* x>
\ o^°
"'i PRO"^
Integrated Review Plan for the Review of the
Ozone National Ambient Air Quality Standards
-------
\
-------
EPA-452/R-19-002
August 2019
Integrated Review Plan for the Review of the Ozone National Ambient Air Quality Standards
U.S. Environmental Protection Agency
Office of Air Quality Planning and Standards
Health and Environmental Impacts Division
Research Triangle Park, NC
and
National Center for Environmental Assessment
Research Triangle Park, NC
-------
DISCLAIMER
This document has been prepared by staff in the U.S. Environmental Protection Agency's
Office of Air Quality Planning and Standards and National Center for Environmental
Assessment. Any findings and conclusions are those of the authors and do not necessarily reflect
the views of the Agency. This document does not represent and should not be construed to
represent any Agency determination or policy. Mention of trade names or commercial products
does not constitute endorsement or recommendation for use.
Questions or comments related to this document should be addressed to Dr. Deirdre
Murphy, U.S. Environmental Protection Agency, Office of Air Quality Planning and Standards,
C504-06, Research Triangle Park, North Carolina 27711 (email: murphy.deirdre@epa.gov).
1
-------
TABLE OF CONTENTS
LIST OF APPENDICES iv
LIST OF FIGURES iv
LIST OF TABLES iv
1 INTRODUCTION 1-1
1.1 LEGISLATIVE REQUIREMENTS 1-1
1.2 OVERVIEW OF THE NAAQS REVIEW PROCESS 1-4
1.3 PLANNED PROCESS AND PROJECTED TIMELINE FOR THIS REVIEW 1-8
2 BACKGROUND 2-1
2.1 PRIOR REVIEWS OF AIR QUALITY CRITERIA AND STANDARDS FOR
PHOTOCHEMICAL OXIDANTS INCLUDING ()3 2-1
2.2 AMBIENT AIR MONITORING AND DATA HANDLING CONVENTIONS FOR
THE CURRENT STANDARDS 2-6
2.2.1 Monitoring Requirements and the Current Monitoring Network 2-6
2.2.2 Data Handling Conventions for Comparison to the Standards 2-9
2.3 OVERVIEW 01 OZONE AIR QUALITY 2-9
3 KEY POLICY-RELEVANT ISSUES FOR THE CURRENT REVIEW 3-1
3.1 THE PRIMARY STANDARD 3-2
3.1.1 Key Issues Related to the Primary Standard 3-2
3.1.2 Background on the Current Primary Standard (Considerations and Conclusions
in the Last Review) 3-6
3.1.2.1 Considering the Need for Revision 3-9
3.1.2.2 Considering Revisions to the Standard 3-14
3.2 THE SECONDARY STANDARD 3-21
3.2.1 Key Issues Related to the Secondary Standard 3-21
3.2.2 Background on the Current Secondary Standard (Considerations and
Conclusions in the Last Review) 3-25
3.2.2.1 Considering the Need for Revision 3-27
3.2.2.2 Considering Revisions to the Standard 3-30
4 SCIENCE ASSESSMENT 4-1
4.1 PURPOSE OF THE ISA 4-1
4.2 ORGANIZATION 01 THE ISA 4-1
4.3 ASSESSMENT APPROACH 4-2
4.3.1 Introduction 4-2
4.3.2 Scope of the ISA 4-4
4.3.3 Literature Search and Identification of Relevant Studies 4-6
4.3.3.1 Systematic Literature Search 4-6
li
-------
4.3.3.2 Initial Screening (Level 1) of Studies from Literature Search 4-7
4.3.3.3 Criteria of In-Scope Studies 4-9
4.3.4 Discipline-Specific Scoping, Searching and Screening 4-9
4.3.4.1 Atmospheric Science 4-9
4.3.4.2 Exposure Assessment 4-9
4.3.4.3 Health - Experimental Studies 4-10
4.3.4.4 Health - Observational (Epidemiologic) Studies 4-12
4.3.4.5 Welfare Effects - Ecological Studies 4-14
4.3.4.6 Welfare - Effects on Climate 4-16
4.3.5 Identification of Policy-Relevant Studies 4-17
4.3.6 Evaluation of Individual Study Quality 4-18
4.3.7 Integration of Evidence and Determination of Causality 4-19
4.3.8 Quality Management 4-22
4.4 SPECIFIC SCIENCE ISSUES TO BE ADDRESSED IN THE ISA 4-23
4.4.1 Causality Determinations from 2013 ISA 4-24
4.4.2 Ambient Concentrations of O3 4-26
4.4.3 Human Exposure 4-26
4.4.4 Health Effects 4-27
4.4.5 At-Risk Lifestages and Populations and Public Health Impact 4-29
4.4.6 Welfare Effects 4-30
4.4.6.1 Ecological Effects 4-30
