described in detail
in section VI.B.1.a of the proposal, visibility scientists have
developed an algorithm, known as the IMPROVE algorithm,\147\ to
estimate light extinction using routinely monitored fine particle
(PM2.5) speciation and coarse particle mass
(PM10-2.5) data, as well as data on relative humidity. There
is both an original and a revised version of the IMPROVE algorithm
(Pitchford et al., 2007). The revised version was developed to address
observed biases in the predictions using the original algorithm under
very low and very high

[[Page 3187]]

light extinction conditions.\148\ These IMPROVE algorithms are
routinely used to calculate light extinction levels on a 24-hour basis
in Federal Class I areas under the Regional Haze Program.
---------------------------------------------------------------------------

\147\ The algorithm is referred to as the IMPROVE algorithm
because it was developed specifically to use the aerosol monitoring
data generated at network sites and with equipment specifically
designed to support the IMPROVE program and was evaluated using
IMPROVE optical measurements at the subset of sites that make those
measurements (Malm et al., 1994).
\148\ These biases were detected by comparing light extinction
estimates generated from the IMPROVE algorithm to direct optical
measurements in a number of rural Federal Class I areas.
---------------------------------------------------------------------------

In either version of the IMPROVE algorithm, the concentration of
each of the major aerosol components is multiplied by a dry extinction
efficiency value and, for the hygroscopic components (i.e., ammoniated
sulfate and ammonium nitrate), also multiplied by an additional factor
to account for the water growth to estimate these components'
contribution to light extinction. Summing the contribution of each
component gives the estimate of total light extinction per unit
distance denoted as the light extinction coefficient (bext), as shown
below for the original IMPROVE algorithm.

bext [ap] 3 x f(RH) x [Sulfate]
+ 3 x f(RH) x [Nitrate]
+ 4 x [Organic Mass]
+ 10 x [Elemental Carbon]
+ 1 x [Fine Soil]
+ 0.6 x [Coarse Mass]
+ 10

Light extinction (bext) is in units of Mm^-1,
the mass concentrations of the components indicated in brackets are in
units of [mu]g/m\3\, and f(RH) is the unitless water growth term that
depends on relative humidity. The final term of 10 Mm^-1 is
known as the Rayleigh scattering term and accounts for light scattering
by the natural gases in unpolluted air. Despite the simplicity of this
algorithm, it performs reasonably well and permits the contributions to
light extinction from each of the major components (including the water
associated with the sulfate and nitrate compounds) to be separately
approximated. Inspection of the PM component-specific terms in the
simple original IMPROVE algorithm shows that most of the
PM2.5 components contribute 5 times or more light extinction
than a similar concentration of PM10-2.5.
The f(RH) term in the original algorithm reflects the increase in
light scattering caused by particulate sulfate and nitrate under
conditions of high relative humidity. Particles with hygroscopic
components (e.g., particulate sulfate and nitrate) contribute more
light extinction at higher relative humidity than at lower relative
humidity because they change size in the atmosphere in response to
ambient relative humidity conditions. For relative humidity below 40
percent the f(RH) value is 1, but it increases to 2 at approximately 66
percent, 3 at approximately 83 percent, 4 at approximately 90 percent,
5 at approximately 93 percent, and 6 at approximately 95 percent
relative humidity. The result is that both particulate sulfate and
nitrate are more efficient per unit mass in light extinction than any
other aerosol component for relative humidity above approximately 85
percent where their total light extinction efficiency exceeds the 10
m\2\/g associated with elemental carbon (EC). PM containing elemental
or black carbon (BC) absorbs light as well as scattering it, making it
the component with the greatest light extinction contributions per unit
of mass concentration, except for the hygroscopic components under
these high relative humidity conditions.\149\
---------------------------------------------------------------------------

\149\ The IMPROVE algorithm does not explicitly separate the
light-scattering and light-absorbing effects of elemental carbon.
---------------------------------------------------------------------------

As noted above, subsequent to the development of the original
IMPROVE algorithm, an alternative algorithm (variously referred to as
the ``revised algorithm'' or the ``new algorithm'' in the literature)
was developed. The revised IMPROVE algorithm is different from the
original algorithm in several important ways. First, the revised
algorithm employs a more complex split-component mass extinction
efficiency to correct biases believed to be related to particle size
distributions.\150\ Specifically, the revised algorithm incorporates
terms to account for particles representing the different dry
extinction and water uptake from two size modes of sulfate, nitrate and
organic mass.\151\ Second, the revised algorithm uses a different
multiplier for organic carbon for purposes of estimating organic
carbonaceous material to better represent aged aerosol found in remote
areas.\152\ In addition, the revised algorithm includes a term for
hygroscopic sea salt that can be important for remote coastal areas,
and site-specific Rayleigh light scattering terms in place of a
universal Rayleigh light scattering value. As noted in section VI.B.1.a
of the proposal, the revised IMPROVE algorithm can yield higher
estimates of current light extinction levels in urban areas on days
with relatively poor visibility as compared to the original algorithm
(Pitchford, 2010). This difference is primarily attributable to the
split-component mass extinction efficiency treatment in the revised
algorithm. This revised algorithm was evaluated at 21 remote locations
and is generally used by RPOs and States for implementation of the
Regional Haze Rule.
---------------------------------------------------------------------------

\150\ In either version of the IMPROVE algorithm, the
concentration of each of the major aerosol components is multiplied
by a dry extinction efficiency value and, for the hygroscopic
components (i.e., ammoniated sulfate and ammonium nitrate), also
multiplied by an additional factor to account for the water growth
to estimate these components' contribution to light extinction. Both
the dry extinction efficiency and water growth terms have been
developed by a combination of empirical assessment and theoretical
calculation using typical particle size distributions associated
with each of the major aerosol components.
\151\ The relative contributions of sulfate, nitrate, and
organic mass concentrations to visibility impairment with the
revised algorithm are different than with the original algorithm due
to the combination of the dry extinction coefficient and f(RH)
functions for derived concentrations of small and large particles.
The apportionment of the total fine particle concentration of each
of the three PM2.5 components into the concentrations of
the small and large size fractions was empirically developed for
remote areas. The fraction of the fine particle component that is in
the large mode is estimated by dividing the total concentration of
the component by 20 [mu]g/m\3\. If the total concentration of a
component exceeds 20 [mu]g/m\3\, all of it is assumed to be in the
large mode.
\152\ The revised IMPROVE algorithm uses a multiplier of 1.8 for
rural areas instead of 1.4 as used in the original algorithm for the
mean ratio of organic mass to organic carbon.
---------------------------------------------------------------------------

ii. Temporal Variations of Light Extinction
Particulate matter concentrations and light extinction in urban
environments vary from hour to hour throughout the 24-hour day due to a
combination of diurnal changes in meteorological conditions and
systematic changes in emissions activity (e.g., rush hour traffic).
Various factors combine to make early morning the most likely time for
peak urban light extinction; although the net effects of the systematic
urban- and larger-scale variations mean that peak daytime PM light
extinction levels can occur any time of day, in many areas they occur
most often in early morning hours (U.S. EPA, 2010b, sections 3.4.2 and
3.4.3; Figures 3-9, 3-10, and 3-12). This temporal pattern in urban
areas contrasts with the general lack of a strong diurnal pattern in PM
concentrations and light extinction in most Federal Class I areas,
reflective of a relative lack of local sources as compared to urban
areas. The use in the Regional Haze Program of 24-hour average
concentrations in the IMPROVE algorithm is consistent with this general
lack of a strong diurnal pattern in Federal Class I areas.
iii. Periods During the Day of Interest for Assessment of Visibility
As noted in sections VI.B.1.b and VI.B.1.c of the proposal, daytime
visibility has dominated the attention of

[[Page 3188]]

those who have studied the visibility effects of air pollution,
particularly in urban areas. The EPA recognizes, however, that
physically PM light extinction behaves the same at night as during the
day and can contribute to nighttime visibility effects by enhancing the
scattering of anthropogenic light, contributing to the ``skyglow''
within and over populated areas, adding to the total sky brightness,
and contributing to the reduction in contrast of stars against the
background. However, little research has been conducted on nighttime
visibility, and the state of the science is not comparable to that
associated with daytime visibility impairment, particularly in terms of
the impact on human welfare. The Policy Assessment notes that the
science is not available at this time to support adequate
characterization specifically of nighttime PM light extinction
conditions and the related effects on public welfare (U.S. EPA, 2011a,
p. 4-18). Therefore the EPA has focused its assessments of PM
visibility impacts in urban areas on daylight hours during this review.
iv. Exposure Durations of Interest
As noted in section VI.B.1.d of the proposal, the roles that
exposure duration and variations in visual air quality within any given
exposure period play in determining the acceptability or
unacceptability of a given level of visual air quality have not been
investigated via preference studies. In the preference studies
available for this review, subjects were simply asked to rate the
acceptability or unacceptability of each image of a haze-obscured
scene, without being provided any suggestion of assumed duration or of
assumed conditions before or after the occurrence of the scene
presented. Preference and/or valuation studies show that atmospheric
visibility conditions can be quickly assessed and preferences
determined. The EPA is unaware of any studies that characterize the
extent to which different frequencies and durations of exposure to
visibility conditions contribute to the degree of public welfare impact
that occurs.
The Policy Assessment considered a variety of circumstances that
are commonly expected to occur in evaluating the potential impact of
visibility impairment on the public welfare based on available
information (U.S. EPA, 2011a, pp. 4-19 to 4-20). In some circumstances,
such as infrequent visits to scenic vistas in natural or urban
environments, people are motivated specifically to take the opportunity
to view a valued scene and are likely to do so for many minutes to
hours to appreciate various aspects of the vista they choose to view.
However, the public has many more opportunities to notice visibility
conditions on a daily basis in settings associated with performing
daily routines (e.g., during commutes and while working, exercising, or
recreating outdoors). As noted in the Policy Assessment, information
regarding the fraction of the public that has only one or a few
opportunities to experience visibility during the day, or on the role
the duration of the observed visibility conditions has on wellbeing
effects associated with those visibility conditions, is not available
(U.S. EPA, 2011a, p. 4-20). However, it is possible that people with
limited opportunities to experience visibility conditions on a daily
basis would receive the entire impact of the day's visual air quality
based on the visibility conditions that occur during the short time
period when they can see it. Since this group could be affected on the
basis of observing visual air quality conditions for periods as short
as one hour or less, and because during each daylight hour there are
some people outdoors, commuting, or near windows, the Policy Assessment
judged that it would be appropriate to use the maximum hourly value of
PM light extinction during daylight hours for each day for purposes of
evaluating the adequacy of the current suite of secondary standards.
Other observers may have access to visibility conditions throughout the
day. For this group, it might be that an hour with poor or
``unacceptable'' visibility can be offset by one or more other hours
with clearer conditions. Therefore, the proposal acknowledged that it
might also be appropriate to consider a multi-hour daylight exposure
period.
v. Periods of Fog and Rain
As discussed in section VI.C of the proposal, the EPA also
recognized that it is appropriate to give special treatment to periods
of fog and rain when considering whether current PM2.5
standards adequately protect public welfare from PM-related visibility
impairment. Visibility impairment occurs during periods with fog or
precipitation irrespective of the presence or absence of PM. Therefore,
it is logical that periods with naturally impaired visibility due to
fog or precipitation should not be treated as having PM-impaired
visibility. There are multiple ways to adjust visibility data to reduce
the effects of fog and precipitation. In the Visibility Assessment,
following the advice of CASAC, the EPA evaluated the effect of
excluding daylight hours for which relative humidity was greater than
90 percent from analyses in order to avoid precipitation and fog
confounding estimates of PM visibility impairment. For the 15 urban
areas included in the Visibility Assessment, the EPA found that a 90
percent relative humidity cutoff criterion was effective in that on
average less than 6 percent of the daylight hours were removed from
consideration, yet those hours had on average ten times the likelihood
of rain, six times the likelihood of snow/sleet, and 34 times the
likelihood of fog compared with hours with 90 percent or lower relative
humidity. In the Regional Haze program, the EPA utilizes monthly
average relative humidity values based on 10 years of climatological
data to reduce the effect of fog and precipitation. This approach
focuses on longer-term averages for each monitoring site and thereby
eliminates the effect of very high humidity conditions on visibility at
those locations.
b. Public Perception of Visibility Impairment
As described in section VI.B.2 of the proposal, there are two main
types of studies that evaluate the public perception of urban
visibility impairment: urban visibility preference studies and urban
visibility valuation studies. As noted in the Integrated Science
Assessment, ``[b]oth types of studies are designed to evaluate
individuals' desire (or demand) for good visual air quality (VAQ) where
they live, using different metrics to evaluate demand. Urban visibility
preference studies examine individuals' demand by investigating what
amount of visibility degradation is unacceptable while economic studies
examine demand by investigating how much one would be willing to pay to
improve visibility'' (U.S. EPA, 2009a, p. 9-66). Because of the limited
number of new studies on urban visibility valuation, the Integrated
Science Assessment cites to the discussion in the 2004 Criteria
Document of the various methods one can use to determine the economic
valuation of changes in visibility, which include hedonic valuation,
contingent valuation and contingent choice, and travel cost.
Contingent valuation studies are a type of stated preference study
that measures the strength of preferences and expresses that preference
in dollar values. Contingent valuation studies often include payment
vehicles that require respondents to consider implementation costs and
their ability to pay for visibility improvements in their responses.
This study design

[[Page 3189]]

aspect is critical because the EPA cannot consider implementations
costs in setting either primary or secondary NAAQS. Therefore in
considering the information available to help inform the standard-
setting process, the EPA has focused on the public perception studies
that do not embed consideration of implementation costs. Nonetheless,
the EPA recognizes that valuation studies do provide additional
evidence that the public is experiencing losses in welfare due to
visibility impairment.\153\ The public perception studies are described
in detail below.
---------------------------------------------------------------------------

\153\ In the regulatory impact analysis (RIA) accompanying this
rulemaking, the EPA describes a revised approach to estimate urban
residential visibility benefits that applies the results of several
contingent valuation studies. The EPA is unable to apply the public
perception studies to estimate benefits because they do not provide
sufficient information on which to develop monetized benefits
estimates. Specifically, the public perception studies do not
provide preferences expressed in dollar values, even though they do
provide additional evidence that the benefits associated with
improving residential visibility are not zero. As previously noted
in this preamble, the RIA is done for informational purposes only,
and the proposed decisions on the NAAQS in this rulemaking are not
in any way based on consideration of the information or analyses in
the RIA.
---------------------------------------------------------------------------

In order to identify levels of visibility impairment appropriate
for consideration in setting secondary PM NAAQS to protect the public
welfare, the Visibility Assessment comprehensively examined information
that was available in this review regarding people's stated preferences
regarding acceptable and unacceptable visual air quality.
Light extinction is an atmospheric property that by itself does not
directly translate into a public welfare effect. Instead, light
extinction becomes meaningful in the context of the impact of
differences in visibility on the human observer. This has been studied
in terms of the acceptability or unacceptability expressed for the
visibility impact of a given level of light extinction by a human
observer. The perception of the visibility impact of a given level of
light extinction occurs in conjunction with the associated
characteristics and lighting conditions of the viewed scene.\154\ Thus,
a given level of light extinction may be perceived differently by
observers looking at different scenes or the same scene with different
lighting characteristics. Likewise, different observers looking at the
same scene with the same lighting may have different preferences
regarding the associated visual air quality. When scene and lighting
characteristics are held constant, the perceived appearance of a scene
(i.e., how well the scenic features can be seen and the amount of
visible haze) depends only on changes in light extinction. This has
been demonstrated using the WinHaze model (Molenar et al., 1994) that
uses image processing technology to apply user-specified changes in
light extinction values to the same base photograph with set scene and
lighting characteristics.
---------------------------------------------------------------------------

\154\ By ``characteristics of the scene'' the EPA means the
distance(s) between the viewer and the object(s) of interest, the
shapes and colors of the objects, the contrast between objects and
the sky or other background, and the inherent interest of the
objects to the viewer. Distance is particularly important because at
a given value of light extinction, which is a property of air at a
given point(s) in space, more light is actually absorbed and
scattered when light passes through more air between the object and
the viewer.
---------------------------------------------------------------------------

Much of what is known about the acceptability of levels of
visibility comes from survey studies in which participants were asked
questions about their preference or the value they place on various
visibility levels as displayed to them in scenic photographs and/or
WinHaze images with a range of known light extinction levels. The
Visibility Assessment (U.S. EPA, 2010b, chapter 2) reviewed the limited
number of urban visibility preference studies currently available
(i.e., four studies) to assess the light extinction levels judged by
the participant to have acceptable visibility for those particular
scenes.
The reanalysis of urban preference studies conducted in the
Visibility Assessment for this review included three completed western
urban visibility preference survey studies plus a pair of smaller focus
studies designed to explore and further develop urban visibility survey
instruments. The three western studies included one in Denver, Colorado
(Ely et al., 1991), one in the lower Fraser River valley near
Vancouver, British Columbia (BC), Canada (Pryor, 1996), and one in
Phoenix, Arizona (BBC Research & Consulting, 2003). A pilot focus group
study was also conducted for Washington, DC (Abt Associates Inc.,
2001). In response to an EPA request for public comment on the Scope
and Methods Plan (74 FR 11580, March 18, 2009), comments were received
(Smith, 2009) about the results of a new focus group study of scenes
from Washington, DC, that had been conducted on subjects from both
Houston, Texas, and Washington, DC, using scenes, methods and
approaches similar to the method and approach employed in the EPA pilot
study (Smith and Howell, 2009). When taken together, these studies from
the four different urban areas included a total of 852 individuals,
with each individual responding to a series of questions while viewing
a set of images of various urban visual air quality conditions.
The approaches used in the four studies were similar and were all
derived from the method first developed for the Denver urban visibility
study. In particular, the studies all used a similar group interview
type of survey to investigate the level of visibility impairment that
participants described as ``acceptable.'' In each preference study,
participants were initially given a set of ``warm up'' exercises to
familiarize them with how the scene in the photograph or image appears
under different VAQ conditions. The participants next were shown 25
randomly ordered photographs (images), and asked to rate each one based
on a scale of 1 (poor) to 7 (excellent). They were then shown the same
photographs or images again, in the same order, and asked to judge
whether each of the photographs (images) would violate what they would
consider to be an appropriate urban visibility standard (i.e. whether
the level of impairment was ``acceptable'' or ``unacceptable''). The
term ``acceptable'' was not defined, so that each person's response was
based on his/her own values and preferences for VAQ. However, when
answering this question, participants were instructed to consider the
following three factors: (1) The standard would be for their own urban
area, not a pristine national park area where the standards might be
stricter; (2) The level of an urban visibility standard violation
should be set at a VAQ level considered to be unreasonable,
objectionable, and unacceptable visually; and (3) Judgments of
standards violations should be based on visibility only, not on health
effects. While the results differed among the four urban areas, results
from a rating exercise show that within each preference study,
individual survey participants consistently distinguish between photos
or images representing different levels of light extinction, and that
more participants rate as acceptable images representing lower levels
of light extinction than they do images representing higher levels.
Given the similarities in the approaches used, the EPA staff
concluded that it was reasonable to compare the results to identify
overall trends in the study findings and to conclude that this
comparison can usefully inform the selection of a range of levels for
use in further analyses. However, the staff also noted that variations
in the specific materials and methods used in each study introduce
uncertainties that should also be considered when interpreting the
results

[[Page 3190]]

of these comparisons. Key differences between the studies include the
following: (1) Scene characteristics; (2) image presentation methods
(e.g., projected slides of actual photos, projected images generated
using WinHaze (a significant technical advance in the method of
presenting visual air quality conditions), or use of a computer monitor
screen; (3) number of participants in each study; (4) participant
representativeness of the general population of the relevant
metropolitan area; and (5) specific wording used to frame the questions
used in the group interview process.
In the Visibility Assessment, each study was evaluated separately
and figures developed to display the percentage of participants that
rated the visual air quality depicted in each photograph as
``acceptable.'' Ely et al. (1991) introduced a ``50% acceptability''
criterion analysis of the Denver preference study results. The 50
percent acceptability criterion is designed to identify the visual air
quality level (defined in terms of deciviews or light extinction) that
best divides the photographs into two groups: Those with a visual air
quality rated as acceptable by the majority of the participants, and
those rated not acceptable by the majority of participants. The
Visibility Assessment adopted this criterion as a useful index for
comparison between studies. The results of each analysis were then
combined graphically to allow for visual comparison. This information
was then carried forward into the Policy Assessment. Figure 5 presents
the graphical summary of the results of the studies in the four cities
and draws on results previously presented in Figures 2-3, 2-5, 2-7, and
2-11 of chapter 2 in the Visibility Assessment. Figure 5 also contains
lines at 20 dv and 30 dv that generally identify a range where the 50
percent acceptance criteria occur across all four of the urban
preference studies (U.S. EPA, 2011a, p. 4-24). Out of the 114 data
points shown in Figure 5, only one photograph (or image) with a visual
air quality below 20 dv was rated as acceptable by less than 50 percent
of the participants who rated that photograph.\155\ Similarly, only one
image with a visual air quality above 30 dv was rated acceptable by
more than 50 percent of the participants who viewed it.\156\
---------------------------------------------------------------------------

\155\ Only 47 percent of the British Columbia participants rated
a 19.2 dv photograph as acceptable.
\156\ In the 2001 Washington, DC study, a 30.9 dv image was used
as a repeated slide. The first time it was shown 56 percent of the
participants rated it as acceptable, but only 11 percent rated it as
acceptable the second time it was shown. The same visual air quality
level was rated as acceptable by 4 percent of the participants in
the 2009 study (Test 1). All three points are shown in Figure 5.
\157\ Top scale shows light extinction in inverse megameter
units; bottom scale in deciviews. Logit analysis estimated response
functions are shown as the color-coded curved lines for each of the
four urban areas.
[GRAPHIC] [TIFF OMITTED] TR15JA13.004

As Figure 5 above shows, each urban area has a separate and unique
response curve that appears to indicate that it is distinct from the
others.\158\ These curves are the result of a logistical regression
analysis using a logit model of the greater than 19,000 ratings of haze
images as acceptable or unacceptable. The model results can be used to
estimate the visual air quality in terms of dv values where the
estimated response functions cross the 50 percent acceptability level,
as well as any alternative criteria levels. Selected examples of these
are shown in Table 4-

[[Page 3191]]

1 of the Policy Assessment (U.S. EPA, 2011a; U.S. EPA, 2010b, Table 2-
4). This table shows that the logit model results also support the
upper and lower ends of the range of 50th percentile acceptability
values (e.g., near 20 dv for Denver and near 30 dv for Washington, DC)
already identified in Figure 5.
---------------------------------------------------------------------------

\158\ At present, data is only available for four urban areas,
as presented in Figure 5 and discussed throughout this section.
Additional research could help inform whether the range identified
by combining the results of the studies depicted in Figure 5 is more
broadly representative.
---------------------------------------------------------------------------

Based on the composite results and the effective range of 50th
percentile acceptability across the four urban preference studies shown
in Figure 5 and Table 4-1 of the Policy Assessment, benchmark levels of
(total) light extinction were selected in a range from 20 dv to 30 dv
(75 to 200 Mm^-1) \159\ for the purpose of provisionally
assessing whether visibility conditions would be considered acceptable
(i.e., less than the low end of the range), unacceptable (i.e., greater
than the high end of the range), or potentially acceptable (within the
range) based on the very limited public preference information. A
midpoint of 25 dv (120 Mm^-1) was also selected for use in
the assessment. This level is also very near to the 50th percentile
criterion value from the Phoenix study (i.e., 24.2 dv), which is by far
the best of the four studies in terms of the fit of the data to the
response curve and the representativeness of study participants. Based
on the currently available information, the Policy Assessment concluded
that the use of 25 dv to represent the middle of the distribution of
results seemed well supported (U.S. EPA, 2011a, p. 4-25).
---------------------------------------------------------------------------

\159\ These values were rounded from 74 Mm^-1 and 201
Mm^-1 to avoid an implication of greater precision than is
warranted. Note that the middle value of 25 dv when converted to
light extinction is 122 Mm^-1 is rounded to 120
Mm^-1 for the same reason. Assessments conducted for the
Visibility Assessment and the first and second drafts of the Policy
Assessment used the unrounded values. The Policy Assessment
considered the results of assessment using unrounded values to be
sufficiently representative of what would result if the rounded
values were used that it was unnecessary to redo the assessments.
That is why some tables and figures in the Policy Assessment
reflected the unrounded values.
---------------------------------------------------------------------------

These three benchmark values provide a low, middle, and high set of
light extinction conditions that are used to provisionally define
daylight hours with urban haze conditions that have been judged
unacceptable by at least 50 percent of the participants in one or more
of these preference studies. As discussed above, PM light extinction is
taken to be (total) light extinction minus the Rayleigh scatter,\160\
such that the low, middle, and high levels correspond to PM light
extinction levels of about 65 Mm^-1, 110 Mm^-1, and
190 Mm^-1. In the Visibility Assessment, these three light
extinction levels were called Candidate Protection Levels (CPLs). This
term was also used in the Policy Assessment and in the proposal notice.
It is important to note, however, that the degree of protection
provided by a secondary NAAQS is not determined solely by any one
component of the standard but by all the components (i.e., indicator,
averaging time, form, and level) being applied together. Therefore, the
Policy Assessment noted that the term CPL is meant only to indicate
target levels of visibility within a range that the EPA staff felt
appropriate for consideration that could, in conjunction with other
elements of the standard, including indicator, averaging time, and
form, potentially provide an appropriate degree of visibility
protection.
---------------------------------------------------------------------------

\160\ Rayleigh scatter is light scattering by atmospheric gases
which is on average about 10 Mm^-1.
---------------------------------------------------------------------------

In characterizing the Policy Assessment's confidence in each CPL
and across the range, a number of issues were considered (U.S. EPA,
2011a, p. 4-26). Looking first at the two studies that define the upper
and lower bounds of the range, the Policy Assessment considered whether
they represent a true regional distinction in preferences for urban
visibility conditions between western and eastern U.S. There was little
information available to help evaluate the possibility of a regional
distinction especially given that there have been preference studies in
only one eastern urban area. Smith and Howell (2009) found little
difference in preference response to Washington, DC, haze photographs
between the study participants from Washington, DC, and those from
Houston, Texas.\161\ This provides some limited evidence that the value
judgment of the public in different areas of the country may not be an
important factor in explaining the differences in these study results.
---------------------------------------------------------------------------

\161\ The first preference study using WinHaze images of a
scenic vista from Washington, DC was conducted in 2001 using
subjects who were residents of Washington, DC. More recently, Smith
and Howell (2009) interviewed additional subjects using the same
images and interview procedure. The additional subjects included
some residents of the Washington, DC area and some residents of the
Houston, Texas area.
---------------------------------------------------------------------------

In further considering what factors could explain the observed
differences in preferences across the four urban areas, the Policy
Assessment noted that the urban scenes used in each study had different
characteristics (U.S. EPA, 2011a, p. 4-26). For example, each of the
western urban visibility preference study scenes included mountains in
the background while the single eastern urban study did not. It is also
true that each of the western scenes included objects at greater
distances from the camera location than in the eastern study. There is
no question that objects at a greater distance have a greater
sensitivity to perceived visibility changes as light extinction is
changed compared to otherwise similar scenes with objects at a shorter
range. This alone might explain the difference between the results of
the eastern study and those from the western urban studies. Having
scenes with the object of greatest intrinsic value nearer and hence
less sensitive in the eastern urban area compared with more distant
objects of greatest intrinsic value in the western urban areas could
further explain the difference in preference results.
Another question considered was whether the high CPL value that is
based on the eastern preference results is likely to be generally
representative of urban areas that do not have associated mountains or
other valued objects visible in the distant background. Such areas
would include the middle of the country, many areas in the eastern
U.S., and possibly some areas in the western U.S. as well.\162\ Based
on the currently available information, the Policy Assessment concluded
that the high end of the CPL range (30 dv) is an appropriate level to
consider (U.S. EPA, 2011a, p. 4-27).
---------------------------------------------------------------------------

\162\ In order to examine this issue, an effort would have to be
made to see if scenes in such areas could be found that would be
generally comparable to the western scenes (e.g., scenes that
contain valued scenic elements at more sensitive distances than that
used in the eastern study). This is only one of a family of issues
concerning how exposure to urban scenes of varying sensitivity
affects public perception for which no preference study information
is currently available.
---------------------------------------------------------------------------

With respect to the low end of the range, the Policy Assessment
considered factors that might further refine its understanding of the
robustness of this level. The Policy Assessment concluded that
additional urban preference studies, especially with a greater variety
in types of scenes, could help evaluate whether the lower CPL value of
20 dv is generally supportable (U.S. EPA, 2011a, p. 4-27). Further, the
reason for the noisiness in data points around the curves apparent in
both the Denver and British Columbia results compared to the smoother
curve fit of Phoenix study results could be explored. One possible
explanation discussed in the Policy Assessment is that these older
studies use photographs taken at different times of day and on
different days to capture the range of light extinction levels needed
for the preference studies. In contrast, the use of WinHaze in the
Phoenix (and Washington, DC) study reduced variations that affect scene
appearance preference rating and avoided the uncertainty inherent in
using ambient measurements to

[[Page 3192]]

represent sight path-averaged light extinction values. Reducing these
sources of noisiness and uncertainty in the results of future studies
of sensitive urban scenes could provide more confidence in the
selection of a low CPL value.
Based on the above considerations, and recognizing the limitations
in the currently available information, the Policy Assessment concluded
that it is reasonable to consider a range of CPL values including a
high value of 30 dv, a mid-range value of 25 dv, and a low value of 20
dv (U.S. EPA, 2011a, p. 4-27). Based on its review of the second draft
Policy Assessment, CASAC also supported this set of CPLs for
consideration by the EPA in this review. CASAC noted that these CPL
values were based on all available visibility preference data and that
they bound the study results as represented by the 50 percent
acceptability criteria. While recommending that further visibility
preference studies be conducted to reduce remaining uncertainties,\163\
CASAC concluded that this range of levels was ``adequately supported by
the evidence presented'' (Samet, 2010d, p. iii).
---------------------------------------------------------------------------

\163\ ``CASAC has also identified needs for the next review
cycle in terms of further research on a number of topics related to
urban visibility; * * *. In particular, there is a need for the
Agency to conduct additional urban visibility preference studies
over a broad range of urban areas and viewing conditions, to further
evaluate and refine the range of visibility levels considered to be
acceptable in the current assessment.'' (Samet, 2010a)
---------------------------------------------------------------------------

c. Summary of Proposed Conclusions
i. Adequacy of the Current Standards for PM-Related Visibility
Impairment
At the time of proposal, the Administrator provisionally concluded
that the current suite of secondary PM standards is not sufficiently
protective of visual air quality, and that consideration should be
given to an alternative secondary standard that would provide
additional protection against PM-related visibility impairment, with a
focus primarily in urban areas. This proposed conclusion was based on
the information presented in the proposal with regard to the nature of
PM-related visibility impairment, the results of public perception
surveys on the acceptability of varying degrees of visibility
impairment in urban areas, analyses of the number of days that are
estimated to exceed a range of candidate protection levels under
conditions simulated to just meet the current standards, and the advice
of CASAC. This section summarizes key points from section VI.C of the
proposal regarding visibility under current conditions, the degree of
protection afforded by the current standards, and CASAC's advice
regarding the adequacy of the current standards.
As discussed in section VI.C.1 of the proposal, to evaluate
visibility under current conditions the Visibility Assessment and
Policy Assessment estimated PM-related light extinction\164\ levels for
15 urban areas\165\ in the United States. Consistent with the emphasis
in this review on the hourly or multi-hour time periods that might
reasonably characterize the visibility effects experienced by various
segments of the population, these analyses focused on using maximum 1-
hour and 4-hour values of PM light extinction during daylight hours for
purposes of evaluating the degree of visibility impairment. Hourly
average PM-related light extinction was analyzed in terms of both
PM10 and PM2.5 light extinction. For reasons
discussed above, hours with relative humidity greater than 90 percent
were excluded from consideration. Recent visibility conditions in these
urban areas were then compared to the CPLs identified above. The
Visibility Assessment, which focused on PM10 light
extinction in 14 of the 15 urban areas during the 2005 to 2007 time
period,\166\ found that all 14 areas had daily maximum hourly
PM10 light extinction values estimated to exceed even the
highest CPL some of the days. Except for the two Texas areas and the
non-California western urban areas, all of the other urban areas were
estimated to have maximum hourly PM10 concentrations that
exceeded the high CPL on about 20 percent to over 60 percent of the
days. All 14 of the urban areas were estimated to have maximum hourly
PM10 concentrations that exceeded the low CPL on about 40
percent to over 90 percent of the days. In general, areas in the East
and in California tend to have a higher frequency of hourly visibility
conditions estimated to be above the high CPL compared with those in
the western U.S.
---------------------------------------------------------------------------

\164\ PM-related light extinction is used here to refer to the
light extinction caused by PM regardless of particle size;
PM10 light extinction refers to the contribution by
particles sampled through an inlet with a particle size 50 percent
cutpoint of 10 [mu]m diameter; and PM2.5 light extinction
refers to the contribution by particles sampled through an inlet
with a particle size 50 percent cutpoint of 2.5 [mu]m diameter.
\165\ The 15 urban areas are Tacoma, Fresno, Los Angeles,
Phoenix, Salt Lake City, Dallas, Houston, St. Louis, Birmingham,
Atlanta, Detroit, Pittsburgh, Baltimore, Philadelphia, and New York.
\166\ Comments on the second draft Visibility Assessment from
those familiar with the monitoring sites in St. Louis indicated that
the site selected to provide continuous PM10 monitoring,
although less than a mile from the site of the PM2.5
data, was not representative of the urban area and resulted in
unrealistically large PM10-2.5 values. The EPA staff
considered these comments credible and set aside the St. Louis
assessment results for PM10 light extinction. Thus,
results and statements in the Policy Assessment regarding
PM10 light extinction applied to only the other 14 areas.
However, results regarding PM2.5 light extinction in most
cases applied to all 15 study areas because the St. Louis estimates
for PM2.5 light extinction were not affected by the
PM10 monitoring issue.
---------------------------------------------------------------------------

The Policy Assessment repeated the Visibility Assessment-type
modeling based on PM2.5 light extinction and data from the
more recent 2007 to 2009 time period for the same 15 study areas
(including St. Louis). While the estimates of the percentage of daily
maximum hourly PM2.5 light extinction values exceeding the
CPLs were somewhat lower than for PM10 light extinction, the
patterns of these estimates across the study areas was found to be
similar. More specifically, except for the two Texas and the non-
California western urban areas, all of the other urban areas were
estimated to have maximum hourly PM2.5 concentrations that
exceeded the high CPL on about 10 percent up to about 50 percent of the
days based on PM2.5 light extinction, while all 15 areas
were estimated to have maximum hourly PM2.5 concentrations
that exceeded the low CPL on over 10 percent to over 90 percent of the
days.
To evaluate how PM-related visibility would be affected by just
meeting the current suite of PM2.5 secondary standards, the
Policy Assessment applied the proportional rollback approach described
in section VI.C.2 of the proposal to all the PM2.5
monitoring sites in each study area.\167\ After adjusting for
composition, the Policy Assessment applied the original IMPROVE
algorithm to calculate the PM10 light extinction, using
``rolled back'' PM2.5 component concentrations, the current
conditions PM10-2.5 concentration for the day and hour, and
relative humidity for the day and hour.
---------------------------------------------------------------------------

\167\ Phoenix and Salt Lake City met the current
PM2.5 NAAQS under current conditions and required no
reduction.
---------------------------------------------------------------------------

In these analyses, the Policy Assessment estimated both
PM2.5 and PM10 light extinction in terms of both
daily maximum 1-hour average values and multi-hour (i.e., 4-hour)
average values for daylight hours. Figure 4-7 and Table 4-6 of the
Policy Assessment displayed the results of the rollback procedures as a
box and whisker plot of daily maximum daylight 1-hour PM2.5
light extinction and the percentage of daily maximum hourly
PM2.5 light extinction values estimated to exceed the CPLs
when just meeting the current

[[Page 3193]]

suite of PM2.5 secondary standards for all 15 areas
considered in the Visibility Assessment (including St. Louis)
(excluding hours with relative humidity greater than 90 percent). These
displays showed that the daily maximum 1-hour average PM2.5
light extinction values in all of the study areas other than the three
western non-California areas were estimated to exceed the high CPL on
about 8 percent up to over 30 percent of the days and to exceed the
middle CPL on about 30 percent up to about 70 percent of the days,
while all areas except Phoenix were estimated to have daily maximum 1-
hour average PM2.5 light extinction values that exceeded the
low CPL on over 15 percent to about 90 percent of the days. Figure 4-8
and Table 4-7 of the Policy Assessment present results based on daily
maximum 4-hour average values. These displays show that the daily
maximum 4-hour average PM2.5 light extinction values in all
of the study areas other than the three western non-California areas
and the two areas in Texas were estimated to exceed the high CPL on
about 4 percent up to over 15 percent of the days and to exceed the
middle CPL on about 15 percent up to about 45 percent of the days,
while all areas except Phoenix were estimated to have daily maximum 4-
hour average PM2.5 light extinction values that exceeded the
low CPL on over 10 percent to about 75 percent of the days. A similar
set of figures and tables were developed in terms of PM10
light extinction (U.S. EPA, 2011a, Figures 4-5 and 4-6, Tables 4-4 and
4-5).
Taking the results of these analyses focusing on 1-hour and 4-hour
maximum light extinction values into account, the Policy Assessment
concluded that the available information in this review clearly called
into question the adequacy of the current suite of PM2.5
standards in the context of public welfare protection from visibility
impairment, primarily in urban areas, and supported consideration of
alternative standards to provide appropriate protection (U.S. EPA,
2011a, p. 4-39). This conclusion was based in part on the large
percentage of days, in many urban areas, that were estimated to have
maximum 1-hour or 4-hour light extinction values that exceed the range
of CPLs identified for consideration under simulations of conditions
that would just meet the current suite of PM2.5 secondary
standards. In particular, for air quality that was simulated to just
meet the current PM2.5 standards, greater than 10 percent of
the days were estimated to have peak light extinction values that
exceed the highest, least protective CPL of 30 dv in terms of
PM2.5 light extinction for 9 of the 15 urban areas, based on
1-hour average values, and would thus likely fail to meet a 90th
percentile-based standard at that level. For these areas, the percent
of days estimated to have maximum 1-hour values that exceed the highest
CPL ranged from over 10 percent to over 30 percent. Similarly, when the
middle CPL of 25 dv was considered, greater than 30 percent up to
approximately 70 percent of the days were estimated to have peak light
extinction that exceeded that CPL in terms of PM2.5 light
extinction, for 11 of the 15 urban areas, based on 1-hour average
values. Based on a 4-hour averaging time, 5 of the areas were estimated
to have at least 10 percent of the days with peak light extinction
exceeding the highest CPL in terms of PM2.5 light
extinction, and 8 of the areas were estimated to have at least 30
percent of the days with peak light extinction exceeding the middle CPL
in terms of PM2.5 light extinction. For the lowest CPL of 20
dv, the percentages of days with 4-hour maximum light extinction
estimated to exceed that CPL are even higher for all cases considered.
Based on all of the above, the Policy Assessment concluded that PM
light extinction estimated to be associated with just meeting the
current suite of PM2.5 secondary standards in many areas
across the country exceeded levels and percentages of days that could
reasonably be considered to be important from a public welfare
perspective (U.S. EPA, 2011a, p. 4-40).
Further, the Policy Assessment concluded that use of the current
indicator of PM2.5 mass, in conjunction with the current 24-
hour and annual averaging times, is clearly called into question for a
national standard intended to protect public welfare from PM-related
visibility impairment (U.S. EPA, 2011a, p. 4-40). This is because such
a standard is inherently variable in the degree of protection provided
because of regional differences in relative humidity and species
composition of PM2.5, which are critical factors in the
relationship between the mix of fine particles in the ambient air and
the associated impairment of visibility. The Policy Assessment noted
that this concern was one of the important elements in the court's
decision to remand the PM2.5 secondary standards set in 2006
to the Agency.
Thus, in addition to concluding that the available information
clearly calls into question the adequacy of the protection against PM-
related visibility impairment afforded by the current suite of
PM2.5 standards, the Policy Assessment also concluded that
it clearly calls into question the appropriateness of each of the
current standard elements: indicator, averaging time, form, and level
(U.S. EPA, 2011a, p. 4-40).
After reviewing the information and analysis in the second draft
Policy Assessment, CASAC concluded that the ``currently available
information clearly calls into question the adequacy of the current
standards and that consideration should be given to revising the suite
of standards to provide increased public welfare protection'' (Samet,
2010d, p. iii). CASAC noted that the detailed estimates of hourly PM
light extinction associated with just meeting the current standards
``clearly demonstrate that current standards do not protect against
levels of visual air quality which have been judged to be unacceptable
in all of the available urban visibility preference studies.'' Further,
CASAC stated, with respect to the current suite of secondary
PM2.5 standards, that ``[T]he levels are too high, the
averaging times are too long, and the PM2.5 mass indicator
could be improved to correspond more closely to the light scattering
and absorption properties of suspended particles in the ambient air''
(Samet, 2010d, p. 9).
After considering the available evidence and the advice of CASAC,
the Administrator concluded at the time of proposal that such
information did provide an appropriate basis to inform a conclusion as
to whether the current standards afford adequate protection against PM-
related visibility impairment in urban areas. The Administrator took
into account the information discussed above with regard to the nature
of PM-related visibility impairment, the results of public perception
surveys on the acceptability of varying degrees of visibility
impairment in urban areas, analyses of the number of days on which peak
1-hour or 4-hour light extinction values are estimated to exceed a
range of candidate protection levels under conditions simulated to just
meet the current standards, and the advice of CASAC. She noted the
clear causal relationship between PM in the ambient air and impairment
of visibility, the evidence from the visibility preference studies, and
the rationale for determining a range of candidate protection levels
based on those studies. She also noted the relatively large number of
days when maximum 1-hour or 4-hour light extinction values were
estimated to exceed the three candidate protection levels, including
the highest level of 30 dv, under the current standards. While
recognizing the limitations in the available information on public