4.4.6.2 Effects on Climate 4-32
4.5 SCIENTIFIC AND PUBLIC REVIEW 4-33
4.5.1 Peer Input Workshop 4-33
4.5.2 Peer Review 4-33
5 QUANTITATIVE RISK AND EXPOSURE ASSESSMENTS 5-1
5.1 ASSESSMENTS INFORMING REVIEW OF THE PRIMARY STANDARD 5-3
5.1.1 Overview of Assessments in Last Review 5-5
5.1.1.1 Exposure-based Risk Analyses 5-9
5.1.1.2 Air Quality Epidemiologic Study Based Risk Analyses 5-15
5.1.2 Assessments for this Review 5-17
5.1.2.1 Key Considerations 5-17
5.1.2.2 Initial Plans for the Current Review 5-18
5.2 ASSESSMENTS INFORMING REVIEW OF THE SECONDARY STANDARD .... 5-21
5.2.1 Overview of Assessments in Last Review 5-22
5.2.1.1 Growth-related Assessments 5-23
5.2.1.2 Foliar Injury Assessments 5-26
5.2.1.3 Additional Air Quality/Exposure and E-R Analyses 5-27
5.2.2 Assessments for this Review 5-29
iii
-------
5.2.2.1 Key Considerations 5-30
5.2.2.2 Initial Plans for the Current Review 5-31
6 POLICY ASSESSMENT 6-1
7 PROPOSED AND FINAL DECISIONS 7-1
8 REFERENCES 8-1
LIST OF APPENDICES
5 A. Limitations and uncertainties of exposure and risk analyses developed in the last review of
the primary standard and consideration of related newly available information and tools.
5B. Limitations and uncertainties of air quality, exposure and risk analyses developed in the
last review of secondary standard, and consideration of related newly available
information and tools.
LIST OF FIGURES
Figure 2-1.Map of U.S. ambient air O3 monitoring sites reporting data to the EPA during the
2015-2017 period 2-8
Figure 2-2.Trends in anthropogenic emissions ofNOx (2002-2017) 2-12
Figure 2-3.Trends in anthropogenic emissions of VOC (2002-2017) 2-12
Figure 2-4.2015-2017 O3 design values across the U.S 2-14
Figure 2-5.Trends in annual 4th highest daily maximum 8-hour average O3 at all sites across
the U.S. with complete data (2000-2017) 2-15
Figure 3-1. Overview of general approach for review of the primary O3 standard 3-5
Figure 3-2. Overview of general approach for review of the secondary O3 standard 3-24
Figure 4-1. General process for development of Integrated Science Assessments 4-3
Figure 5-1. Summary of health risk assessment approaches that have been employed in
NAAQS reviews 5-4
Figure 5-2. Conceptual model for 2014 O3 health risk assessment 5-6
Figure 5-3. Analytical approach for exposure-based risk analyses 5-10
Figure 5-4. Analytical approach for epidemiologic-based analyses 5-17
LIST OF TABLES
Table 1-1. Projected timeline for completion of the review 1-9
Table 4-1. PECOS tool to define the parameters and provide a framework for identifying
relevant experimental studies 4-11
Table 4-2. PECOS tool to define the parameters and provide a framework for identifying
relevant epidemiologic studies 4-13
Table 4-3. PECOS tool to define the parameters and provide a framework for identifying
relevant ecological studies 4-16
iv
-------
Table 4-4. PECOS tool to define the parameters and provide a framework for identifying
relevant studies on the effects of tropospheric 03 on climate 4-17
Table 4-5. Weight of evidence determinations 4-20
v
-------
1 INTRODUCTION
The U.S. Environmental Protection Agency (EPA) is conducting a review of the air
quality criteria and the national ambient air quality standards (NAAQS) for photochemical
oxidants including ozone (O3). This Integrated Review Plan (IRP) contains the current plans for
this review. The review will provide an integrative assessment of relevant scientific information
and will focus on key aspects of the O3 NAAQS, including the basic elements of the standards:
the indicator,1 averaging time, form,2 and level. These elements, which together serve to define
each ambient air quality standard, are considered collectively in evaluating the protection to
public health and public welfare afforded by the standards.