[[Page 3194]]

perceptions of the acceptability of varying degree of visibility
impairment and the information on the number of days estimated to
exceed the CPLs, she concluded that such information provided an
appropriate basis to inform a conclusion as to whether the current
standards provide adequate protection against PM-related visibility
impairment in urban areas. Based on these considerations, and placing
great importance on the advice of CASAC, the Administrator
provisionally concluded that the current standards are not sufficiently
protective of visual air quality, and that consideration should be
given to an alternative secondary standard that would provide
additional protection against PM-related visibility impairment, with a
focus primarily in urban areas.
Having reached this conclusion, the Administrator also stated at
the time of proposal that the current indicator of PM2.5
mass, in conjunction with the current 24-hour and annual averaging
times, is not well suited for a national standard intended to protect
public welfare from PM-related visibility impairment. As noted in the
proposal, the current standards do not incorporate information on the
concentrations of various species within the mix of ambient particles,
nor do they incorporate information on relative humidity, both of which
play a central role in determining the relationship between the mix of
PM in the ambient air and impairment of visibility. Such considerations
were reflected both in CASAC's advice to set a distinct secondary
standard that would more directly reflect the relationship between
ambient PM and visibility impairment and in the court's remand of the
current secondary PM2.5 standards. Based on the above
considerations, at the time of proposal the Administrator provisionally
concluded that the current secondary PM2.5 standards, taken
together, are neither sufficiently protective nor suitably structured
to provide an appropriate degree of public welfare protection from PM-
related visibility impairment, primarily in urban areas. This led the
EPA to consider alternative standards by looking at each of the
elements of the standards--indicator, averaging time, form, and level--
as discussed below.
ii. Indicator
At the time of proposal, the EPA considered three alternative
indicators for a PM2.5 standard designed to protect against
visibility impairment: The current PM2.5 mass indicator;
directly measured PM2.5 light extinction; and calculated
PM2.5 light extinction. Directly measured PM2.5
light extinction is a measurement (or combination of measurements) of
the light absorption and scattering caused by PM2.5 under
ambient conditions. Calculated PM2.5 light extinction uses
the IMPROVE algorithm to calculate PM2.5 light extinction
using measured PM2.5 mass, speciated PM2.5 mass,
and measured relative humidity. The Policy Assessment evaluated each of
these alternatives, finally concluding that consideration should be
given to establishing a new calculated PM2.5 light
extinction indicator (U.S. EPA, 2011a, p. 4-51).
As discussed in section VI.D.1 of the proposal, the Policy
Assessment concluded that consideration of the use of either directly
measured PM2.5 light extinction or calculated
PM2.5 light extinction as an indicator is justified because
light extinction is a physically meaningful measure of the
characteristic of ambient PM2.5 that is most relevant and
directly related to PM-related visibility effects (U.S. EPA, 2011a, p.
4-41). Further, as noted above, PM2.5 is the component of PM
responsible for most of the visibility impairment in most urban areas.
In these areas, the contribution of PM10-2.5 is a minor
contributor to visibility impairment most of the time. The Policy
Assessment also indicated that the available evidence demonstrated a
strong correspondence between calculated PM2.5 light
extinction and PM-related visibility impairment, as well as the
significant degree of variability in visibility protection across the
U.S. allowed by a PM2.5 mass indicator. The Policy
Assessment recognized that while in the future it would be appropriate
to consider a direct measurement of PM2.5 light extinction
it was not an appropriate option in this review because a suitable
specification of the equipment and associated performance verification
procedures cannot be developed in the time frame for this review.
(a) PM2.5 Mass
In terms of utilizing a PM2.5 mass indicator, the
proposal noted that PM2.5 mass monitoring methods are in
widespread use, including the FRM involving the collection of periodic
(usually 1-day-in-6 or 1-day-in-3) 24-hour filter samples. However,
these routine monitoring activities do not include measurement of the
full water content of the ambient PM2.5 that contributes,
often significantly, to visibility impacts. Further, the
PM2.5 mass concentration monitors do not provide information
on the composition of the ambient PM2.5, which plays a
central role in the relationship between PM-related visibility
impairment and ambient PM2.5 mass concentrations. Additional
analyses discussed in the proposal that looked at the contribution of
PM2.5 to total PM-related light extinction (defined in terms
of hourly PM10 calculated light extinction) indicate that
there is a poor correlation between hourly PM10 light
extinction and hourly PM2.5 mass principally due to the
impact of the water content of the particles on light extinction, which
depends on both the composition of the PM2.5 and the ambient
relative humidity. Both composition and especially relative humidity
vary during a single day, as well as from day-to-day, at any site and
time of year. Also, there are systematic regional and seasonal
differences in the distribution of ambient humidity and
PM2.5 composition conditions that make it impossible to
select a PM2.5 concentration that generally would correspond
to the same PM-related light extinction levels across all areas of the
nation. Analyses discussed in the proposal quantify the projected
uneven protection that would result from the use of 1-hour average
PM2.5 mass as the indicator.
(b) Directly Measured PM2.5 Light Extinction
PM light extinction has a nearly one-to-one relationship to light
extinction, unlike PM2.5 mass concentration. As explained
above, PM2.5 is the component responsible for the large
majority of PM light extinction in most places and times.
PM2.5 light extinction can be directly measured using
several instrumental methods, some of which have been used for decades
to routinely monitor the two components of PM2.5 light
extinction (light scattering and absorption) or to jointly measure both
as total light extinction (from which Rayleigh scattering is subtracted
to get PM2.5 light extinction). As noted at the time of
proposal, there are a number of advantages to direct measurements of
light extinction for use in a secondary standard relative to estimates
of PM2.5 light extinction calculated using PM2.5
mass and speciation data. These include greater accuracy of direct
measurements with shorter averaging times and overall greater
simplicity when compared to the need for measurements of multiple
parameters to calculate PM light extinction.
In evaluating whether direct measurement of PM2.5 or
PM10 light extinction is appropriate to consider in the
context of this PM NAAQS review, the EPA solicited comment from the
Ambient Air Monitoring and Methods

[[Page 3195]]

Subcommittee (AAMMS) of CASAC. The CASAC AAMMS recommended that
consideration of direct measurement should be limited to
PM2.5 light extinction, and that although instruments
suitable for this purpose are commercially available at present,
research is expected to produce even better instruments in the near
term. The CASAC AAMMS advised against choosing any currently available
commercial instrument, or even a general measurement approach, as an
FRM because to do so could discourage development of other potentially
superior approaches. Instead, the CASAC AAMMS recommended that the EPA
develop performance-based approval criteria for direct measurement
methods in order to put all approaches on a level playing field.
At the present time, the EPA has not undertaken to develop and test
such performance-base approval criteria. The EPA anticipates that if an
effort were begun it would take at least several years before such
criteria would be ready for regulatory use. Thus, the Policy Assessment
concluded that while in the future it would be appropriate to consider
a direct measurement of PM2.5 light extinction, or the sum
of separate measurements of light scattering and light absorption, as
the indicator for the secondary PM2.5 standard, this is not
an appropriate option in this review because a suitable specification
of the equipment or appropriate performance-based verification
procedures cannot be developed in the time frame for this review (U.S.
EPA, 2011a, p. 4-51, -52).
(c) Calculated PM2.5 Light Extinction
For the reasons discussed above, the Policy Assessment concluded
that a calculated PM2.5 light extinction indicator would be
the preferred approach. PM2.5 light extinction can be
calculated from PM2.5 mass, combined with speciated
PM2.5 mass concentration data plus relative humidity data,
as is presently routinely done on a 24-hour average basis under the
Regional Haze Program using data from the rural IMPROVE monitoring
network. This same calculation procedure, using a 24-hour average
basis, could be used for a NAAQS focused on protecting against PM-
related visibility impairment primarily in urban areas. This approach
would use the type of data that is routinely collected from the urban
CSN \168\ in combination with monthly average relative humidity data
based on long-term climatological means as used in the Regional Haze
Program (U.S. EPA, 2011a, Appendix G, section G.2). The proposal
discussed the complex approach utilized in the Visibility Assessment
for calculating hourly PM2.5 light extinction \169\ and
discussed various simplified approaches for calculating these hourly
values that were analyzed in the Policy Assessment. The Policy
Assessment concluded that each of these simplified approaches provided
reasonably good estimates of PM2.5 light extinction and each
would be appropriate to consider as the indicator for a distinct hourly
or multi-hour secondary standard (U.S. EPA, 2011a, p. 4-48). The
proposal also recognized that the Policy Assessment identified a number
of variations on these simplified approaches that it would be
appropriate to consider, including:
---------------------------------------------------------------------------

\168\ About 200 sites in the CSN routinely measure 24-hour
average PM2.5 chemical components using filter-based
samplers and chemical analysis in a laboratory, on either a 1-day-
in-3 or 1-day-in-6 schedule (U.S. EPA, 2011a, Appendix B, section
B.1.3).
\169\ As noted at the time of proposal, the sheer size of the
ambient air quality, meteorological, and chemical transport modeling
data files involved with the Visibility Assessment approach would
make it very difficult for state agencies or any interested party to
consistently apply such an approach on a routine basis for the
purpose of implementing a national standard defined in terms of the
Visibility Assessment approach.

(1) The use of the split-component mass extinction efficiency
approach from the revised IMPROVE algorithm\170\
---------------------------------------------------------------------------

\170\ If the revised IMPROVE algorithm were used to define the
calculated PM2.5 mass-based indicator, it would not be
possible to algebraically reduce the revised algorithm to a two-
factor version as described above and in Appendix F of the Policy
Assessment for the simplified approaches. Instead, five component
fractions would be determined from each day of speciated sampling,
and then either applied to hourly measurements of PM2.5
mass on the same day or averaged across a month and then applied to
measurements of PM2.5 mass on each day of the month.
---------------------------------------------------------------------------

(2) The use of more refined value(s) for the organic carbon
multiplier \171\
---------------------------------------------------------------------------

\171\ An organic carbon (OC)-to-organic mass (OM) multiplier of
1.6 was used for the assessment, which was found to produce a value
of OM comparable to the one derived with the original, albeit more
complex, Visibility Assessment method.
---------------------------------------------------------------------------

(3) The use of the reconstructed 24-hour PM2.5 mass
(i.e., the sum of the five PM2.5 components from
speciated monitoring) as a normalization value for the hourly
measurements from the PM2.5 instrument as a way of better
reflecting ambient nitrate concentrations
(4) The use of historical monthly or seasonal, or regional,
speciation averages

Overall, the analyses conducted for the Visibility Assessment and
Policy Assessment indicated that the use of a calculated
PM2.5 light extinction indicator would provide a much higher
degree of uniformity in terms of the degree of protection from
visibility impairment across the country than a PM2.5 mass
indicator, because a calculated PM2.5 light extinction
indicator would directly incorporate the effects of humidity and
PM2.5 composition differences between various regions.
Further, the proposal noted that the Policy Assessment concluded that
consideration could be given to defining a calculated PM2.5
light extinction indicator on either a 24-hour or a sub-daily basis
(U.S. EPA, 2011a, p. 4-52). However, the Policy Assessment noted that
approval of continuous FEM monitors has been based only on 24-hour
average, not hourly, PM2.5 mass. In addition, there are
mixed results of data quality assessments on a 24-hour basis for these
monitors, as well as the near absence of performance data for sub-daily
averaging periods. Thus, while it is possible to utilize data from
PM2.5 continuous FEMs on a 1-hour or multi-hour (e.g., 4-
hour) basis, these factors increase the uncertainty of utilizing
continuous methods to support 1-hour or 4-hour PM2.5 mass
measurements as an input to the light extinction calculation.
Therefore, as noted at the time of proposal, until issues regarding the
comparability of 24-hour PM2.5 mass values derived from
continuous FEMs and filter-based FRMs \172\ are resolved, there is
reason to be cautious about relying on a calculation procedure that
uses hourly PM2.5 mass values reported by continuous FEMs in
combination with speciated PM2.5 mass values from 24-hour
filter-based samplers.
---------------------------------------------------------------------------

\172\ Filter-based FRMs are designed to adequately quantify the
amount of PM2.5 collected over 24-hours. They cannot be
presumed to be appropriate for quantifying average concentrations
over 1-hour or 4-hour periods.
---------------------------------------------------------------------------

(d) CASAC Advice
In reviewing the second draft Policy Assessment, CASAC stated that
it ``overwhelmingly * * * would prefer the direct measurement of light
extinction,'' recognizing it as the property of the atmosphere that
most directly relates to visibility effects (Samet, 2010d, p. iii).
CASAC noted that ``[I]t has the advantage of relating directly to the
demonstrated harmful welfare effect of ambient PM on human visual
perception.'' However, CASAC also concluded that the calculated
PM2.5 light extinction indicator ``appears to be a
reasonable approach for estimating hourly light extinction'' (Samet,
2010d, p. 11). Further, based on CASAC's understanding of the time that
would be required to develop an FRM for this indicator, CASAC agreed
with the staff preference presented in the second draft Policy
Assessment for a calculated PM2.5 light extinction
indicator. CASAC noted that ``[I]ts reliance on procedures that

[[Page 3196]]

have already been implemented in the CSN and routinely collected
continuous PM2.5 data suggest that it could be implemented
much sooner than a directly measured indicator'' (Samet, 2010d, p.
iii).\173\
---------------------------------------------------------------------------

\173\ In commenting on the second draft Policy Assessment, CASAC
did not have an opportunity to review the assessment of continuous
PM2.5 FEMs compared to collocated FRMs (Hanley and Reff,
2011) as presented and discussed in the final Policy Assessment
(U.S. EPA, 2011a, p. 4-50).
---------------------------------------------------------------------------

(e) Administrator's Proposed Conclusions on Indicator
At the time of proposal, while agreeing with CASAC that a directly
measured PM light extinction indicator would provide the most direct
link between PM in the ambient air and PM-related light extinction, the
Administrator provisionally concluded that this was not an appropriate
option in this review because a suitable specification of currently
available equipment or performance-based verification procedures cannot
be developed in the time frame of this review. Taking all of the above
considerations and CASAC advice into account, the Administrator
provisionally concluded that a new calculated PM2.5 light
extinction indicator, similar to that used in the Regional Haze Program
(i.e., using an IMPROVE algorithm as translated into the deciview
scale), was the appropriate indicator to replace the current
PM2.5 mass indicator. Such an indicator, referred to as a
PM2.5 visibility index, would appropriately reflect the
relationship between ambient PM and PM-related light extinction, based
on the analyses discussed in the proposal and incorporation of factors
based on measured PM2.5 speciation concentrations and
relative humidity data. In addition, selection of this type of
indicator would address, in part, the issues raised in the court's
remand of the 2006 p.m.2.5 standards. The Administrator also
noted that such a PM2.5 visibility index would afford a
relatively high degree of uniformity of visual air quality protection
in areas across the country by virtue of directly incorporating the
effects of differences in PM2.5 composition and relative
humidity across the country.
Based on these above considerations, the Administrator proposed to
set a distinct secondary standard for PM2.5 defined in terms
of a PM2.5 visibility index (i.e., a calculated
PM2.5 light extinction indicator, translated into the
deciview scale) to protect against PM-related visibility impairment
primarily in urban areas. The Administrator proposed that such an index
be based on the original IMPROVE algorithm in conjunction with monthly
average relative humidity data based on long-term climatological means
as used in the Regional Haze Program. The EPA solicited comment on all
aspects of the proposed indicator, especially:

(1) The proposed use of a PM2.5 visibility index
rather than a PM10 visibility index which would include
an additional term for coarse particles;
(2) Using the revised IMPROVE algorithm rather than the original
IMPROVE algorithm;
(3) The use of alternative values for the organic carbon
multiplier in conjunction with either the original or revised
IMPROVE algorithm;
(4) The use of historical monthly, seasonal, or regional
speciation averages;
(5) Alternative approaches to determining relative humidity; and
(6) Simplified approaches to generating hourly PM2.5
light extinction values for purposes of calculating an hourly or
multi-hour indicator.

iii. Averaging Times
In this review, as discussed in section VI.D.2 of the proposal,
consideration of appropriate averaging times for use in conjunction
with a PM2.5 visibility index was informed by information
related to the nature of PM visibility effects and the nature of inputs
to the calculation of PM2.5 light extinction, as discussed
above. The EPA considered both sub-daily (1- and 4-hour averaging
times) and 24-hour averaging times. In considering sub-daily averaging
times, the EPA has also considered what diurnal periods and ambient
relative humidity conditions would be appropriate to consider in
conjunction with such an averaging time.
As an initial matter, the Policy Assessment considered sub-daily
averaging times. Taking into account what is known from available
studies concerning how quickly people experience and judge visibility
conditions, the possibility that some fraction of the public
experiences infrequent or short periods of exposure to ambient
visibility conditions, and the typical rate of change of the path-
averaged PM light extinction over urban areas, the initial analyses
conducted as part of the Visibility Assessment focused on a 1-hour
averaging time. In its review of the first draft Policy Assessment,
CASAC agreed that a 1-hour averaging time would be appropriate to
consider, noting that PM effects on visibility can vary widely and
rapidly over the course of a day and such changes are almost
instantaneously perceptible to human observers (Samet, 2010c, p. 19).
The Policy Assessment noted that this view related specifically to a
standard defined in terms of a directly measured PM light extinction
indicator, in that CASAC also noted that a 1-hour averaging time is
well within the instrument response times of the various currently
available and developing optical monitoring methods.
However, CASAC also advised that if a PM2.5 mass
indicator were to be used, it would be appropriate to consider
``somewhat longer averaging times--2 to 4 hours--to assure a more
stable instrumental response'' (Samet, 2010c, p. 19). In considering
this advice, the Policy Assessment concluded that since a calculated
PM2.5 light extinction indicator relies in part on measured
PM2.5 mass, it would be appropriate to consider a multi-hour
averaging time on the order of a few hours (e.g. 4-hours). A multi-hour
averaging time might reasonably characterize the visibility effects
experienced by the segment of the population who have access to
visibility conditions often or continuously throughout the day. For
this segment of the population, it may be that their perception of
visual air quality reflects some degree of offsetting an hour with poor
visual air quality with one or more hours of clearer visual conditions.
Further, the Policy Assessment recognized that a multi-hour averaging
time would have the effect of averaging away peak hourly visibility
impairment, which can change significantly from one hour to the next
(U.S. EPA, 2011a, p. 4-53; U.S. EPA, 2010b, Figure 3-12).
In considering either 1-hour or multi-hour averaging times, the
Policy Assessment recognized that no data are available with regard to
how the duration and variation of time a person spends outdoors during
the daytime impacts his or her judgment of the acceptability of
different degrees of visibility impairment. As a consequence, it is not
clear to what degree, if at all, the protection levels found to be
acceptable in the public preference studies would change for a multi-
hour averaging time as compared to a 1-hour averaging time. Thus, the
Policy Assessment concluded that it is appropriate to consider a 1-hour
or multi-hour (e.g., 4-hour) averaging time as the basis for a sub-
daily standard defined in terms of a calculated PM2.5 light
extinction indicator (U.S. EPA, 2011a, p. 4-53).
In addition, as discussed above, some data quality uncertainties
have been observed with regard to hourly data collected by FEMs.
Specifically, as part of the review of data from all continuous FEM
PM2.5 instruments operating at state/local monitoring sites,
the Policy Assessment noted that the occurrence of questionable
outliers in 1-

[[Page 3197]]

hour data submitted to AQS from continuous FEM PM2.5
instruments had been observed at some of these sites (Evangelista,
2011). Some of these outliers were questionable simply by virtue of
their extreme magnitude, as high as 985 [mu]g/m\3\, whereas other
values were questionable because they were isolated to single hours
with much lower values before and after, a pattern that is much less
plausible than if the high concentrations were more sustained.\174\ The
Policy Assessment noted that any current data quality problems might be
resolved in the normal course of monitoring program evolution as
operators become more adept at instrument operation and maintenance and
data validation or by improving the approval criteria and testing
requirements for continuous instruments. Regardless, the Policy
Assessment noted that multi-hour averaging of FEM data could serve to
reduce the effects of such outliers relative to the use of a 1-hour
averaging time.
---------------------------------------------------------------------------

\174\ Similarly questionable hourly data were not observed in
the 2005 to 2007 continuous PM2.5 data used in the
Visibility Assessment, all of which came from early-generation
continuous instruments that had not been approved as FEMs. However,
only 15 sites and instruments were involved in the Visibility
Assessment analyses, versus about 180 currently operating FEM
instruments submitting data to AQS. Therefore, there were more
opportunities for very infrequent measurement errors to be observed
in the larger FEM data set.
---------------------------------------------------------------------------

The Policy Assessment noted that there are significant reasons to
consider using PM2.5 light extinction calculated on a 24-
hour basis to reduce the various data quality concerns described above
with respect to relying on continuous PM2.5 monitoring data.
However, the Policy Assessment recognized that 24 hours is far longer
than the hourly or multi-hour time periods that might reasonably
characterize the visibility effects experienced by various segments of
the population, including both those who do and do not have access to
visibility conditions often or continuously throughout the day. Thus,
the Policy Assessment concluded that the appropriateness of considering
a 24-hour averaging time would depend upon the extent to which PM-
related light extinction calculated on a 24-hour average basis would be
a reasonable and appropriate surrogate for PM-related light extinction
calculated on a sub-daily basis.
To examine this relationship, the EPA conducted comparative
analyses of 24-hour and 4-hour averaging times in conjunction with a
calculated PM2.5 indicator. For these analyses, 4-hour
average PM2.5 light extinction was calculated based on using
the Visibility Assessment approach. The 24-hour average
PM2.5 light extinction was calculated using the original
IMPROVE algorithm and long-term relative humidity conditions to
calculate PM2.5 light extinction. Based on these
analyses,\175\ which are presented and discussed in Appendix G of the
Policy Assessment, scatter plots comparing 24-hour and 4-hour
calculated PM2.5 light extinction were constructed for each
of the 15 cities included in the Visibility Assessment and for all 15
cities pooled together (U.S. EPA, 2011a, Figures G-4 and G-5). Though
there was some scatter around the regression line for each city because
the calculated 4-hour light extinction values included day-specific and
hour-specific influences that are not captured by the simpler 24-hour
approach, these analyses generally showed good correlation between 24-
hour and 4-hour average PM2.5 light extinction, as evidenced
by reasonably high city-specific and pooled R\2\ values, generally in
the range of over 0.6 to over 0.8.\176\ This suggested that
PM2.5 light extinction calculated on a 24-hour basis is a
reasonable and appropriate surrogate to PM2.5 light
extinction calculated on a sub-daily basis.
---------------------------------------------------------------------------

\175\ These analyses are also based on the use of a 90th
percentile form, averaged over 3 years, as discussed below in
section VI.D.3 and in section 4.3.3 of the Policy Assessment (U.S.
EPA, 2011a).
\176\ The EPA staff noted that the R\2\ value (0.44) for Houston
was notably lower than for the other cities.
---------------------------------------------------------------------------

Taking the above considerations and CASAC's advice into account,
the Policy Assessment concluded that it would be appropriate to
consider a 24-hour averaging time, in conjunction with a calculated
PM2.5 light extinction indicator and an appropriately
specified standard level, as discussed below. By using site-specific
daily data on PM2.5 composition and site-specific long-term
relative humidity conditions, this 24-hour average indicator would
provide more consistent protection from PM2.5-related
visibility impairment than would a secondary PM2.5 NAAQS
based only on 24-hour or annual average PM2.5 mass. In
particular, this approach would account for the systematic difference
in humidity conditions between most eastern states and most western
states. The Policy Assessment also concluded that it would also be
appropriate to consider a multi-hour, sub-daily averaging time, for
example a period of 4 hours, in conjunction with a calculated
PM2.5 light extinction indicator and with further
consideration of the data quality issues discussed above. Such an
averaging time, to the extent that data quality issues can be
appropriately addressed, would be more directly related to the short-
term nature of the perception of visibility impairment, short-term
variability in PM-related visual air quality, and the short-term nature
(hourly to multiple hours) of relevant exposure periods for segments of
the viewing public. Such an averaging time would still result in an
indicator that is less sensitive than a 1-hour averaging time to short-
term instrument variability with respect to PM2.5 mass
measurement. In conjunction with consideration of a multi-hour, sub-
daily averaging time, the Policy Assessment concluded that
consideration should be given to including daylight hours only and to
applying a relative humidity screen of approximately 90 percent to
remove hours in which fog or precipitation is much more likely to
contribute to the observed visibility impairment (U.S. EPA, 2011a, p.
4-58). Recognizing that a 1-hour averaging time would be even more
sensitive to data quality issues, including short-term variability in
hourly data from currently available continuous monitoring methods, the
Policy Assessment concluded that it would not be appropriate to
consider a 1-hour averaging time in conjunction with a calculated
PM2.5 light extinction indicator in this review (U.S. EPA,
2011a, p. 4-58).
As noted above, in its review of the first draft Policy Assessment,
CASAC concluded that PM effects on visibility can vary widely and
rapidly over the course of a day and such changes are almost
instantaneously perceptible to human observers (Samet, 2010c, p. 19).
Based in part on this consideration, CASAC agreed that a 1-hour
averaging time would be appropriate to consider in conjunction with a
directly measured PM light extinction indicator, noting that a 1-hour
averaging time is well within the instrument response times of the
various currently available and developing optical monitoring methods.
At that time, CASAC also advised that if a PM2.5 mass
indicator were to be used, it would be appropriate to consider
``somewhat longer averaging times--2- to 4-hours--to assure a more
stable instrumental response'' (Samet, 2010c, p. 19). Thus, CASAC's
advice on averaging times that would be appropriate for consideration
was predicated in part on the capabilities of monitoring methods that
were available for the alternative indicators discussed in the draft
Policy Assessment. CASAC's views on a multi-hour averaging time would
also apply to the calculated PM2.5 light extinction
indicator since hourly PM2.5 mass measurements are also
required for this

[[Page 3198]]

indicator when calculated on a sub-daily basis.
It is important to note that at the time it provided advice on
suitable averaging times, CASAC did not have the benefit of EPA's
subsequent assessment of the data quality issues associated with the
use of continuous FEMs as the basis for hourly PM2.5 mass
measurements. Furthermore, since CASAC only commented on the first and
second drafts of the Policy Assessment, neither of which included
discussion of a calculated PM2.5 indicator based on a 24-
hour averaging time, CASAC did not have a basis to offer advice
regarding a 24-hour averaging time. In addition, the 24-hour averaging
time is not based on consideration of 24-hours as a relevant exposure
period, but rather as a surrogate for a sub-daily period of 4 hours,
which is consistent with CASAC's advice concerning an averaging time
associated with the use of a PM2.5 mass indicator.
Taking into account the information discussed above with regard to
analyses and conclusions presented in the final Policy Assessment the
Administrator recognized that hourly or sub-daily, multi-hour averaging
times, within daylight hours and excluding hours with relative humidity
above approximately 90 percent, are more directly related than a 24-
hour averaging time to the short-term nature of the perception of PM-
related visibility impairment and the relevant exposure periods for
segments of the viewing public. On the other hand, she recognized that
data quality uncertainties have recently been associated with currently
available instruments that would be used to provide the hourly
PM2.5 mass measurements that would be needed in conjunction
with an averaging time shorter than 24-hours. As a result, while the
Administrator recognized the desirability of a sub-daily averaging
time, she had strong reservations about proposing to set a standard at
this time in terms of a sub-daily averaging time.
In considering the information and analyses related to
consideration of a 24-hour averaging time, the Administrator recognized
that the Policy Assessment concluded that PM2.5 light
extinction calculated on a 24-hour averaging basis is a reasonable and
appropriate surrogate for sub-daily PM2.5 light extinction
calculated on a 4-hour average basis. In light of this finding and the
views of CASAC based on its reviews of the first and second drafts of
the Policy Assessment, the Administrator proposed to set a distinct
secondary standard with a 24-hour averaging time in conjunction with a
PM2.5 visibility index.
iv. Form
As discussed in section VI.D.3 of the proposal, the ``form'' of a
standard defines the air quality statistic that is to be compared to
the level of the standard in determining whether the standard is
achieved. The form of the current 24-hour PM2.5 NAAQS is
such that the level of the standard is compared to the 3-year average
of the annual 98th percentile value of the measured indicator. The
purpose in averaging for three years is to provide stability from the
occasional effects of inter-annual meteorological variability that can
result in unusually high pollution levels for a particular year. The
use of a multi-year percentile form, among other things, makes the
standard less subject to the possibility of transient violations caused
by statistically unusual indicator values, thereby providing more
stability to the air quality management process that may enhance the
practical effectiveness of efforts to implement the NAAQS. Also, a
percentile form can be used to take into account the number of times an
exposure might occur as part of the judgment on protectiveness in
setting a NAAQS. For all of these reasons, the Policy Assessment
concluded it would be appropriate to consider defining the form of a
distinct secondary standard in terms of a 3-year average of a specified
percentile air quality statistic (U.S. EPA, 2011a, p. 4-58).
The urban visibility preference studies that provided results
leading to the range of CPLs being considered in this review offer no
information that addresses the frequency of time that visibility levels
should be below those values. Given this lack of information, and
recognizing that the nature of the public welfare effect is one of
aesthetics and/or feelings of well-being, the Policy Assessment
concluded that it would not be appropriate to consider eliminating all
exposures above the level of the standard and that allowing some number
of hours/days with reduced visibility can reasonably be considered
(U.S. EPA, 2011a, p. 4-59). In the Visibility Assessment, 90th, 95th,
and 98th percentile forms were assessed for alternative PM light
extinction standards (U.S. EPA, 2010b, section 4.3.3). In considering
these alternative percentiles, the Policy Assessment noted that the
Regional Haze Program targets the 20 percent most impaired days for
improvements in visual air quality in Federal Class I areas. If
improvement in the 20 percent most impaired days were similarly judged
to be appropriate for protecting visual air quality in urban areas, a
percentile well above the 80th percentile would be appropriate to
increase the likelihood that all days in this range would be improved
by control strategies intended to attain the standard. A focus on
improving the 20 percent most impaired days suggests that the 90th
percentile, which represents the median of the distribution of the 20
percent worst days, would be an appropriate form to consider.
Strategies that are implemented so that 90 percent of days have visual
air quality that is at or below the level of the standard would
reasonably be expected to lead to improvements in visual air quality
for the 20 percent most impaired days. Higher percentile values within
the range assessed could have the effect of limiting the occurrence of
days with peak PM-related light extinction in urban areas to a greater
degree. In considering the limited information available from the
public preference studies, the Policy Assessment found no basis to
conclude that it would be appropriate to consider limiting the
occurrence of days with peak PM-related light extinction in urban areas
to a greater degree.
Another aspect of the form discussed in the proposal for a sub-
daily averaging time was whether to include all daylight hours or only
the maximum daily daylight hour(s). The maximum daily daylight 1-hour
or multi-hour form would be most directly protective of the welfare of
people who have limited, infrequent or intermittent exposure to
visibility during the day (e.g., during commutes), but spend most of
their time without an outdoor view. For such people a view of poor
visibility during their morning commute may represent their perception
of the day's visibility conditions until the next time they venture
outside during daylight, which may be hours later or perhaps the next
day. Other people have exposure to visibility conditions throughout the
day. For those people, it might be more appropriate to include every
daylight hour in assessing compliance with a standard, since it is more
likely that each daylight hour could affect their welfare.
The Policy Assessment did not have information regarding the
fraction of the public that has only one or a few opportunities to
experience visibility during the day, nor did it have information on
the role the duration of the observed visibility conditions has on
wellbeing effects associated with those visibility conditions. However,
it is logical to conclude that people with limited opportunities to
experience visibility conditions on a daily basis

[[Page 3199]]

would experience the entire impact associated with visibility based on
their short-term exposure. The impact of visibility for those who have
access to visibility conditions often or continuously during the day
may be based on varying conditions throughout the day.
In light of these considerations, the analyses conducted as part of
the Visibility Assessment analyses included both the maximum daily hour
and the all daylight hours forms. The Policy Assessment noted that
there is a close correspondence between the level of protection
afforded for all 15 urban areas by a maximum daily daylight 1-hour
approach using the 90th percentile form and an all daylight hours
approach combined with the 98th percentile form (U.S. EPA, 2010b,
section 4.1.4). This suggested that reductions in visibility impairment
required to meet either form of the standard would provide protection
to both fractions of the public (i.e., those with limited opportunities
and those with greater opportunities to view PM-related visibility
conditions). CASAC generally supported consideration of both types of
forms without expressing a preference based on its review of
information presented in the second draft Policy Assessment (Samet,
2010d, p. 11).
In conjunction with a calculated PM2.5 light extinction
indicator and alternative 24-hour or sub-daily (e.g., 4-hour) averaging
times, based on the above considerations, and given the lack of
information on and the high degree of uncertainty over the impact on
public welfare of the number of days with visibility impairment over a
year, the Policy Assessment concluded that it would be appropriate to
give primary consideration to a 90th percentile form, averaged over
three years (U.S. EPA, 2011a, p. 4-60). Further, in the case of a
multi-hour, sub-daily alternative standard, the Policy Assessment
concluded that it would be appropriate to give primary consideration to
a form based on the maximum daily multi-hour period in conjunction with
the 90th percentile form (U.S. EPA, 2011a, p. 4-60). This sub-daily
form would be expected to provide appropriate protection for various
segments of the population, including those with limited opportunities
during a day and those with more extended opportunities over the
daylight hours to experience PM-related visual air quality.
Though CASAC did not provide advice as to a specific form that
would be appropriate, it took note of the alternative forms considered
in that document and encouraged further analyses in the final Policy
Assessment that might help to clarify a basis for selecting from within
the range of forms identified. In considering the available information
and the conclusions in the final Policy Assessment in light of CASAC's
comments, at the time of proposal the Administrator concluded that a
90th percentile form, averaged over 3 years, is appropriate, and
proposed such a form in conjunction with a PM2.5 visibility
index and a 24-hour averaging time.
v. Level
As discussed in section VI.D.4 of the proposal, in considering
appropriate levels for a 24-hour standard defined in terms of a
PM2.5 visibility index and an 90th percentile form, averaged
over 3 years, the Policy Assessment took into account the evidence- and
impact-based considerations discussed above, with a focus on the
results of public perception and attitude surveys related to the
acceptability of various levels of visual air quality and on the
important limitations in the design and scope of such available
studies. The Policy Assessment considered a variety of approaches for
identifying appropriate levels for such a standard, including utilizing
both adjusted and unadjusted CPLs derived from the visibility
preference studies.
The Policy Assessment interpreted the results from the visibility
preferences studies conducted in four urban areas to define a range of
low, middle, and high CPLs for a sub-daily standard (e.g., 1- to 4-hour
averaging time) of 20, 25, and 30 dv, which are approximately
equivalent to PM2.5 light extinction of values of 65, 110,
and 190 Mm^-1. The CASAC generally supported this approach,
noting that the ``EPA staff's approach for translating and presenting
the technical evidence and assessment results is logically conceived
and clearly presented. The 20-30 deciview range of levels chosen by EPA
staff as `Candidate Protection Levels' is adequately supported by the
evidence presented'' (Samet, 2010d, p. 11).\177\ The Policy Assessment
also recognized that to define a range of alternative levels that would
be appropriate to consider for a 24-hour calculated PM2.5
light extinction standard, it would be appropriate to consider whether
some adjustment to these CPLs is warranted since these preference
studies cannot be directly interpreted as applying to a 24-hour
exposure period (as noted above and in Policy Assessment section
4.3.1). Considerations related to such adjustments are more
specifically discussed below.
---------------------------------------------------------------------------

\177\ In 2009, the DC Circuit remanded the secondary
PM2.5 standards to the EPA in part because the Agency
failed to identify a target level of protection, even though EPA
staff and CASAC had identified a range of target levels of
protection that were appropriate for consideration. The court
determined that the Agency's failure to identify a target level of
protection as part of its final decision was contrary to the statute
and therefore unlawful, and that it deprived EPA's decision-making
of a reasoned basis. See 559F. 3d at 528-31; see also section VI.A.2
above and the Policy Assessment, section 4.1.2.
---------------------------------------------------------------------------

In considering alternative levels for a sub-daily standard based
directly on the four preference study results, the Policy Assessment
noted that the individual low and high CPLs are in fact generally
reflective of the results from the Denver and Washington, DC studies
respectively, and the middle CPL is very near to the 50th percentile
criteria result from the Phoenix study, which was by far the best of
the studies, providing somewhat more support for the middle CPL.
In considering the results from the four visibility preference
studies, the Policy Assessment recognized that currently available
studies are limited in that they were conducted in only four areas,
three in the U.S. and one in Canada. Further, the Policy Assessment
recognized that available studies provide no information on how the
duration and variation of time a person spends outdoors during the
daytime may impact their judgment of the acceptability of different
degrees of visibility impairment. As such, there is a relatively high
degree of uncertainty associated with using the results of these
studies to inform consideration of a national standard for any specific
averaging time. Nonetheless, the Policy Assessment concluded, as did
CASAC, that these studies are appropriate to use for this purpose (U.S.
EPA, 2011a, p. 4-61).
Using approaches described in section VI.C.4 of the proposal, the
Policy Assessment explored various approaches to adjusting the CPLs
derived from the preference studies to inform alternative levels for a
24-hour standard. These various approaches, based on analyses of 2007-
2009 data from the 15 urban areas assessed in the Visibility
Assessment, focused on estimating CPLs for a 24-hour standard that
would provide generally equivalent protection as that provided by a 4-
hour standard with CPLs of 20, 25, and 30 dv. In conducting these
analyses, staff initially expected that the values of 24-hour average
PM2.5 light extinction and daily maximum daylight 4-hour
average PM2.5 light extinction would differ on any given
day, with the shorter term peak value generally being larger. This
would mean that, in concept, the level of a 24-hour standard should
include a

[[Page 3200]]

downward adjustment compared to the level of a 4-hour standard to
provide generally equivalent protection. As discussed more fully in
section G.5 of Appendix G and summarized below, this initial
expectation was not found to be the case across the range of CPLs
considered. In fact, as shown in Tables G-7 and G-8 of Appendix G and
in the corrected version of Table G-6 found in Frank et al.
(2012b),\178\ in considering estimates aggregated or averaged over all
15 cities as well as the range of city-specific estimates for the
various approaches considered, these analyses indicated that the
generally equivalent 24-hour levels ranged from somewhat below the 4-
hour level to just above the 4-hour level for each of the CPLs.\179\ In
all cases, the range of city-specific estimates of generally equivalent
24-hour levels included the 4-hour level for each of the CPLs of 20,
25, and 30 dv. As noted in the proposal, looking more broadly at these
results could support consideration of using the same CPL for a 24-hour
standard as for a 4-hour standard, recognizing that there is no one
approach that can most closely identify a generally equivalent 24-hour
standard level in each urban area for each CPL. The use of such an
unadjusted CPL for a 24-hour standard would place more emphasis on the
relatively high degree of spatial and temporal variability in relative
humidity and fine particle composition observed in urban areas across
the country, so as to reduce the potential of setting a 24-hour
standard level that would require more than the intended degree of
protection in some areas.
---------------------------------------------------------------------------

\178\ Note that the city-specific ranges shown in Table G-6,
Appendix G of the Policy Assessment are incorrectly stated for
Approaches C and E. Drawing from the more detailed and correct
results for Approaches C and E presented in Tables G-7 and G-8,
respectively, the city-specific ranges in Table G-6 for Approach C
should be 17-21 dv for the CPL of 20 dv; 21-25 dv for the CPL of 25
dv; and 24-30 dv for the CPL of 30 dv; the city-specific ranges in
Table G-6 for Approach E should be 17-21 dv for the CPL of 20 dv;
21-26 dv for the CPL of 25 dv; and 25-31 dv for the CPL of 30 dv. In
the EPA's reanalysis comparing 4- vs. 24-hour values, Frank et al.
(2012b) recreated Table G-6 using the correct values from Tables G-7
and G-8.
\179\ As discussed in more detail in Appendix G of the Policy
Assessment, some days have higher values for 24-hour average light
extinction than for daily maximum 4-hour daylight light extinction,
and consequently an adjusted ``equivalent'' 24-hour CPL can be
greater than the original 4-hour CPL. This can happen for two
reasons. First, the use of monthly average historical RH data will
lead to cases in which the f(RH) values used for the calculation of
24-hour average light extinction are higher than all or some of the
four hourly values of f(RH) used to determine daily maximum 4-hour
daylight light extinction on the same day. Second, PM2.5
concentrations may be greater during non-daylight periods than
during daylight hours.
---------------------------------------------------------------------------

In considering the appropriate level of a secondary standard
focused on protection from PM-related urban visibility impairment based
on either a 24-hour or a multi-hour, sub-daily (e.g., 4-hour) averaging
time, the EPA has been mindful of the important limitations in the
available evidence from public preference studies. These uncertainties
and limitations are due in part to the small number of stated
preference studies available for this review; the relatively small
number of study participants and the extent to which the study
participants may not be representative of the broader study area
population in some of the studies; and the variations in the specific
materials and methods used in each study such as scene characteristics,
the range of VAQ levels presented to study participants, image
presentation methods and specific wording used to frame the questions
used in the group interviews. In addition the EPA has noted that the
scenic vistas available on a daily basis in many urban areas across the
country generally may not have the inherent visual interest or the
distance between viewer and object of greatest intrinsic value as in
the Denver and Phoenix preference studies, and that there is the
possibility that there could be regional differences in individual
preferences for VAQ.
It is also important to note that as in past reviews, the EPA is
considering a national visibility standard in conjunction with the
Regional Haze Program as a means of achieving appropriate levels of
protection against PM-related visibility impairment in urban, non-
urban, and Federal Class I areas across the country. The EPA recognizes
that programs implemented to meet a national standard focused primarily
on the visibility problems in urban areas can be expected to improve
visual air quality in surrounding non-urban areas as well, as would
programs now being developed to address the requirements of the
Regional Haze Program established for protection of visual air quality
in Federal Class I areas. The EPA also believes that the development of
local programs, such as those in Denver and Phoenix, can continue to be
an effective and appropriate approach to provide additional protection,
beyond that afforded by a national standard, for unique scenic
resources in and around certain urban areas that are particularly
highly valued by people living in those areas.
The Policy Assessment concluded that it is appropriate to give
primary consideration to alternative standard levels toward the upper
end of the ranges identified above for 24-hour and sub-daily standards,
respectively (U.S. EPA, 2011a, p. 4-63). Thus, the Policy Assessment
concluded it is appropriate to consider the following alternative
levels: A level of 28 dv or somewhat below, down to 25 dv, for a
standard defined in terms of a calculated PM2.5 light
extinction indicator, a 90th percentile form, and a 24-hour averaging
time; and a standard level of 30 dv or somewhat below, down to 25 dv,
for a similar standard but with a 4-hour averaging time (U.S. EPA,
2011a, p. 4-63). The Policy Assessment judged that such standards would
provide appropriate protection against PM-related visibility impairment
primarily in urban areas. The Policy Assessment noted that CASAC
generally supported consideration of the 20-30 dv range as CPLs and,
more specifically, that support for consideration of the upper part of
the range of the CPLs derived from the public preference studies was
expressed by some CASAC Panel members during the public meeting on the
second draft Policy Assessment. The Policy Assessment concluded that
such a standard would be appropriate in conjunction with the Regional
Haze Program to achieve appropriate levels of protection against PM-
related visibility impairment in areas across the country (U.S. EPA,
2011a, p. 4-63).
Based on the considerations discussed above and in section VI.D.4
of the proposal, and taking into account the advice of CASAC, at the
time of proposal the Administrator concluded that it would be
appropriate to establish a target level of protection--for a standard
defined in terms of a PM2.5 visibility index; a 90th
percentile form averaged over 3 years; and a 24-hour averaging time--
equivalent to the protection afforded by such a sub-daily (i.e., 4-
hour) standard at a level of 30 dv, which is the upper end of the range
of CPLs identified in the Policy Assessment and generally supported by
CASAC. More specifically, the Administrator provisionally concluded
that a 24-hour level of either 30 dv or 28 dv could be construed as
providing such a degree of protection, and that either level was
supported by the available information and was generally consistent
with the advice of CASAC. Thus, the EPA proposed two options for the
level of a new 24-hour standard (defined in terms of a PM2.5
visibility index and a 90th percentile form, averaged over 3 years) to
provide appropriate protection from PM-related visibility impairment:
Either 30 dv or 28 dv. As noted in the proposal, the option of setting
such a 24-hour standard at a level of 30 dv would reflect recognition
that there is considerable spatial and

[[Page 3201]]

temporal variability in the key factors that determine the value of the
PM2.5 visibility index in any given urban area, such that
there is a relatively high degree of uncertainty as to the most
appropriate approach to use in selecting a 24-hour standard level that
would be generally equivalent to a specific 4-hour standard level.
Selecting a 24-hour standard level of 30 dv would reflect a judgment
that such substantial degrees of variability and uncertainty should be
reflected in a higher standard level than would be appropriate if the
underlying information were more consistent and certain. Alternatively,
the option of setting such a 24-hour standard at a level of 28 dv would
reflect placing more weight on statistical analyses of aggregated data
from across the study cities and not placing as much emphasis on the
city-to-city variability as a basis for determining an appropriate
degree of protection on a national scale.
The information available for the Administrator to consider when
setting the secondary PM standard raises a number of uncertainties.
While CASAC supported moving forward with a new standard on the basis
of the available information, CASAC also recognized these
uncertainties, referencing the discussion of key uncertainties and
areas for future research in the second draft of the Policy Assessment.
In discussing areas of future research, CASAC stated that: ``The range
of 50% acceptability values discussed as possible standards are based
on just four studies (Figure 4-2), which, given the large spread in
values, provide only limited confidence that the benchmark candidate
protection levels cover the appropriate range of preference values.
Studies using a range of urban scenes (including, but not limited to,
iconic scenes--``valued scenic elements'' such as those in the
Washington, DC study), should also be considered'' (Samet, 2010d, p.
12). The EPA solicited comment on how the Administrator should weigh
those uncertainties as well as any additional comments and information
to inform her consideration of these uncertainties.
In addition, the EPA solicited comment on a number of other issues
related to the level of the standard, including:

(1) Both of the proposed levels and the various approaches to
identifying generally equivalent levels upon which the alternative
proposed levels are based.
(2) A broader range of levels down to 25 dv in conjunction with
a 24-hour averaging time.
(3) A range of alternative levels from 30 to 25 dv in
conjunction with a sub-daily (e.g., 4-hour) averaging time.
(4) The strengths and limitations associated with the public
preference studies and the use of these studies to inform the
selection of a range of levels that could be used to provide an
appropriate degree of public welfare protection when combined with
the other elements of the standard (i.e. indicator, form and
averaging time).
(5) Specific aspects of the public preference studies, including
the extent to which the 50 percent acceptability criterion is an
appropriate basis for establishing target protection levels in the
context of establishing a distinct secondary NAAQS to address PM-
related visibility impairment in urban areas; how the variability
among preference studies in the extent to which study participants
may be representative of the broader study area population should be
weighed in the context of considering these studies in reaching
proposed conclusions on a distinct secondary NAAQS; and the extent
to which the ranges of VAQ levels presented to participants in each
of the studies may have influenced study results and on how this
aspect of the study designs should appropriately be weighed in the
context of considering these studies in the context of this review.
vi. Administrator's Proposed Conclusions Regarding PM Standards To
Protect Visibility
At the time of proposal, based on the considerations described
above, the Administrator proposed to revise the suite of secondary PM
standards by adding a distinct standard for PM2.5 to address
PM-related visibility impairment, focused primarily on visibility in
urban areas. This proposed visibility standard was to be defined in
terms of a PM2.5 visibility index, which would use measured
PM2.5 mass, combined with PM2.5 speciation data
and relative humidity data, to calculate PM2.5 light
extinction, translated into the deciview (dv) scale; a 24-hour
averaging time; a 90th percentile form, averaged over 3 years; and a
level of 28-30 dv.
vii. Related Technical Analysis
At the time of proposal, the EPA conducted a two-pronged technical
analysis of the relationships between the proposed PM2.5
visibility index standard and the current 24-hour PM2.5
mass-based standard (Kelly, et al., 2012a). This analysis was designed
to provide technical information to inform key issues related to
implementing a distinct secondary standard for visibility as proposed.
Specifically, the EPA recognized that significant technical issues were
likely to arise for new or modified emissions sources conducting air
quality analyses for purposes of demonstrating that they would not
cause or contribute to a violation of the visibility standard under the
Prevention of Significant Deterioration (PSD) program. Such a
demonstration for the proposed secondary PM2.5 visibility
index standard could require each PSD applicant to predict, via air
quality modeling, the increase in visibility impairment, in terms of
the proposed PM2.5 visibility index, that would result from
the proposed source's emissions in conjunction with an assessment of
existing air quality (visibility impairment) conditions in terms of the
proposed PM2.5 visibility index. The EPA noted that if this
demonstration were to be attempted using the six-step procedure that
the EPA proposed to use for calculating PM2.5 visibility
index design values from monitored air concentrations of
PM2.5 components, significant technical issues with the
modeling procedures could arise.
To address these technical issues, the EPA sought to explore
whether sources that met the requirements pertaining to the 24-hour
mass-based standard of 35 [micro]g/m\3\ would also meet the
requirements pertaining to the proposed visibility index standard. As
described in Kelly et al. (2012a), the first prong of the analysis
addressed aspects of a PSD significant impact analysis by evaluating
whether an individual source's impact resulting in a small increase in
the ambient PM2.5 concentration would produce a comparably
small increase in visibility impairment. This analysis included
estimates of PM2.5 speciation profiles based on direct
PM2.5 emission profiles for a broad range of source
categories and for theoretical upper and lower bound scenarios.
The second prong of the analysis addressed aspects of a PSD
cumulative impact analysis by exploring the relationship between the
three-year design values for the existing 24-hour PM2.5
standard and coincident design values for the proposed PM2.5
visibility index standard based on recent air quality data. This aspect
of the analysis indicated that increases in 24-hour PM2.5
design values generally correspond to increases in visibility index
design values, and vice-versa. The analysis further explored the
appropriateness of using a demonstration that a source does not cause
or contribute to a violation of the 24-hour PM2.5 standard
as a surrogate for a demonstration that a source does not cause or
contribute to a violation of the proposed secondary PM2.5
visibility index standard. This analysis was based on 2008 to 2010 air
quality data, and compared the proposed level of 35 [mu]g/m\3\ for the
24-hour PM2.5 standard and for illustrative purposes an
alternative standard level of 12 [micro]g/m\3\ for the annual
PM2.5 standard with the

[[Page 3202]]

proposed levels of 28 or 30 dv for the secondary PM2.5
visibility index standard with a 24-hour averaging time and a 90th
percentile form. The results indicated that all (for the 30 dv level)
or nearly all (for the 28 dv level) areas in attainment of the 24-hour
PM2.5 standard would also have been in attainment of the
proposed secondary PM2.5 visibility index standard.
Based on this technical analysis, the EPA proposed that there is
sufficient evidence that a demonstration that a source does not cause
or contribute to a violation of the mass-based 24-hour PM2.5
standard serves as a suitable surrogate for demonstrating that a source
does not cause or contribute to a violation of the proposed secondary
24-hour PM2.5 visibility index standard under the PSD
program. As such, the EPA proposed to conclude that many or all sources
undergoing PSD review for PM2.5 could rely upon their
analysis for demonstrating that they do not cause or contribute to a
violation of the mass-based 24-hour PM2.5 standard to also
show that they do not cause or contribute to a violation of the
proposed secondary PM2.5 visibility index standard, if a
distinct visibility standard were finalized.
Although this proposed ``surrogacy policy'' was designed to address
an implementation-related issue, the second prong of the technical
analysis addresses the broader technical question of the relationship
between the existing 24-hour PM2.5 standard and the proposed
PM2.5 visibility index standard in terms of the degree of
protection likely to be afforded by each standard. Specifically, the
analysis indicated that depending on the level of the proposed
PM2.5 visibility index standard, the existing 24-hour
PM2.5 mass-based standard would be as protective or in some
areas more protective of visibility than a distinct secondary standard
set within the range of levels proposed. Commenters on the proposed
PM2.5 visibility index explored the implications of this
analysis at length, as discussed further below in section VI.C.1.f. For
this reason, the analysis is described in some detail here.
Kelly et al. (2012a) noted that the relationship between design
values for the 24-hour PM2.5 standard and the proposed
secondary visibility index standard is not obvious a priori because of
differences in design value calculations for the standards. However,
closer examination of this relationship indicated that increases or
decreases in 24-hour PM2.5 design values correspond,
respectively, to increases or decreases in visibility index values.
Specifically, based on measurements from 102 sites with complete data
from 2008-2010, Kelly et al. (2012a) found linear correlations between
the 24-hour PM2.5 design values and the visibility index
design values with r\2\ values ranging from 0.65 to 0.98 across these
sites, with an average r\2\ value of 0.75 across all U.S. sites.
Moreover, the data indicated that no design value existed where the
visibility index design value exceeded 30 dv, but the 24-hour
PM2.5 standard level of 35 [micro]g/m\3\ was attained.
Visibility index design values for certain sites in the Industrial
Midwest were shown to exceed 28 dv despite the fact that the 24-hour
PM2.5 design values for these sites were below 35 [micro]g/
m\3\. This was attributed to the combination of high nitrate and
sulfate fractions, substantial RH adjustment factors, and
PM2.5 distribution characteristics that led to relatively
high visibility index design values for a given 24-hour
PM2.5 design value for counties in the Industrial
Midwest.\180\ Kelly et al. (2012a) concluded that the ``overall, design
values based on 2008-2010 data suggest that counties that attain 24-
hour PM2.5 NAAQS level of 35 [micro]g/m\3\ would attain the
proposed secondary PM2.5 visibility index NAAQS level of 30
dv and generally attain the level of 28 dv'' (pp. 17-18). In addition,
the Kelly et al. analysis indicated that at sites that violated both
the 24-hour PM2.5 level and the proposed visibility index 30
dv level, the proposed level of 30 dv would likely be attained if
PM2.5 concentrations were reduced such that the 24-hour
PM2.5 level of 35 [micro]g/m\3\ was attained (Kelly et al.,
2012a, p.15).\181\ A key implication of this analysis, therefore, was
that within the range of levels proposed by the EPA for a visibility
index standard (28-30 dv), the 24-hour PM2.5 standard of 35
[micro]g/m\3\ would be controlling in almost all (at 28 dv) or all (at
30 dv) instances.
---------------------------------------------------------------------------

\180\ Kelly et al. (2012a) also noted that ``Regional reductions
in sulfate PM2.5 due to emission controls planned as part
of national rules as well as emission reductions associated with
potential annual standard violations are expected to improve
visibility in this region'' (p. 17).
\181\ The analysis also showed that attaining the 24-hour
PM2.5 standard level of 35 [micro]g/m\3\ would result in
achieving a lower PM2.5 visibility index level in certain
areas of the country, largely western areas, than would be achieved
in other areas of the country. This is due to differences in the
composition of ambient PM2.5 and the lower relative
humidity in those areas.
---------------------------------------------------------------------------

2. Other (Non-Visibility) PM-related Welfare Effects
In the 2006 review, the EPA concluded that there was insufficient
information to consider a distinct secondary standard based on PM-
related impacts to ecosystems, materials damage and soiling, and
climatic and radiative processes (71 FR 61144, October 17, 2006).
Specifically, there was a lack of evidence linking various non-
visibility welfare effects to specific levels of ambient PM. In that
review, to provide a level of protection for these welfare-related
effects, the secondary standards were set equal to the revised primary
standards to directionally improve the level of protection afforded
vegetation, ecosystems, and materials (71 FR 61210, October 17, 2006).
This section briefly outlines key conclusions discussed more fully
in section VI.E of the proposal regarding the non-visibility welfare
effects of PM. These conclusions relate to the climate, ecological
(including effects on plants, soil and nutrient cycling, wildlife and
water) and materials damage effects of PM. For all of these effects,
the Policy Assessment concluded that there is insufficient information
at this time to revise the current suite of secondary standards. It is
important to note that the Policy Assessment explicitly excluded
discussion of the effects associated with deposited particulate matter
components of NOX and SOx and their
transformation products which are addressed fully in the joint review
of the secondary NO2 and SO2 NAAQS.
a. Evidence of Other Welfare Effects Related to PM
With regard to the role of PM in climate, the proposal noted that
there is considerable ongoing research focused on understanding aerosol
contributions to changes in global mean temperature and precipitation
patterns. The Integrated Science Assessment concluded ``that a causal
relationship exists between PM and effects on climate, including both
direct effects on radiative forcing and indirect effects that involve
cloud feedbacks that influence precipitation formation and cloud
lifetimes'' (U.S. EPA, 2009a, section 9.3.10). These effects are
discussed in more detail in section VI.E.1 of the proposal, which
provides information on the major aerosol components of interest for
climate processes, including black carbon (BC), organic carbon (OC),
sulfates, nitrates, and mineral dusts, and the nature, magnitude, and
direction (e.g., cooling vs. warming) of various aerosol impacts on
climate.\182\ The Policy Assessment concluded that aerosols alter
climate processes directly through radiative forcing and by indirect
effects on cloud brightness, changes in precipitation, and

[[Page 3203]]

possible changes in cloud lifetimes (U.S. EPA, 2011a, p. 5-10).
Further, the Policy Assessment noted that the major aerosol components
that contribute to climate processes (i.e. BC, OC, sulfate, nitrate and
mineral dusts) vary in their reflectivity, forcing efficiencies and
even in the direction of climate forcing, though there is an overall
net climate cooling associated with aerosols in the global atmosphere
(U.S. EPA, 2009a, section 9.2.10). The Policy Assessment concluded that
the current mass-based PM2.5 and PM10 secondary
standards were not an appropriate or effective means of focusing
protection against PM-associated climate effects due to these
differences in components (U.S. EPA, 2011a, p. 5-11). In addition, in
light of the significant uncertainties in current scientific
information and the lack of sufficient data, the Policy Assessment
concluded it is not currently feasible to conduct a quantitative
analysis for the purpose of informing revisions of the current
secondary PM standards based on climate (U.S. EPA, 2011a, p. 5-11).
Overall the Policy Assessment concluded that there is insufficient
information at this time to base a national ambient standard on climate
impacts associated with current ambient concentrations of PM or its
constituents (U.S. EPA, 2011a, p. 5-11, -12).\183\
---------------------------------------------------------------------------

\182\ Atmospheric PM is referred to as aerosols in the remainder
of this section to be consistent with the Integrated Science
Assessment.
\183\ This conclusion would apply for both the secondary
(welfare-based) and the primary (health-based) standards.
---------------------------------------------------------------------------

With regard to ecological effects, the proposal noted that several
ecosystem components (e.g., plants, soils and nutrient cycling,
wildlife and water) are impacted by PM air pollution, which may alter
the services provided by affected ecosystems. Ecological effects
include both direct effects due to deposition (e.g., wet, dry or
occult) to vegetation surfaces and indirect effects occurring via
deposition to ecosystem soils or surface waters where the deposited
constituents of PM then interact with biological organisms. Some of the
ecological effects considered in this review include direct effects to
metabolic processes of plant foliage; contribution to total metal
loading resulting in alteration of soil biogeochemistry and
microbiology, and plant and animal growth and reproduction; and
contribution to total organics loading resulting in bioaccumulation and
biomagnification across trophic levels. Section VI.E.2 of the proposal
summarizes key findings related to:

(1) Impacts on plants and the ecosystem services they provide
due to deposition of PM to vegetative surfaces, which alters the
radiation received by the plant, and uptake of deposited PM
components by plants from soil or foliage, which can lead to stress
and decreased photosynthesis;
(2) Impacts on ecosystem support services such as nutrient
cycling, products such as crops and the regulation of flooding and
water quality;
(3) Impacts on wildlife, especially due to biomagnification of
heavy metals (especially Hg) up the food chain and bioconcentration
of POPs and PBDEs; and
(4) Impacts of deposited PM, especially metals and organics, on
the ecosystem services provided by water bodies, including primary
production, provision of fresh water, regulation of climate and
floods, recreational fishing and water purification.

The proposal noted that the Integrated Science Assessment had
concluded that ecological evidence is sufficient to conclude that a
causal relationship is likely to exist between deposition of PM and a
variety of effects on individual organisms and ecosystems (U.S. EPA,
2009a, sections 2.5.3 and 9.4.7), and also noted that vegetation and
other ecosystem components are affected more by particulate chemistry
than size fraction. However, the proposal also pointed to the
Integrated Science Assessment conclusion that it is generally difficult
to characterize the nature and magnitude of effects and to quantify
relationships between ambient concentrations of PM and ecosystem
response due to significant data gaps and uncertainties as well as
considerable variability that exists in the components of PM and their
various ecological effects. There are few studies that link ambient PM
concentrations to observed effect. Most direct ecosystem effects
associated with particulate pollution occur in severely polluted areas
near industrial point sources (quarries, cement kilns, metal smelting)
(U.S. EPA, 2009a, sections 9.4.3 and 9.4.5.7).
Based on the evidence available at this time, the proposal noted
the following key conclusions in the Policy Assessment:

(1) A number of significant environmental effects that either
have already occurred or are currently occurring are linked to
deposition of chemical constituents found in ambient PM.
(2) Ecosystem services can be adversely impacted by PM in the
environment, including supporting, provisioning, regulating and
cultural services.
(3) The lack of sufficient information to relate specific
ambient concentrations of particulate metals and organics to a
degree of impairment of a specific ecological endpoint hinders the
identification of a range of appropriate indicators, levels, forms
and averaging times of a distinct secondary standard to protect
against associated effects.
(4) Data from regionally-based ecological studies can be used to
establish probable local, regional and/or global sources of
deposited PM components and their concurrent effects on ecological
receptors.

The proposal noted that the Policy Assessment had concluded that
the currently available information is insufficient for purposes of
assessing the adequacy of the protection for ecosystems afforded by the
current suite of PM secondary standards or establishing a distinct
national standard for ambient PM based on ecosystem effects of
particulates not addressed in the NOX/SOX
secondary review (e.g., metals, organics) (U.S. EPA, 2011a, p. 5-24).
Furthermore, the Policy Assessment had concluded that in the absence of
information providing a basis for specific standards in terms of
particle composition, the observations continue to support retaining an
appropriate degree of control on both fine and coarse particles to help
address effects to ecosystems and ecosystem components associated with
PM (U.S. EPA, 2011a, p. 5-24).
With regard to materials damage, the proposal discussed effects
associated with deposition of PM, including both physical damage
(materials damage effects) and impaired aesthetic qualities (soiling
effects). As with the other categories of welfare effects discussed
above, the Integrated Science Assessment concluded that evidence is
sufficient to support a causal relationship between PM and effects on
materials (U.S. EPA, 2009a, sections 2.5.4 and 9.5.4). The deposition
of PM can physically affect materials, adding to the effects of natural
weathering processes, by potentially promoting or accelerating the
corrosion of metals, by degrading paints and by deteriorating building
materials such as stone, concrete and marble (U.S. EPA, 2009a, section
9.5). In addition, the deposition of ambient PM can reduce the
aesthetic appeal of buildings and objects through soiling. The Policy
Assessment made the following observations:

(1) Materials damage and soiling that occur through natural
weathering processes are enhanced by exposure to atmospheric
pollutants, most notably sulfur dioxide and particulate sulfates.
(2) While ambient particles play a role in the corrosion of
metals and in the weathering of materials, no quantitative
relationships between ambient particle concentrations and rates of
damage have been established.
(3) While soiling associated with fine and course particles can
result in increased cleaning frequency and repainting of surfaces,
no quantitative relationships between particle characteristics and
the frequency of cleaning or repainting have been established.
(4) Limited new data on the role of microbial colonizers in
biodeterioration

[[Page 3204]]

processes and contributions of black crust to soiling are not
sufficient for quantitative analysis.
(5) While several studies in the PM Integrated Science
Assessment and NOX/SOX Integrated Science
Assessment suggest that particles can promote corrosion of metals
there remains insufficient evidence to relate corrosive effects to
specific particulate levels or to establish a quantitative
relationship between ambient PM and metal degradation. With respect
to damage to calcareous stone, numerous studies suggest that wet or
dry deposition of particles and dry deposition of gypsum particles
can enhance natural weathering processes.

The Policy Assessment concluded that none of the new evidence in
this review called into question the adequacy of the current standards
for protecting against material damage effects, that such effects could
play no quantitative role in determining whether revisions to the
secondary PM NAAQS are appropriate at this time, and that observations
continue to support retaining an appropriate degree of control on both
fine and coarse particles to help address materials damage and soiling
associated with PM (U.S. EPA, 2011a, p. 5-29).
b. CASAC Advice
In advising the EPA regarding the non-visibility welfare effects,
CASAC stated that it ``concurs with the Policy Assessment's conclusions
that while these effects are important, and should be the focus of
future research efforts, there is not currently a strong technical
basis to support revisions of the current standards to protect against
these other welfare effects'' (Samet, 2010c). More specifically, with
regard to climate impacts, CASAC concluded that while there is
insufficient information on which to base a national standard, the
causal relationship is established and the risk of impacts is high, so
further research on a regional basis is urgently needed (Samet, 2010c,
p. 5). CASAC also noted that reducing certain aerosol components could
lead to increased radiative forcing and regional climate warming while
having a beneficial effect on PM-related visibility. As a consequence,
CASAC noted that a secondary standard directed toward reducing PM-
related visibility impairment has the potential to be accompanied by
regional warming if light scattering aerosols are preferentially
targeted.
With regard to ecological effects, CASAC concluded that the
published literature is insufficient to support a national standard for
PM effects on ecosystem services (Samet, 2010c, p.23). CASAC noted that
the best-established effects are related to particles containing
nitrogen and sulfur, which are being considered in the EPA's ongoing
review of the secondary NAAQS for NOX/SOX. With
regard to PM-related effects on materials, CASAC concluded that the
published literature, including literature published since the last
review, is insufficient either to call into question the current level
of the standard or to support any specific national standard for PM
effects on materials (Samet, 2010c, p.23). Nonetheless, with regard to
both types of effects, CASAC noted the importance of maintaining an
appropriate degree of control of both fine and coarse particles to
address such effects, even in the current absence of sufficient
information to develop a standard.
c. Summary of Proposed Decisions Regarding Other Welfare Effects
Based on the above considerations and the advice of CASAC, at the
time of proposal the Administrator provisionally concluded that it
would not be appropriate to establish any distinct secondary PM
standards to address other non-visibility PM-related welfare effects,
including ecological effects, effects on materials, and climate
impacts. Nonetheless, the Administrator concurred with the conclusions
of the Policy Assessment and CASAC advice that it is important to
maintain an appropriate degree of control of both fine and coarse
particles to address such effects. Noting that there is an absence of
information that would support any different standards, the
Administrator proposed generally to retain the current suite of
secondary PM standards \184\ to address non-visibility welfare effects.
Specifically, the Administrator proposed to retain all aspects of the
current secondary 24-hour PM2.5 and PM10
standards. With regard to the secondary annual PM2.5
standard, the Administrator proposed to retain the level of the current
standard and to revise the form of the standard by removing the option
for spatial averaging consistent with this change to the primary annual
PM2.5 standard.
---------------------------------------------------------------------------

\184\ As summarized in section VI.A and Table 1 above, the
current suite of secondary PM standards includes annual and 24-hour
PM2.5 standards and a 24-hour PM10 standard.
---------------------------------------------------------------------------

C. Public Comments on Proposed Decisions Regarding Secondary PM
Standards

The EPA received a large number of comments on its proposed
decisions with regard to secondary PM standards, with the large
majority of those comments focusing on the proposal to set a distinct
standard to protect against visibility impairment, discussed below in
section VI.C.1. Very few commenters addressed the proposal to retain
the existing secondary standards for non-visibility welfare effects,
discussed below in section VI.C.2. As discussed in section VI.D. below,
the Administrator has decided to retain the current suite of secondary
PM standards generally, while revising only the form of the secondary
annual PM2.5 standard to remove the option for spatial
averaging consistent with this change to the primary annual
PM2.5 standard. The Administrator has also decided, contrary
to what was proposed, not to establish a distinct secondary standard to
address PM-related visibility impairment. This section discusses EPA's
responses to the comments EPA received on its proposal, and the
rationale behind the Administrator's final decisions is discussed in
section VI.D. below.
1. Comments on Proposed Secondary Standard for Visibility Protection
a. Overview of Comments
Among those commenting on the proposal to set a distinct secondary
PM2.5 visibility index standard, a large majority of
commenters, including more than 25 state and local agencies; regional
organizations such as NACAA, NESCAUM, and WESTAR; and industry
commenters, such as ACC, API, BP, EPRI, NCBA, NEDA-CAP, NMA, NSSGA, and
UARG, opposed setting a distinct secondary standard for visibility at
this time. Many commenters in this group expressed the view that such a
standard was not needed, primarily on the basis that adequate
protection was provided by the existing 24-hour secondary
PM2.5 standard. Some of these commenters also expressed
legal concerns with the nature of the proposed standard. Other
commenters in this group supported a distinct secondary standard for
visibility in concept, but expressed the view that it was premature to
set such a standard pending collection of additional visibility
preference study data and the resolution of a number of key technical
issues. Support for setting such a distinct secondary standard for
visibility at this time came from a second group of commenters,
including the Department of the Interior (National Park Service),
several states, the Mid-Atlantic/Northeast Visibility Union (MANE-VU),
the National Tribal Air Association (NTAA), environmental organizations
such as the Appalachian Mountain Club, National Parks Conservation
Association, Earthjustice (AMC, et al.) and the League of Women

[[Page 3205]]

Voters of Texas. These commenters argued that the existing secondary
standards are not sufficiently protective of visual air quality, and
that a distinct secondary standard similar to the proposed visibility
index standard is both necessary and appropriate to ensure adequate
protection of visibility.
Commenters in both groups expressed concerns about various aspects
of the proposed distinct secondary standard, including the indicator,
averaging time, level, and form. In addition, a large number of
commenters, including commenters from both groups, expressed concern
and/or confusion over the relationship between the Regional Haze
Program and the proposed distinct secondary standard for visibility,
raising issues such as analytical differences in methods between the
programs, monitoring issues, and other implementation challenges.
A discussion of the significant comments outlined above, including
EPA's responses to the comments, is presented here, with more detailed
discussion in the Response to Comments document. Comments relating to
the specific elements of the proposed standard--indicator, averaging
time, form and level--are discussed in sections VI.C.1.b-e,
respectively. Comments related to the need for a distinct secondary
standard at this time are discussed in section VI.C.f. Legal issues
raised by commenters opposed to setting a secondary standard based on
the proposed visibility index are discussed in section VI.C.g. Finally,
comments related to the relationship between a distinct secondary
standard and the Regional Haze Program are discussed in section
VI.C.h.\185\ While the EPA concludes in section VI.D below to retain
the current suite of secondary PM2.5 standards, the
appropriateness of the protection that would be provided by the
proposed PM2.5 visibility index standard, and the
relationship between this degree of protection and that provided by the
current secondary 24-hour secondary PM2.5 standard, are key
elements in the Administrator's decision, and are discussed below.
---------------------------------------------------------------------------

\185\ Comments pertaining to implementation issues, which the
Administrator may not consider in making decisions about setting
national ambient air quality standards, are discussed in the
Response to Comments document, as are comments regarding monitoring
issues related to the proposed distinct visibility index standard.
---------------------------------------------------------------------------

b. Indicator
Numerous commenters, both those supporting a distinct secondary
standard and those opposed to setting such a standard, expressed views
on the suitability of utilizing a PM2.5 calculated light
extinction indicator for the standard as proposed. While these groups
of commenters differed in terms of their views on the appropriateness
of using calculated PM2.5 light extinction as the basis for
the indicator rather than relying on direct measurements of
PM2.5 light extinction, commenters from both groups
expressed concern over specific elements of the proposed method of
calculating PM2.5 light extinction. In particular,
commenters expressed differing views on which IMPROVE algorithm should
be utilized; whether it is appropriate to exclude coarse particles from
the indicator; and whether the proposed protocols for incorporating
data on relative humidity and PM2.5 species are
appropriate.\186\
---------------------------------------------------------------------------

\186\ Some commenters expressed concern about the omission of
other contributors to visibility impairment from the visibility
index, as discussed in the Response to Comments document.
---------------------------------------------------------------------------

i. Comments on Calculated vs. Directly Measured Light Extinction
The majority of commenters in both groups noted the uncertainties
associated with relying on a calculated light extinction indicator and
stated a preference for utilizing direct light extinction measurements.
However, recognizing the limitations on applying direct measurements at
present, commenters supporting the proposal to set a distinct standard
argued that relying on ``calculated light extinction is a reasonable
first approach'' (DOI, p. 2). These commenters pointed to the advice of
CASAC, which had acknowledged that it was not possible for the EPA to
develop an FRM for direct measurement of light extinction within the
time frame of this review and had concluded that relying on a
calculated PM2.5 light extinction indicator represented a
reasonable approach that could be implemented sooner than a directly
measured indicator. These commenters generally supported the proposal
to adopt a calculated PM2.5 light extinction indicator, at
least as an interim approach.
Commenters opposed to setting a distinct standard generally argued
that it was inappropriate to rely on a calculated light extinction
indicator rather than direct measurements. Some of these commenters
argued that the proposed calculated light extinction indictor is ill
suited for a bright line standard because the method uses average
humidity and a reconstructed visibility measurement calculated from
PM2.5 speciation filter analysis, rather than measuring what
is actually observed by individuals. A number of commenters advocated
postponing setting a distinct standard until an approach based on
direct light extinction measurements can be adopted. Many of these
commenters stated that relying on direct light extinction measurements
would enable a standard to be based on a shorter averaging time, either
1-hour or sub-daily (4 to 6 hours), consistent with the more
instantaneous nature of perceptions of visual air quality and the
advice of CASAC in this review.
The EPA generally agrees with commenters that an indicator based on
directly measured light extinction would provide the most direct link
between PM in the ambient air and PM-related light extinction. However,
as noted at the time of proposal and in accordance with the advice of
CASAC, the EPA has concluded that this is not an appropriate option in
this review because a suitable specification of currently available
equipment or performance-based verification procedures could not be
developed in the time frame of this review. Moreover, CASAC concluded
that relying on a calculated PM2.5 light extinction
indicator based on PM2.5 chemical speciation and relative
humidity data represented a reasonable approach. The inputs that are
necessary include measurements that are available through existing
monitoring networks and approved protocols. Thus, the EPA remains
confident that the available evidence demonstrates that a strong
correspondence exists between calculated PM2.5 light
extinction and PM-related visibility impairment. Furthermore, CASAC
agreed, noting that the proposed calculated PM2.5 light
extinction indicator based on the original IMPROVE algorithm ``appears
to be a reasonable approach for estimating hourly light extinction''
(Samet, 2010d, p. 11) and ``its reliance on procedures that have
already been implemented in the CSN and routinely collected continuous
PM2.5 data suggest that it could be implemented much sooner
than a directly measured indicator'' (Samet, 2010d, p. iii). Thus it
would not be appropriate to postpone setting a distinct secondary
standard until an approach based on direct light extinction
measurements could be adopted.
ii. Comments on Specific Aspects of Calculated Light Extinction
Indicator
Some commenters, even those supporting the adoption of a calculated
light extinction indicator, also expressed concern over specific
aspects of the proposed indicator. First, a

[[Page 3206]]

number of commenters expressed concern over the proposal to use the
original IMPROVE algorithm as the basis for the calculated light
extinction indicator. These commenters noted that the original IMPROVE
algorithm has been shown to have consistent biases at both low and high
levels of light extinction. In particular, these commenters expressed
concern with the algorithm's bias at higher levels of light extinction,
which they pointed out were the conditions that might be encountered on
hazier days in urban areas.
Some commenters supported use of the revised IMPROVE algorithm.
These commenters noted that the revised equation has been through a
peer review which confirmed that it is based on the best science and
corrects the biases inherent in the original algorithm. Commenters also
noted that this revised algorithm has been widely incorporated into
Regional Haze plans, and urged the EPA to use this same equation in the
visibility index for the sake of consistency: ``EPA approved this
approach for regional haze and does not dispute its greater accuracy.
Therefore, a national secondary ambient air quality standard based on
criteria that accurately reflect the latest scientific knowledge
logically should not revert to the original IMPROVE algorithm''
(Oklahoma DEQ, p. 2). Other commenters noted that both the original and
the revised IMPROVE algorithms were designed in support of the Regional
Haze Program which is focused on largely rural Class I areas, and that
neither algorithm is necessarily suitable for urban areas. Noting that
the EPA has not thoroughly evaluated the applicability of either
IMPROVE algorithm in urban areas, these commenters urged additional
research to evaluate the suitability of either algorithm (or an
alternative approach) in urban areas.
Second, a number of commenters argued that exclusion of coarse PM
from the calculated light extinction indicator was inappropriate. These
commenters noted that coarse particulate matter is an important
contributor to visibility impairment in many areas, particularly in the
western U.S., and that the levels of ``acceptable'' visual air quality
derived from the visibility preference studies reflected total light
extinction due to the full mix of particles (including coarse PM) in
ambient air. A few commenters noted that due to the exclusion of coarse
particles, a ``deciview'' calculated for purposes of the proposed
PM2.5 visibility index is inconsistent with the unit as
conventionally defined under the Regional Haze Program. Other
commenters, however, supported the proposal to exclude coarse PM from
the calculated light extinction indicator, noting the important role
that PM2.5 plays in urban visibility and arguing it would be
more difficult to control the contribution of coarse particle sources
such as wind-blown dust to urban visibility impairment.
Third, some commenters questioned why the EPA was proposing to rely
on monthly average relative humidity (f(RH)) values when hourly
humidity data are widely available, particularly in urban areas. One
commenter argued that the EPA's proposed approach involves ``guessing
relative humidity'' rather than relying on accurate, readily available
measurements (Oklahoma DEQ, p. 1). The commenter stated that since
relative humidity is highly variable and weather dependent, the
proposed approach ``effectively undermines the capacity of the
prescribed monitoring regime to identify periods when PM2.5
adversely affects visibility.'' Other commenters supported this view,
noting that relative humidity can vary substantially even within a 24-
hour period, and that light extinction can be very sensitive to these
changes. These commenters recommended that hourly or daily humidity
measurements should be utilized in place of the proposed monthly
average f(RH) values.
Some commenters also recommended that the EPA should utilize a 90
percent relative humidity screen rather than 95 percent cap for
purposes of eliminating periods in which visibility impairment is due
to rain or fog. These commenters claimed that under a 95 percent cap,
both the average f(RH) values and the PM2.5 visibility index
values could be inflated in locations frequently affected by fog and/or
precipitation. These commenters preferred the approach of excluding
hours with relative humidity above 90 percent on the grounds that this
approach would eliminate foggy/rainy hours irrespective of the
frequency of occurrence.
The EPA does not agree with commenters who advocated using the
revised IMPROVE algorithm. Both the original and the revised IMPROVE
algorithms have been evaluated by comparing the calculated estimates of
light extinction with coincident optical measurements. As discussed
above in section VI.B.1.a.i, the revised algorithm was developed to
address observed biases in the predictions using the original algorithm
under very low and very high light extinction conditions, with further
modifications and additions to better account for differences in
particle composition and aging in remote areas.\187\ However, the EPA
does not believe that these same modifications and additions would
necessarily be appropriate for calculating light extinction in urban
areas. Instead, the EPA considers the original algorithm to be suitable
for purposes of calculating urban light-extinction, although some
adjustments may be appropriate for urban environments as well. The
reasons why the original algorithm is suited to urban environments are
discussed further below, along with adjustments that the EPA believes
are likely appropriate based on the current (limited) state of
knowledge.
---------------------------------------------------------------------------

\187\ Specifically, the revised IMPROVE algorithm incorporates
additional terms to account for particles representing the different
dry extinction and water uptake (f(RH)) from two size modes of
sulfate, nitrate and organic mass, as well as adding a term for
hygroscopic sea salt. There are also adjustments for the calculation
of OM as 1.8*OC compared to 1.4*OC in the original algorithm to
better account for the more aged PM organic components found in
remote areas.
---------------------------------------------------------------------------

First, the EPA considers that the multiplier of 1.8 used to convert
OC to OM in the revised IMPROVE algorithm is too high for urban
environments. The EPA is aware that there has been considerable debate
within the research community about the appropriate multiplier to use
to best represent urban environments. As discussed in Appendix F of the
Policy Assessment (U.S. EPA, 2011a), the EPA used the SANDWICH mass
closure approach (Frank, 2006) in the Urban Focused Visibility
Assessment (U.S. EPA, 2010b) for purposes of calculating maximum
daylight hourly PM2.5 light extinction and evaluated which
multiplier would produce 24-hour results most similar to the SANDWICH
approach using 24-hour PM2.5 organic carbon derived from the
new Chemical Speciation Network (CSN) carbon monitoring protocol
established in 2007.\188\ Analyses presented in Appendix F of the
Policy Assessment indicate that a multiplier of 1.6 is most appropriate
for purposes of comparing the hourly PM2.5 light extinction
with calculated 24-hour extinction (see Appendix F, section F.6 for a
full explanation). The EPA also considers this higher multiplier to be
a better approach for urban CSN monitoring sites where the new
measurements of organic carbon tend to be lower than those produced by
the older NIOSH-type monitoring protocol

[[Page 3207]]

(Malm, 2011). A multiplier of 1.6 is now used to calculate OM from OC
measurements at CSN sites.
---------------------------------------------------------------------------

\188\ Starting in 2007, the CSN adopted the IMPROVE monitoring
protocol for the measurement of organic and elemental carbon using
the IMPROVE analytical method and an IMPROVE-like sampler. The
transition was completed in 2009. (See ``Modification of Carbon
Procedures in the Speciation Network,'' http://www.epa.gov/ttn/amtic/files/ambient/pm25/spec/faqcarbon.pdf.)
---------------------------------------------------------------------------

At the time of proposal, the EPA proposed to use the original
IMPROVE algorithm with its 1.4 multiplier for converting OC to OM, but
requested comment on whether this value was appropriate. Comments
received by the Agency generally indicate that the OC-to-OM multiplier
of 1.4 used in the original IMPROVE algorithm is too low for urban
areas. Based on the analyses presented in Appendix F of the Policy
Assessment, the EPA agrees with these commenters. However, the EPA also
believes that it would be inappropriate to use a multiplier as high as
1.8 to convert OC to OM in urban areas. As noted by commenters, the
organic mass contribution to visibility impairment can be large, and
generally OM is significantly larger in urban areas compared to
surrounding rural areas.\189\ Because a large portion of the organic
component of urban PM results from nearby emissions sources, the total
OM mass is generally closer to the measured OC from which it is
derived. This means it is appropriate to use a smaller multiplier to
convert OC to OM in urban areas as compared to the value of 1.8 used in
the revised algorithm, which is tailored to remote areas. The CASAC
noted that urban OM includes fresh emissions and the EPA concluded in
the Visibility Assessment that ``the original version is considered
more representative of urban situations when emissions are still fresh
rather than aged as at remote IMPROVE sites'' (U.S. EPA, 2010b, p. 3-
19). Although the revised algorithm represents the best science of
estimating extinction in remote areas with its aged aerosol, the
commenters did not address how the EPA should modify the revised
algorithm to best represent the more complex and different urban
aerosol, particularly for OM. In light of all of these considerations,
in particular the analyses the EPA conducted for Appendix F of the
Policy Assessment and the fact that the monitoring method for organic
carbon has recently changed in the CSN network, the EPA judges that a
multiplier of 1.6 for urban areas would be most appropriate for
purposes of calculating PM2.5 light extinction in urban
areas.\190\ In formulating this judgment, the EPA recognizes that
neither the original nor the revised IMPROVE algorithm has been tested
for suitability in urban areas and that additional research is
necessary to reduce the uncertainties about the most appropriate value
for the OC to OM multiplier in urban environments. With regard to other
changes between the original and revised IMPROVE algorithms, the EPA
also does not believe that it would be appropriate to include a term
for hygroscopic sea salt for urban light extinction, or to
differentiate between different size modes of sulfate, nitrate, and
organic mass as empirically defined by the revised IMPROVE algorithm.
Unlike in some remote coastal locations, sea salt is not major
contributor to light extinction in urban areas. Moreover, urban sources
of salt include sanding of roads during the winter and those re-
entrained particles are mostly in the coarse size range.
---------------------------------------------------------------------------

\189\ The difference between higher PM2.5 mass in
urban areas compared to surrounding regions, known as the urban
excess, is largely attributed to organic mass (U.S. EPA, 2004b).
\190\ The implications of this shift to a 1.6 multiplier for OC
in urban areas for decisions about averaging time, level, and need
for a distinct secondary standard are discussed further below in
sections VI.C.1.c, VI.C.1.e, and VI.C.1.f, respectively.
---------------------------------------------------------------------------

Like in remote areas, small and large size modes of sulfate,
nitrate and organic mass would exist in the urban environment. However,
the apportionment of the total fine particle concentration of each of
the three PM2.5 components into the concentrations of the
small and large size fractions would likely need a different approach
than that used for remote areas. This is because of the closer
proximity of urban sources to their emissions. This is a particular
concern not only for organic mass, which as explained previously has a
large contribution from nearby urban emission sources, but also for
PM2.5 nitrate whose concentrations are also higher in urban
areas compared to the surrounding regions. Thus, a higher portion of
the total urban concentration may be in the small mode compared to
remote areas and thus a different apportionment algorithm would be
needed.
Finally, the EPA does not consider it necessary to employ site-
specific Rayleigh light scattering terms in place of a universal
Rayleigh light scattering value for purposes of calculating light
extinction in urban areas for purposes of calculating the 90th
percentile values. The site-specific Rayleigh value is most important
to accurately estimate extinction on the best visibility days which is
an essential metric for the regional haze program.
For all of these reasons, the EPA considers the original IMPROVE
algorithm better suited to the task of calculating urban light
extinction than the revised IMPROVE algorithm. However, the EPA does
consider it appropriate to make certain adjustments to the original
algorithm for purposes of calculating urban light extinction. As
discussed above, the EPA believes it is appropriate to use a 1.6
multiplier to convert OC to OM in urban areas. In addition, the EPA
believes it is appropriate to exclude the term for coarse particles
from the equation. The EPA does not agree with commenters who suggested
that coarse particles should be included in the calculated light
extinction indicator. As noted in the proposal, PM2.5 is the
component of PM responsible for most of the visibility impairment in
most urban areas. Currently available data suggest that
PM10-2.5 is a minor contributor to visibility impairment
most of the time, although at some locations (U.S. EPA, 2010b, Figure
3-13 for Phoenix) PM10-2.5 can be a major contributor to
urban visibility effects. While it is reasonable to assume that other
urban areas in the desert southwestern region of the country may have
conditions similar to the conditions shown for Phoenix, in fact few
urban areas conduct continuous PM10-2.5 monitoring. This
significantly increases the difficulty of assessing the role of coarse
particles in urban visibility impairment. For example, among the 15
urban areas assessed in this review, only four areas had collocated
continuous PM10 data allowing calculation of hourly
PM10-2.5 data for 2005 to 2007. In addition,
PM10-2.5 is generally less homogenous in urban areas than
PM2.5 in that coarse particle concentrations exhibit greater
temporal variability and a steeper gradient across urban areas than
fine particles (U.S. EPA, 2009a, p. 3-72). This makes it more
challenging to select sites that would adequately represent urban
visibility conditions. Thus, while it would be possible to include a
PM10-2.5 light extinction term in a calculated light
extinction indicator, as was done in the Visibility Assessment, there
is insufficient information available at this time to assess the impact
and effectiveness of such a refinement in providing public welfare
protection in areas across the country (U.S. EPA, 2011a, pp. 4-41 to 4-
42). Therefore, the EPA concludes that it is not appropriate to set a
standard based on a calculated light extinction indicator that includes
coarse particles at this time, and the calculated indicator should be
based on PM2.5 light extinction.
With regard to the suggestion by some commenters that the
calculated light extinction indicator should be calculated using hourly
humidity data,

[[Page 3208]]

the EPA disagrees that concurrent humidity measurements should be used.
The use of longer-term averages for each monitoring site adequately
captures the seasonal variability of relative humidity and its effects
of visibility impairment, and this approach focuses more on the
underlying aerosol contributions to visibility impairment and less on
the day-to-day variations in humidity. This provides a more stable
indicator for comparison to the NAAQS and one that is more directly
related to the underlying emissions that contribute to visibility
impairment.
With regard to the comments advocating the use of a 90 percent
humidity screen as opposed to a 95 percent humidity cap, the EPA
believes that relying on monthly average relative humidity values based
on 10 years of climatological data appropriately reduces the effect of
fog and precipitation. Although the approach of using a 95 percent
humidity cap, as in the Regional Haze Program, includes some hours with
relative humidity between 90-95 percent, the general approach of using
a longer-term average for each monitoring site effectively eliminates
the effect of very high humidity conditions on visibility at those
locations.
Therefore, taking all of the above considerations and CASAC advice
into account, the EPA continues to conclude that a calculated
PM2.5 light extinction indicator, similar to that used in
the Regional Haze Program (i.e., using an IMPROVE algorithm as
translated into the deciview scale), would be the most appropriate
indicator to replace the current PM2.5 mass indicator for a
distinct secondary standard. Moreover, the EPA continues to conclude
that this calculated indicator should based on the original IMPROVE
algorithm, adjusted to use a 1.6 OC multiplier and exclude the term for
coarse particles, in conjunction with monthly average relative humidity
data (i.e., f(RH) values) based on long-term climatological means as
used in the Regional Haze Program. A PM2.5 visibility index
defined in this way would appropriately reflect the relationship
between ambient PM and PM-related light extinction, based on the
analyses discussed in the proposal and reflecting the aerosol and
relative humidity contributions to visibility impairment by
incorporation of factors based on measured PM2.5 speciation
concentrations and climatological average relative humidity data. In
addition, this type of indicator would address, in part, the issues
raised in the court's remand of the 2006 PM2.5 standards.
Such a PM2.5 visibility index would afford a relatively high
degree of uniformity of visual air quality protection in areas across
the country by virtue of directly incorporating the effects of
differences in PM2.5 composition and relative humidity
across the country.
c. Averaging Time
Few commenters specifically addressed the issue of averaging time.
Those who did generally expressed the view that an hourly or sub-daily
averaging time would be the most appropriate approach, as supported by
CASAC and the EPA's own analyses in this review. These comments were
generally consistent with the emphasis among all commenters on the
desirability of adopting a directly measured light extinction indicator
that could be measured on an hourly or sub-daily time scale. Some
commenters noted that a standard based on a 4-6 hour averaging time
would better capture peak daily light extinction while allowing stable
signal quality; others urged EPA to adopt a 1-hour averaging time in
conjunction with direct measurements. Commenters pointed to significant
limitations associated with using a 24-hour averaging time, including
the uncertainties in translating hourly or sub-daily visibility index
values into 24-hour equivalent values. Some commenters criticized the
analysis presented in the Policy Assessment comparing the 24-hour
calculated light extinction values to the maximum daylight 4-hour
calculated light extinction values. These commenters stated that the
scatter plots and regressions presented in the Policy Assessment
indicate there is considerable variation in the 24-hour vs. 4-hour
relationship, and interpreted this to mean that 24-hour light
extinction values are a poor surrogate for 4-hour values. For example,
several industry commenters cited an analysis which noted that the
correlation coefficient between the 24-hour and 4-hour values was as
low as r\2\ = 0.42 in Houston, and stated that the EPA was being overly
``optimistic'' in concluding that city-specific and pooled r\2\ values
in the range of 0.6 to 0.8 showed good correlation (UARG, Attachment 2,
p. 27).
In addition, some commenters expressed concern over potential bias
and greater uncertainty introduced by the inclusion of nighttime hours,
noting that because relative humidity tends to be higher at night,
inclusion of these hours could cause areas to ``record NAAQS
exceedances that have no corresponding visibility impairment value''
(UARG, p. 36). Commenters also emphasized the poor fit of a 24-hour
averaging time with the near instantaneous judgments about visibility
impairment reflected in the visibility preference studies. Commenters
also noted that there is greater hourly variation in PM concentrations
and resulting visibility conditions in urban areas than in Class I
areas; thus, while the Regional Haze Program uses 24-hour IMPROVE data,
the commenters stated that a shorter averaging time is needed for an
urban-focused PM2.5 visibility standard. Some commenters
objected to a 24-hour averaging time as unsupported by the record in
this review: ``Because the science the Administrator relies on for the
other elements of the proposed visibility standard is tied to short-
term exposures to visibility impairment, the EPA has no basis for
promulgating a standard that uses a 24-hour averaging time'' (API, p.
43). These commenters claimed that while the EPA may not have the
information or infrastructure in place to allow the Agency to set a
standard based on a 1-hour or other sub-daily averaging time, this does
not justify moving to a 24-hour averaging time.
Among commenters supporting the proposed distinct secondary
standard for visibility, many commenters recognized the limitations on
monitoring methods and currently available data that led to the EPA's
proposal to adopt a standard based on a 24-hour averaging time. Most of
these commenters acknowledged that the lack of reliable hourly
speciation data means that a 24-hour averaging time is the only
workable approach for a standard based on calculated light extinction.
Commenters advocating a distinct secondary standard for visibility
therefore generally supported the proposal to adopt a 24-hour averaging
time, at least as an interim approach until a directly measured light
extinction indicator could be adopted in the future. This approach was
also supported by a few industry commenters who noted that since a
visibility index standard would be based on data from the IMPROVE and
CSN monitors, which operate on a 24-hour basis with 1-in-3 (or 1-in-6)
day sampling, ``it is imperative that EPA retain a 24-hour averaging
time if a secondary visibility standard is promulgated'' (API,
Attachment 2, p. 9).
In response to comments supporting a 1-hour or sub-daily (4- to 6-
hour) averaging time in conjunction with a direct light extinction
measurements, the EPA notes that, as discussed above in the response to
comments on indicator, the Agency has concluded

[[Page 3209]]

that a directly measured light extinction indicator is not an
appropriate option in this review, independent of the decision on
averaging time. Having reached the conclusion that a calculated
PM2.5 light extinction indicator would be most appropriate,
the EPA has next considered what averaging time would be most desirable
for such an indicator. As noted in the proposal, the EPA has recognized
that hourly or sub-daily (4- to 6-hour) averaging times, within
daylight hours and excluding hours with high relative humidity, are
more directly related than a 24-hour averaging time to the short-term
nature of the perception of PM-related visibility impairment and the
relevant exposure periods for segments of the viewing public. Thus, the
Agency agrees with commenters' general point that, as a starting
premise, a sub-daily averaging time would generally be preferable.
However, as noted at the time of proposal and discussed above in
section VI.B.1.c, important data quality uncertainties have recently
been identified in association with currently available instruments
that would be used to provide the hourly PM2.5 mass
measurements that would be needed in conjunction with an averaging time
shorter than 24 hours. As a result, at this time the Agency has strong
technical reservations about a secondary standard that would be defined
in terms of a sub-daily averaging time. The data quality issues which
have been identified, including short-term variability in hourly data
from currently available continuous monitoring methods, effectively
preclude adoption of a 1-hour averaging time in this review, given the
sensitivity of a 1-hour averaging time to these data quality
limitations. Even with regard to multi-hour averaging times, the EPA
continues to conclude that the data quality concerns preclude adoption
of a sub-daily averaging time.
Moreover, analyses conducted for the Policy Assessment indicate
that PM2.5 light extinction calculated on a 24-hour average
basis would be a reasonable and appropriate surrogate for
PM2.5 light extinction calculated on a 4-hour basis. The
scatter plots comparing 24-hour and 4-hour calculated PM2.5
light extinction in the Policy Assessment (U.S. EPA, 2011a, Figures G-4
and G-5) do show some scatter around the regression line for each city.
This was to be expected, since the calculated 4-hour light extinction
includes day-specific and hour-specific influences that are not
captured by the simpler 24-hour approach. Overall, however, in the
EPA's view, both the city-specific and pooled 15-city 24-hour vs. 4-
hour comparisons show strong correlation between the two averaging
times. Moreover, the 90th percentile design values calculated for 4-
hour vs. 24-hour light extinction are much more closely correlated than
are the values for individual days in particular urban areas calculated
using these two approaches. Thus, while the EPA agrees with commenters
who pointed out the relatively low correlation between 4- and 24-hour
values in cities such as Houston, the Agency points out that the
correlations of 90th percentile values are much higher, particularly
when one considers the average values across urban areas. In general,
the 90th percentile values line up better and demonstrate closer to a
one-to-one relationship.
The EPA has conducted a reanalysis (Frank et al., 2012b) of the
relationships between estimated 24-hour and 4-hour visibility
impairment based on the variety of metrics discussed in Appendix G of
the Policy Assessment that further supports this finding. The
reanalysis more appropriately considered the uncertainty of the
calculated 4-hour values. It also considered the effect of changing the
OC to OM multiplier used in urban areas with the new CSN monitoring
protocol from 1.4 to 1.6. The revised analysis shows that the 24-hour
values are generally closer to the 4-hour values than originally
estimated.
Since conclusions in the proposal about the relationship between 4-
hour and 24-hour values were drawn not just on the basis of the city-
specific results but also on the more robust 90th percentile values,
the EPA disagrees with commenters who state that the Agency was overly
optimistic in considering 24-hour values an appropriate surrogate for
4-hour values. Also, it is appropriate to focus on the 90th percentile
design value comparison since the design values would determine
attainment status and the degree of improvement in air quality that
could be expected in areas instituting controls to meet the NAAQS.
Therefore the EPA does not agree with commenters who state that a 24-
hour averaging time cannot serve as an appropriate surrogate for sub-
daily periods of visibility impairment. On the contrary, the EPA
continues to conclude, on the basis of this analysis, that
PM2.5 light extinction calculated on a 24-hour basis is a
reasonable and appropriate surrogate for sub-daily PM2.5
light extinction calculated on a 4-hour basis.
The EPA recognizes that the effect of adopting a 24-hour averaging
time may be to smooth out some of the hour-by-hour variability in
visibility index values. (Indeed, this is true if we compare a 4-hour
averaging time to a 1-hour averaging time as well.) Hour-specific
influences which would be evident if an hourly or sub-daily averaging
time were to be used will be masked to some extent when those hours are
averaged together with other hours. This means, in part, that a 24-hour
averaging time may effectively reduce peak values by means of averaging
them together with other hours, which may have lower values. However,
given the well documented variability in hourly visibility conditions,
especially in urban areas, as noted by commenters, it is reasonable to
assume that in some cases peak hours may be significantly influenced by
atypical conditions, making it appropriate to adopt an averaging time
that is sufficiently long to ensure that hour-specific influences are
balanced against more typical conditions. Perhaps even more important
is the concern that many peak hourly measurements may be significantly
influenced by atypical instrument performance; this reinforces the
conclusion that it is appropriate to adopt a longer averaging time, to
ensure that hour-specific uncertainties are balanced against more
robust measurements.
Thus, in agreement with commenters who supported a daily averaging
time, the EPA concludes that a 24-hour averaging time would be
appropriate for a distinct secondary standard based on a calculated
PM2.5 light extinction indicator.
d. Form
The EPA received very few comments with regard to the proposal to
adopt a 90th percentile form, averaged over 3-years, in conjunction
with a PM2.5 visibility index and a 24-hour averaging time.
One commenter stated that it was inappropriate to use a 90th percentile
form, noting that this would result in the exclusion of a minimum of 36
days of data annually. The commenter expressed particular concern that
this proposed approach, in combination with a 24-hour standard based on
an unadjusted CPL, would not capture the worst visibility impairment
and that this would undermine ``the intent of setting a meaningful
secondary visibility standard'' (AMC, et al., p. 2). Another commenter
argued that the EPA had provided no scientific basis for why the 90th
percentile form was suitable, and claimed that the Agency was making
``a somewhat arbitrary judgment that people's welfare would be affected
only if adverse urban visibility were to occur

[[Page 3210]]

more than 10 percent of the time'' (API, Attachment 2, p. 4).
On other hand, a few commenters who appeared to generally support
the proposal to use a 90th percentile form advocated averaging the 90th
percentile values over longer time periods, arguing that averaging over
only 3 years would not provide a stable assessment of visual air
quality in the West because this time period is insufficient to
properly account for western drought and fire cycles. These commenters
pointed to the approach in the Regional Haze Program of averaging
visibility impairment over 5 years, and noted that even within this
longer time period data can be significantly influenced by high
emissions during significant fire years.
The EPA disagrees with all of these comments. With regard to the
comment opposing the 90th percentile form as inappropriately excluding
the worst visibility days, the EPA notes that there is a significant
lack of information on, and a high degree of uncertainty regarding, the
impact on public welfare of the number of days with visibility
impairment over the course of a year. For example, the visibility
preference studies used to derive the range of CPLs considered in this
review offered no information regarding the frequency of time that
visibility levels should be below those values. Based on this
limitation, the EPA concluded in the Policy Assessment that it would
not be appropriate to consider eliminating all exposures above the
level of the standard and that it was reasonable to consider allowing
some number of days with reduced visibility. Recognizing that the
Regional Haze Program focuses attention on the 20 percent worst
visibility days (i.e., those at or above the 80th percentile of
visibility impairment), the EPA continues to believe, as noted in the
proposal, that a percentile well above the 80th percentile would be
appropriate to increase the likelihood that all days in this range
would be improved by control strategies intended to help areas attain
the standard. Focusing on the 90th percentile, which represents the
median of the distribution of the 20 percent worst visibility days,
could be reasonably expected to lead to improvements in visual air
quality on the 20 percent most impaired days. Thus, the EPA has made a
reasoned judgment based on a full consideration of the upper end of the
distribution of visibility impairment conditions and continues to
conclude that it is appropriate to focus on the 90th percentile of
visibility impairment values.
With regard to comments requesting the EPA adopt a longer multi-
year averaging period for the 90th percentile values, the EPA disagrees
that it would be appropriate to average the 90th percentile values over
periods longer than 3 years. The EPA recognizes that a multi-year
percentile form offers greater stability to the air quality management
process by reducing the possibility that statistically unusual
indicator values will lead to transient violations of the standard.
Utilizing a 3-year average form provides stability from the occasional
effects of inter-annual meteorological variability that can result in
unusually high pollution levels for a particular year. The Agency has
adopted this approach in other NAAQS, including the current secondary
24-hour PM2.5 NAAQS, which has a 98th percentile form
averaged over 3 years. However, adopting a multi-year averaging period
longer than 3 years would increase the number of days with visibility
impairment above the target level of protection and would therefore
reduce the protectiveness of the standard. Based on this the EPA does
not believe it would be appropriate to average 90th percentile values
over a period as long as five years. Therefore, the EPA continues to
conclude that a 90th percentile form, averaged over 3 years, would be
appropriate, in conjunction with a calculated PM2.5 light
extinction indicator and a 24-hour averaging time.
e. Level
With regard to level, commenters focused on two main themes. First,
a large number of commenters addressed the information available from
the public preference studies with regard to the acceptability of
various levels of visual air quality. These comments, which are
discussed in subsection VI.C.1.e.i below, address the EPA's use of
visibility preference studies as the basis for the selection of a range
of appropriate levels for the Administrator to consider. Many
commenters challenged the use of these studies as the basis for setting
a distinct secondary standard, arguing that limitations in these
studies rendered them an unsuitable and insufficient basis on which to
establish such a standard. Second, commenters expressed different views
as to what level(s) of a distinct secondary standard would be
appropriate, if the EPA were to set such a standard. These comments
reflected consideration of the results of the public preference studies
as well as analyses conducted in the Visibility Assessment and the
Policy Assessment, as discussed in the proposal. Comments addressing
the appropriateness of specific levels are discussed in subsection
VI.C.1.e.ii below.
i. Comments on Visibility Preference Studies
A majority of commenters expressed the view that the existing
preference studies provide an insufficient basis for selection by the
Administrator of an appropriate level of public welfare visibility
protection for a national standard. These commenters highlighted a
number of limitations and uncertainties (enumerated below) associated
with these studies as support for this view. In contrast, other
commenters felt that despite certain limitations, these studies do
provide a sufficient basis on which the Administrator can select an
appropriate level of a standard to provide national public welfare
visibility protection. The remainder of this section organizes and
discusses these comments under four broad topic areas, including: (a)
Limitations and uncertainties associated with the visibility preference
studies; (b) preference study methods and design; (c) use of preference
study results for determining adversity; (d) the appropriateness of
using regionally varying preference study results to select a single
level for a national standard.
(a) Preference Study Limitations and Uncertainties
A large and diverse number of limitations and uncertainties
associated with the visibility preference studies have been identified
and discussed in the public comments. Many of these same limitations
and uncertainties were also identified and discussed by the EPA in the
various documents developed throughout this review. The most important
and fundamental limitations and uncertainties will be discussed here in
the preamble, while more specific, unique or detailed comments will be
addressed in the Response to Comments document.
The primary or most frequent limitation cited by many commenters
relates to the small number of preference studies that are available in
this review. In particular, some commenters note that these preference
studies cover just four locations, only three of which occur in the
U.S., that the two studies conducted in Washington, DC were pilot
studies, not full preference studies, and/or that three of the
preference studies were conducted in the West, while only one was
conducted in the East, providing only limited geographic coverage.
Typically, these same commenters also pointed out that taken together,
these

[[Page 3211]]

limited studies only included a total of 852 participants, which they
claimed was too small a sample size and unrepresentative nationally.
These commenters thus concluded that there is insufficient information,
both geographically and demographically, upon which to select a
national level of a visibility index for purposes of visibility
protection.
In contrast, several commenters stated support for using the
preference studies, concluding they provide an adequate basis, in spite
of their limited nature. In particular, AMC et al. state:

We believe that these studies provide sufficient results to
inform setting a national visibility standard. While the number of
studies is small, they do incorporate spatial variation and, in the
case of Denver and Phoenix, varied populations* * *. EPA should have
confidence, rather than uncertainty, in the fact that these studies
used different methods and respondents and yield a range of 20-24
dv, with one outlier of 29. (AMC, et al., pp. 6-7)

Regarding the first group of commenters, the EPA notes that it is
well aware of the limited nature of the information, which it has
described in great detail in the Integrated Science Assessment,
Visibility Assessment, and Policy Assessment, as well as in section
VI.B.2 of the proposed rule (77 FR 38973). The EPA further notes,
however, that limited information does not preclude the Administrator
from making judgments based on the best available science, taking into
account the existing uncertainties and limitations associated with that
available science. Thus, in reaching judgments based on the science,
the Administrator appropriately weighs the associated uncertainties.
The CASAC supported this view and concluded that the available
information provided a sufficient basis on which the Administrator
could form a judgment about requisite PM-related public welfare
visibility protection. Specifically, CASAC stated ``[t]he 20-30
deciview range of levels chosen by EPA staff as `Candidate Protection
Levels' is adequately supported by the evidence presented'' (Samet,
2010b, p. iii). As discussed in the proposed rule (77 FR 38990), the
Administrator recognized and explicitly took into account the
uncertainties and limitations in the science in determining an
appropriate degree of protection when she proposed a level at the upper
end of the recommended range. As discussed below, the Administrator
continues to be mindful of these uncertainties and limitations in
reaching her final determination regarding what constitutes an
appropriate degree of protection with respect to PM-related visibility
impairment.
With respect to the comments of AMC et al., the EPA agrees that
these studies provide a sufficient basis to inform the Administrator's
judgments regarding an appropriate level of protection from PM-related
visibility impairment, but she recognizes that these studies, which are
the only studies before her, are a limited source of information.
However, the EPA does not agree that the Washington, DC, results
represent an outlier, and thus the EPA believes these results are
appropriately included in the range identified for the Administrator to
consider.
Some commenters made the point that the EPA relied on much of this
same evidence to reach the conclusion in 2006 that the information was
too limited to allow selection of a national standard. For example, API
stated:

[T]he bulk of the VAQ preference studies were available during
the previous PM NAAQS review and were considered by the Agency in
its establishment of the 2006 p.m. secondary NAAQS * * *. The
Proposed Rule does not mention this fact and does not explain why
many of these same studies now compel EPA to propose this new
secondary NAAQS * * *. The Proposed Rule notes in passing that,
since the last review of the PM NAAQS, `limited information that has
become available regarding the characterization of public
preferences in urban areas has provided some new perspectives on the
usefulness of this information in informing the selection of target
levels of urban visibility protection.' 77 Fed. Reg. at 38969/2. It
is a serious oversight that the Proposed Rule makes no attempt to
explain what that information is or how it affects the
interpretation of the VAQ preference studies. This `limited
information' is an apparent reference to information provided by Dr.
Anne Smith. (API, p. 37)

The EPA disagrees with these commenters. First, the EPA disagrees
that it failed to distinguish between studies that were available in
the previous review and the current review. The discussion in section
VI.A.1 of the proposal specifically identifies the studies from Denver,
Phoenix and British Columbia (77 FR 38967/2) as being considered in the
last review. The EPA further disagrees with the implication that it is
being circumspect about identifying the ``limited information that has
become available regarding the characterization of public preferences
in urban areas.'' Beginning in section VI.A.3 of the proposed rule (77
FR 38969), the EPA was clear about what information, both preexisting
and new, it relied upon in this review to inform its views and provide
the basis for its proposal. In section VI.B.2, the EPA elaborates on
the specific information, tools, methods and data which are considered
in relation to the public preference studies, including the new
information available since the last review.
As noted above and in the proposal, in addition to the substantial
PM urban air quality information and analyses new to this review, there
are three other sources of information that have specifically
``provided some new perspectives on the usefulness of'' the preference
studies ``in informing the selection of target levels of urban
visibility protection'' (77 FR 38969). They include: (1) Results from
additional urban visibility preference study experiments conducted for
Washington, DC by Smith and Howell (2009) which added to the preference
data for that location and shed light on the role of location in
preference responses; (2) a review and reanalysis (Stratus Consulting,
2009) of the urban visibility public preference studies from the four
urban areas, including the newly available Smith and Howell (2009)
experiments which examined the similarities and differences between the
studies and evaluated the potential significance of those differences
on the study results; and (3) additional analyses, including most
importantly a logit analysis (Deck and Lawson, 2010, as discussed in
Chapter 2 and Appendix J of the Visibility Assessment), which was
requested and reviewed by CASAC, which showed that each city's
responses represented unique and statistically different curves. Taken
together, these sources contributed to the EPA's current knowledge and
understanding of each survey study's results, the appropriateness of
comparing each study's results to the others, and the key uncertainties
relevant to data interpretation. In addition, in the last review the
decision to not adopt a distinct secondary standard was remanded as
contrary to law and failing to provide a reasoned explanation for the
decision. As such it is not appropriate for purposes of comparison with
the Administrator's judgment and reasoning in this review.
(b) Preference Study Methods and Design
In addition to the limitations and uncertainties noted above, many
comments also asserted the methodologies used in the preference studies
are fundamentally flawed. Many commenters cited some of the same issues
that have already been identified by the EPA as sources of uncertainty
and potential factors in producing the statistically different study
results (see section VI.B.1.b above). As noted above,

[[Page 3212]]

the EPA is well aware of the issues raised regarding the adequacy of
the preference studies to serve as a basis for a secondary NAAQS (see
77 FR 38975) and solicited comment on how these uncertainties should be
considered (see 77 FR 38990). Most of these same commenters also
pointed to an assessment of the preference studies methodology provided
by Smith and Howell (2009) as the basis for their views, as indicated
by the following comments:

Smith and Howell (2009) show that VAQ preference study outcomes
are malleable and depend entirely on the design of the study.
Accordingly, such studies do not identify any meaningful threshold
of acceptable visibility conditions. Despite Smith and Howell's
conclusions, EPA continues to assert that the VAQ preference studies
can be used to identify minimally acceptable visibility conditions
even though the Agency has never provided any valid scientific basis
for discounting the Smith and Howell (2009) results. (API, p. 38)
Well-controlled preference studies discussed by Anne Smith of
Charles River Associates at the March 2010 CASAC meeting
demonstrated that the judgment of panel members was affected by the
order in which photographs were presented and tendency to identify
the middle of the range of visibility degredation as a threshold of
acceptability. This points to a potential flaw in these studies and
that artifacts caused by these tendencies may have influenced study
results. Dismissing these inherent flaws in the existing preference
studies and then using these studies to set a secondary NAAQS is
arbitrary and capricious. (API, Attachment 2, p. 12)
EPA also fails to acknowledge that the only study conducted
since the last review rebuts the validity of the VAQ preference
studies previously conducted. (UARG, Attachment 2, p. 28)

As is explained in a more detailed discussion in the Response to
Comments document, the EPA disagrees that the study conducted by Smith
and Howell (2009) supports the conclusion that the preference study
methodologies were fundamentally flawed; however, the EPA notes that
their experiments do identify areas where additional research would be
useful to further inform our limited understanding of public
preferences in urban areas. The EPA views the Smith and Howell
experiments as increasing the EPA's knowledge and understanding of the
findings of the 2001 Washington, DC focus group pilot study (Abt, 2001)
in several important ways, although this information still remains
limited overall. Specifically, the Smith and Howell results suggest:
(1) The 2001 results, while based on a small sample size of 9, were
consistent with results from a larger sample of the general Washington,
DC population; (2) an individual's preferences for visibility in one
location may not depend on whether they live in that location; and (3)
presentation method (i.e., changing from slide projection to computer
monitor) did not appear to affect the reported preferences.
(c) Preference Study Results and Adversity
A number of comments were received regarding the EPA's use of
preference study results to make the determination that adverse
PM2.5-related visibility effects on the public welfare are
occurring. In this context, several commenters questioned whether the
EPA had made the case that unacceptable levels of visual air quality
based on preference study results alone can be equated with an adverse
public welfare effect. These commenters suggested that unless
preference study information is linked to personal comfort and well-
being or other associated welfare effects, it cannot form the basis of
a determination of adversity. For example, Kennecott Utah Copper LLC
stated that:

Thus, EPA seemingly was building the foundation for a
determination of what constitutes an adverse effect on visibility in
the context of public welfare. However * * * EPA subsequently veered
toward an oversimplified focus on public acceptance of visibility
conditions * * *. EPA's discussion of visibility in the Policy
Assessment and its proposed rule in the Federal Register focuses
entirely on ``acceptable'' and ``unacceptable'' visual air quality
and make no mention of an ``adverse effect'' in the context of
visibility. EPA's reliance on only 3 urban preference studies
represents a paucity of data and a wholesale abandonment of any
effort to seek a scientifically measurable adverse effect.
(Kennecott Utah Copper LLC, p. 26)

In response, the EPA first notes that the definition of effects on
welfare included in section 302(h) of the CAA identifies both
visibility and the broader category of effects on personal comfort and
well-being as effects on welfare. In setting a secondary standard to
address visibility impairment, the EPA considers the effect on the
public from impairment of visibility as a separate and distinct welfare
effect in its own right. The EPA is not required to translate this into
terms of personal comfort and well-being, as visibility impairment is
designated explicitly by Congress as an effect on welfare. While there
may be a large degree of overlap among these different welfare effects,
the EPA properly focuses on evaluating all of the information before
the Agency on the effect visibility impairment has on the public,
whether or not this impairment would also be categorized as having an
adverse effect on personal comfort and well-being. It is in the context
of all of this information that the EPA makes the judgment as to the
appropriate degree of protection from known and anticipated adverse
effects on the public from visibility impairment. The EPA recognizes
that there is uncertainty about the degree of adversity to the public
welfare associated with PM-related visibility impairment. However a
secondary standard is designed to provide protection from ``known or
anticipated'' adverse effects, and a bright line determination of
adversity is not required in judging the requisite degree of protection
under section 109(b)(2). Furthermore, the EPA disagrees that it has
abandoned its consideration of visibility-related impacts on the
welfare effect of personal comfort and well-being, as is made clear in
the following quote:

Research has demonstrated that people are emotionally affected
by low visual air quality, that perception of pollution is
correlated with stress, annoyance, and symptoms of depression, and
that visual air quality is deeply intertwined with a ``sense of
place,'' affecting people's sense of the desirability of a
neighborhood (U.S. EPA, 2009a, section 9.2.4). Though it is not
known to what extent these emotional effects are linked to different
periods of exposure to poor visual air quality, providing additional
protection against short-term exposures to levels of visual air
quality considered unacceptable by subjects in the context of the
preference studies would be expected to provide some degree of
protection against the risk of loss in the public's ``sense of well-
being.'' (77 FR 38973/1, emphasis added)

The approach taken to address such qualitative, but policy-
relevant, information in this review is the same as in other NAAQS
reviews. The review is initiated with a comprehensive assessment of all
possible public health and welfare effects associated with PM in the
Integrated Science Assessment. Then policy-relevant effects for which
there is sufficient quantitative information to allow a determination
of the change in risks associated with incremental changes in air
quality are assessed (in this review, in the Visibility Assessment) and
used to provide a quantitative basis to inform the selection of an
appropriate range of levels for further consideration in the Policy
Assessment. In the Policy Assessment, the EPA considers all important
policy-relevant evidence and information, both quantitative and
qualitative, in making recommendations regarding the range of policy
options appropriate for the Administrator to consider. It is in the
context of all of this information that the Administrator

[[Page 3213]]

makes her final judgment as to the appropriate degree of protection
from known and anticipated adverse effects on the public from
visibility impairment.
Another issue raised in the comments regarding adversity is the
EPA's decision to use the 50 percent acceptability criterion from the
public preference studies in determining candidate protection levels of
visibility impairment for the selection of a national level of
visibility protection. For example, AMC et al. recommended ``a 75%
acceptability criterion as a target that is in line with protecting the
broader public from the negative effects of visibility impairment''
(AMC, et al., p. 9).
In the Visibility Assessment, the EPA noted that the use of the 50
percent acceptance level for urban visibility was first presented in
Ely et al. (1991) (U.S. EPA, 2010b, p. 2-5). Ely discussed the use of
the 50 percent acceptability criterion as a reasonable basis for
setting an urban visibility standard.

The standard was determined based on a 50% acceptability
criterion, that is, the standard was set at the level of extinction
that would divide the slides into two groups: those judged
acceptable and those judged unacceptable by a majority of the people
in the study. The criterion is politically reasonable because it
defines the point where a majority of the study participants begin
to judge slides as representing unacceptable visibility. It is also
consistent with psychological scaling theory which indicates that a
``true score'' exceeds a standard when more than 50% of the
``observed scores'' exceed that standard. (Ely et al., 1991, p. 11)

As Ely described, the 50 percent acceptability criterion and the
preference study conducted by Ely were used as the basis for setting
the level of the Denver Visibility Standard in 1990. That same
criterion was judged appropriate and selected for use in the Phoenix
preference study (BBC research, 2003) and as the basis for setting the
level of the Phoenix Visibility Standard in 2003. Most recently, the 50
percent acceptability criterion has been recommended by the British
Columbia Visibility Coordinating Committee as the basis for the
visibility standard currently under consideration by British Columbia,
Canada. Furthermore, CASAC supported this approach, while recognizing
the uncertainty associated with this issue. Specifically, CASAC agreed
that ``the 50th percentile for the acceptability criteria is logical,
given the noted similarities in methodologies employed in the 4 study
areas. * * * In terms of choosing a specific percentile from the
preference studies, we note that there may not be a ``preferred'' one,
but in assessing preference studies to propose a PM secondary NAAQS,
the 50th percentile is sufficient, as it is the basis for existing
visibility indexes used in the Denver/Colorado Front Range and Phoenix
metropolitan areas'' (Samet, 2009c, pp. 8-9). Therefore, after
considering the information that served as the original basis for its
selection as described in Ely et al., 1991, and given its acceptance
and use in existing visibility programs, the EPA continues to conclude,
consistent with the advice of CASAC, that it is reasonable to use the
50 percent acceptability criterion in determining target levels of
protection from visibility impairment.
(d) Appropriateness of using regionally varying preference study
results to select a single level for a national standard.
A number of commenters raised concerns regarding the bases for and
implications of the differences observed in the preference study
results, concluding that these results were due to regionally varying
factors and thus could not be used to set a national standard. For
example, some commenters asserted that because the confidence intervals
around the four 50 percent acceptability levels do not overlap at all,
and because there are variations in preference study designs and
inherent differences in the visual setting among cities and panels, the
four preference curves and their associated 50 percent dv values are
city-specific and statistically different. The commenters concluded,
therefore, that it was inappropriate to aggregate the 50th percentile
dv values from multiple studies and that they should instead be
evaluated individually.
Other commenters expressed the related view that the preference
study results cannot be used to set a national standard for visibility
impairment because the results show that visibility preferences vary
regionally. For example, API stated that:

The `one-size-fits-all' approach * * * is not viable because it
does not account for regional and city-specific factors that have
been made evident in the disparity of preference study data * * *.
It is well known, for example, that the level of light extinction to
which people in different areas of the country are accustomed, as
well as the urban setting, are the primary factors that affect a
person's visual perception of an urban vista. Thus, the degree to
which extinction threshold can be related to human welfare is
inevitably regionally-dependent. (API, Attachment 2, p. 4)

Some of these commenters argued that because acceptable visual air
quality is regionally dependent, it would be more appropriate to
develop distinct visibility standards at the state or local level.
Others pointed out that areas which lack ``important visibility
vistas'' might not need standards at all, since flat areas without
significant terrain have a limited maximum visual range (NEDA/CAP, p.
3).
Other commenters stated that due to regionally varying factors,
such as relative humidity, it is not possible to select a single level
for a national standard to protect visibility across the United States.
In particular, these commenters pointed to differences between Eastern
and Western areas, arguing that a single national standard could not
offer the appropriate degree of protection in locations with distinct
characteristics. For example:

[T]he proposed method falls short because it is not temporally
or geographically representative enough to have any meaning * * *.
The uncertainty evidenced in these studies and the non-uniformity
between the western and eastern vistas makes it impossible at this
time to set an acceptable light extinction value that would
appropriately address visibility concerns in non-Class I areas. (New
York DOH/DEC, pp. 5-6)

The EPA agrees that the preference curves and the 50 percent dv
levels are separate and distinct data points representing four
different VAQ preference curves for four unique urban scenes. However,
the EPA does not consider the fact that the four curves are distinct as
a weakness of the approach or a reason that the results cannot be
compared. In addition, the EPA does not agree that the study results
necessarily support a conclusion that preferences are regionally
dependent. In particular, the EPA notes that the results of Smith and
Howell (2009) which show that participants in Houston and Washington,
DC did not have significantly different views on acceptable air quality
in Washington, DC, provide limited support for the conclusion that
people's preferences differ less because of where they live and more
because of the scene they are viewing.
On the other hand, the existing literature indicates that people's
preferences for VAQ depend in large part on the characteristics and
sensitivity of the scene being viewed. The EPA understands there is a
wide variety or range of urban scenes within the United States. These
sensitive urban scenes include those with natural vistas such as the
Colorado Rocky Mountains as well as those with iconic man-made urban
structures like the Washington Monument. The EPA believes that the
scenes presented in the four urban areas

[[Page 3214]]

include important types of sensitive valued urban scenes and therefore,
when considered together, can inform the selection of a level of
acceptable urban VAQ at the national scale, taking into account the
variation across the country evidenced in the studies. This is
discussed further in the next section, below.
The EPA does agree with commenters that there are regionally
varying factors that are important to take into account when setting a
national standard for visibility protection. Section VI.A above
regarding the history of the secondary PM NAAQS review discusses the
evolution of the EPA's understanding regarding the regional differences
in PM concentrations, relative humidity and other factors. As a result,
the current review has gone to great lengths to address these factors,
leading to the EPA's proposal to use the IMPROVE algorithm to calculate
light extinction in order to take into account the varying effects of
relative humidity and speciated PM. While this approach does not result
in a uniform level of ambient PM2.5, it does ensure a
nationally uniform level of visibility protection. The EPA refers the
reader to other sections of the final rule, including sections
VI.B.1.a, VI.B.1.c, VI.C.1.b and VI.C.1.f, and the Response to Comments
document for a more detailed response as to how it is taking these
variables into account.
ii. Specific Comments on Level
The EPA received relatively few comments endorsing a specific level
for a distinct secondary standard for visibility. In general,
commenters who opposed setting a distinct secondary standard at this
time did not address the question of what level would be appropriate if
the EPA were to set a distinct secondary standard for visibility;
similarly, commenters who supported adopting a distinct secondary
standard at this time generally did not recommend a specific level.
However, a few commenters did provide comments in support of a specific
level or range of levels, with some commenters advocating standards at
the upper end of the range of proposed levels (i.e., 30 dv), while
others supported levels below the lower end of the proposed range
(i.e., below 28 dv).
As discussed above, a large number of commenters argued that the
currently available data are insufficient to determine what constitutes
a standard that would be neither more nor less protective than
necessary and that no standard should be set at this time. These
commenters pointed to the limitations and uncertainties in the
preference studies discussed above as the basis for this claim. These
commenters pointed to significant variation in the results of the
preference studies in support of their arguments that the studies
should not be used to derive a level for a distinct secondary standard
for visibility. For example, one consultant cited by several industry
commenters argued that the proposed level of 28 or 30 dv did not
reflect the substantial difference in visibility preferences between
the East and the West reflected in the preference studies (UARG,
Attachment 2, p. 11), and that it did not reflect the full range of
preferences (i.e., potential 50 percent acceptability levels) likely to
exist nationwide (UARG, Attachment 2, p. 19). This commenter further
objected to the EPA's proposal for a level of 28 or 30 dv on the
grounds that the EPA had inaccurately adjusted 4-hour values into 24-
hour values. Based on his analysis, the consultant concluded that ``a
range of adjusted values from 28 to 32 dv is needed'' to account for
the majority of the spread between the 4-hour vs. 24-hour equivalent
values at the upper end of the distribution of values.
A number of commenters questioned whether the proposed range of
levels was appropriate. One industry commenter claimed that the EPA had
not explicitly justified why a standard within the proposed range was
requisite, stating that ``EPA makes no attempt to explain how the
proposed level of the standard is neither lower nor higher than
necessary to protect public welfare'' (NSSGA, p. 15). Arizona DEQ noted
that since the proposed calculated light extinction indicator excluded
coarse particles and Rayleigh scattering, the proposed levels of 28 or
30 dv were inconsistent with the visibility preference studies, which
considered total light extinction. Noting these perceived problems with
the proposed range of levels, a few commenters noted that if the EPA
were to set a distinct secondary standard, the level should be set no
lower than 30 dv, ``to account for inconsistent value judgments, a
great deal of spatial and temporal variability, and a very high level
of uncertainty'' (Texas CEQ, p. 7).
In contrast, some commenters supporting the EPA's proposal for a
distinct secondary standard for visibility stated that the proposed
range of levels from 28-30 dv was insufficiently protective based on a
24-hour averaging time, and recommended a lower level for the
visibility index standard. These commenters expressed the view that the
proposed levels of 28 or 30 dv represented neither adequate surrogates
for equivalent 4-hour values, as the EPA claimed, nor sufficiently
protective levels based on recent air quality data. Several commenters
stated that the EPA's own analyses suggested that a standard set at a
level of 28 or 30 dv was insufficiently protective based on a 24-hour
averaging time. One commenter emphasized that the Policy Assessment had
indicated a level between 25-28 dv was appropriate for a standard
calculated on a 24-hour average, and encouraged the EPA to adopt a
standard level of 25 dv. Several environmental groups provided comments
stating that a 24-hour average would underestimate a 4-hour value by
13-42 percent and certain areas of the country--particularly the
Northeast--would be affected disproportionately. These commenters
suggested that a 24-hour PM2.5 visibility index standard
should be set at a level of 18.6-20 dv. The Department of the Interior
pointed to recent air quality data indicating that visibility on the
20% worst days in several large metropolitan areas, including
Birmingham, Fresno, New York City, Phoenix, and Washington, DC was
below 29 dv. While noting that these calculations were based on IMPROVE
calculations which include contributions from coarse PM mass, DOI
expressed the view that the proposed level of 28 to 30 dv would not
provide adequate visibility protection compared to the current 24-hour
PM2.5 standard of 35 [micro]g/m\3\ and recommended that the
standard be set at a level of 25 dv consistent with the results of the
Phoenix visibility preference study.
In contrast, the states of Arizona and Colorado submitted comments
arguing that the visibility preference studies conducted in Phoenix and
Denver, respectively, were designed to address a specific local problem
and that the results of these studies were not an appropriate basis for
selecting the level of a national standard. For example, Arizona DEQ
noted:

The cited studies were conducted considering total light
extinction; including extinction resulting from particulate matter
and Rayleigh scattering. Visibility impairment due to coarse
particulate matter can be an important contributor in Arizona,
specifically in the Phoenix area where ongoing measurements have
been made. Therefore, ADEQ believes that the proposed levels of the
secondary visibility standard are inconsistent with applicable urban
studies. (Arizona DEQ, p. 2)

Similarly, the Colorado Department of Public Health and the Environment
noted that the Denver visibility standard was designed to address
``brown clouds'', i.e., strong inversions that occur in the Denver
metropolitan area, and that this standard ``is based on a

[[Page 3215]]

specific view of Denver'' associated with particular sight paths and
direct measurement methods. The commenter stated that this standard
``is applicable only to this location,'' and that these limitations
make it potentially unsuitable for application as ``a national
secondary standard, particularly a proposed standard that does not use
a direct measurement method'' (Colorado DPHE, p. 2).
While acknowledging the uncertainties and limitations associated
with the visibility preference studies as discussed above, the EPA
continues to conclude, as did CASAC, that the preference studies are
appropriate to use as the basis for selecting a target level of
protection from visibility impairment. However, the EPA agrees with
commenters who emphasize the high degree of variability in visibility
conditions and the potential variability in visibility preferences
across different parts of the country. In light of the associated
uncertainty, as noted in the proposal, the Administrator judged it
appropriate to establish a target level of protection equivalent to the
upper end of the range of Candidate Protection Levels (CPLs) identified
in the Policy Assessment and generally supported by CASAC. Thus, the
EPA proposed to set a 24-hour visibility index standard that would
provide protection equivalent to the protection afforded by a 4-hour
standard set at a level of 30 dv. In light of the comments received on
the proposal, in particular comments emphasizing the uncertainty and
variability in the results of the public preference studies, the EPA
continues to conclude that this approach is warranted, and that it is
appropriate to set a target level of protection equivalent to the
protection that would be afforded by a 4-hour, 30 dv visibility index
standard.
Moreover, the EPA disagrees with commenters who argued that the
EPA's approach for translating 4-hour CPLs into equivalent 24-hour
values was inappropriate. In adjusting 4-hour values for purposes of
defining an appropriate level for a 24-hour standard, the EPA noted at
the time of proposal that there were multiple approaches for estimating
generally equivalent levels on a city-specific or national basis. While
expressing the view that it was appropriate to consider the two
approaches with the highest r\2\ values (Approaches A and B in Appendix
G of the Policy Assessment),\191\ which used regressions of 90th
percentile light extinction values, the EPA determined it would also be
appropriate to consider the city-specific estimates resulting from
Approaches C and E which showed greater variability than the aggregated
estimates. Approaches C and E generated a range of city-specific
estimates of generally equivalent 24-hour levels that encompassed the
range of levels considered appropriate for 4-hour CPLs, including the
CPL of 30 dv at the upper end of that range. This information provided
support for using the same CPL for a 24-hour standard as for a 4-hour
standard, since no single approach could generate an equivalent 24-hour
standard level in each urban area for each CPL. The EPA continues to
conclude, as it did at the time of proposal, that using an unadjusted
4-hour CPL for purposes of establishing a target level of protection
for a 24-hour standard is appropriate because this approach places more
emphasis on the relatively high degree of spatial and temporal
variability in relative humidity and fine particle composition observed
in urban areas across the country, consistent with EPA's reanalysis
discussed below.
---------------------------------------------------------------------------

\191\ In particular, EPA staff expressed a preference for
Approach B in the Policy Assessment. However, in light of the
additional information provided by the other approaches explored in
Appendix G of the Policy Assessment and the reanalysis in Frank, et
al. (2012b), the EPA judges it more appropriate to consider the
range of values resulting from all five analytical approaches for
purposes of informing decisions about the equivalent level of a 24-
hour standard.
---------------------------------------------------------------------------

The EPA has conducted a reanalysis (Frank et al., 2012b) of the
relationships between estimated 24-hour and 4-hour visibility
impairment based on the variety of metrics discussed in Appendix G of
the Policy Assessment. The reanalysis has more appropriately considered
the uncertainty of the calculated 4-hour values. The revised analysis
shows that the 24-hour equivalent level is generally closer to the 4-
hour value at the upper end of the range of CPLs than originally
estimated, as can be seen in the results for Approaches B, C, and
D.\192\ For example, the reanalysis indicates that Approach B yields an
adjusted 24-hour CPL of 29 dv\193\ as generally being equivalent to a
4-hour CPL of 30 dv, while Approach C yields a 24-hour equivalent CPL
of 29 dv averaged across cities and a range of city-specific values
from 25-36 dv.^{194 195} Not only are the 90th percentile and
pooled average values closer to the 4-hour CPL of 30 dv, the range of
city-specific results shows a wider spread that clearly encompasses the
unadjusted 4-hour value of 30 dv near the midpoint of the city-specific
range. This provides support for concluding that the EPA's approach to
translating of 4-hour CPLs into equivalent 24-hour values was
appropriate, and that it is appropriate to use unadjusted 4-hour values
for purposes of selecting a level for a standard based on a 24-hour
averaging time.\196\
---------------------------------------------------------------------------

\192\ Approach E as presented in the Policy Assessment is based
on the median values for each city; these results are not affected
by the regression analyses. Therefore, Approach E was not included
in the reanalysis, and the results remain unchanged from those
reported in the corrected Table G-6 as reported in Frank, et al.,
2012b.
\193\ In Appendix G of the Policy Assessment, a 24-hour adjusted
CPL of 28 dv was estimated to be equivalent to a 4-hour value of 30
dv under Approach B (annual 90th percentile values regression).
\194\ In Appendix G of the Policy Assessment, under Approach C
(all-days city-specific regression), a 24-hour adjusted CPL of 27 dv
was estimated to be equivalent to a 4-hour CPL of 30 dv when
averaged across cities, while city-specific values were estimated to
range from 24-30 dv.
\195\ In the reanalysis, Approach D (all days pooled regression)
generated results of 28 dv for the 24-hour CPL equivalent to a 4-
hour value of 30 dv as compared to a value of 27 dv in the original
analysis described in Appendix G.
\196\ The analysis in Appendix G of the Policy Assessment used
the 4-hour light extinction value treated as the independent (x-
axis) variable in an ordinary least squares regression. The EPA now
concludes that this regression approach was not the most appropriate
approach because that variable has error and in fact may be more
uncertain than the calculated 24-hour extinction values. The Frank
et al. (2012b) reanalysis uses an orthogonal regression instead of
ordinary least squares regression and results in slopes closer to
the 1:1 line for all the results, particularly for Dallas, TX.
Furthermore, consistent with the EPA's conclusion that a higher
multiplier for converting OC to OM would be appropriate (see section
VI.C.1.b.ii above), the reanalysis substitutes a 1.6 multiplier for
converting OC to OM in the calculation of 24-hour values instead of
the value of 1.4 that was used in calculating 24-hour values for
Appendix G. The higher multiplier is more consistent with the
SANDWICH approach used to calculate the 4-hour values found in
Appendix G. See Frank et al. (2012b) for a more detailed
explanation.
---------------------------------------------------------------------------

Moreover, the EPA disagrees with commenters who argue that the
currently available evidence is sufficient to justify establishing a
target level of protection at 25 dv or below. The EPA recognizes that
25 dv represents the middle of the range of 50 percent acceptability
levels from the 4 cities studied, and represents the 50 percent
acceptability level from the Phoenix study, which the Agency has
acknowledged as the best of the four studies in terms of having the
least noise in the preference study results and the most representative
selection of participants. The EPA also notes the caveats discussed in
the proposal regarding whether it would be appropriate to interpret
results from the western studies as generally representative of a
broader range of scenic vistas in urban areas across the country. The
Policy Assessment noted significant differences in the

[[Page 3216]]

characteristics of the urban scenes used in each study, with western
urban visibility preference study scenes including mountains in the
background and objects at greater distances, while scenes in the
eastern study did not. Since objects at a greater distance have a
greater sensitivity to perceived visibility changes as light extinction
changes compared to otherwise similar scenes with objects at a shorter
range, this likely explains part of the difference between the results
of the eastern study and results of the western studies. In the
proposal, the EPA noted that the scenic vistas available on a daily
basis in many urban areas across the country generally do not have the
inherent visual interest or the distance between viewer and object of
greatest intrinsic value as in the Denver and Phoenix preference
studies. Also, the Agency takes note of the caution expressed by
Colorado and Arizona about using the results of the Denver and Phoenix
preference studies, which were aimed at addressing specific local
visibility problems, to inform the choice of level for a national
standard. Therefore, the Agency considers it reasonable to conclude,
especially in light of the significant uncertainties, that it is
appropriate to place less weight on the western preference results and
that the high CPL value (30 dv) that is based on the eastern preference
results is likely to be more representative of urban areas that do not
have associated mountains or other valued objects visible in the
distant background. These areas would include the middle of the country
and many areas in the eastern U.S., as well as some western areas. As a
result, the EPA concludes that it is more appropriate to establish a
target level of