This document is organized into eight chapters. Chapter 1 presents introductory
information on the legislative requirements for reviews of the NAAQS, an overview of the
review process, and a summary of the status and projected schedule for the current review.
Chapter 2 provides background information on prior reviews of the criteria and standards for
photochemical oxidants, including O3, key aspects of the ambient air monitoring requirements,
and an overview of current O3 air quality. Chapter 3 presents the general approach and a set of
policy-relevant questions intended to focus this review on the critical scientific and policy issues.
Chapters 4 through 7 discuss the planned scope and organization of key assessment documents,
the planned approaches for preparing the documents, and plans for scientific and public review
of the documents. The complete citations for references cited throughout the document are
provided in chapter 8.
1.1 LEGISLATIVE REQUIREMENTS
Two sections of the Clean Air Act (CAA) govern the establishment 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 pollutants "emissions of which, in his 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 he
"plans to issue air quality criteria... " (42 U.S.C. § 7408(a)(1)). Air quality criteria are intended
1 The "indicator" of a standard defines the chemical species or mixture that is to be measured in determining
whether an area attains the standard. The indicator of the current NAAQS for photochemical oxidants is O3
2 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. For example, the form of the annual PM2 5 NAAQS is the three-
year average of the weighted annual mean PM2 5 concentrations, while the form of the current three-month Pb
NAAQS is a three-month average concentration not to be exceeded during a three-year period.
1-1
-------
to "accurately reflect the latest scientific knowledge useful in indicating the kind and extent of all
identifiable effects on public health or welfare which may be expected from the presence of [a]
pollutant in the ambient air... " 42 U.S.C. § 7408(a)(2).
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 [42
U.S.C. § 7409(a)], Section 109(b)(1) defines primary standards as ones "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."3 Under section
109(b)(2), a secondary standard 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."4
In setting primary and secondary standards that are "requisite" to protect public health
and welfare, respectively, as provided in section 109(b), the EPA's task is to establish standards
that are neither more nor less stringent than necessary. 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, "[attainability and technological feasibility are
not relevant considerations in the promulgation of national ambient air quality standards."
American Petroleum Institute v. Costle, 665 F.2d 1176, 1185 (D.C. Cir. 1981). At the same time,
courts have clarified the EPA may consider "relative proximity to peak background ...
concentrations" as a factor in deciding how to revise the NAAQS in the context of considering
standard levels within the range of reasonable values supported by the air quality criteria and
judgments of the Administrator. American Trucking Associations, Inc. v. EPA, 283 F.3d 355, 379
(D.C. Cir. 2002).
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), cert, denied, 449 U.S. 1042 (1980);
3 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).
4 Under CAA section 302(h) (42 U.S.C. § 7602(h)), effects on welfare include, but are not limited to, "effects on
soils, water, crops, vegetation, manmade 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."
1-2
-------
American Petroleum Institute v. Costle, 665 F.2d at 1186 (D.C. Cir. 1981), cert, denied, 455 U.S.
1034 (1982); Coalition of Battery Recyclers Ass'n v. EPA, 604 F.3d 613, 617-18 (D.C. Cir.
2010); Mississippi v. EPA, 744 F.3d 1334, 1353 (D.C. Cir. 2013). 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 include 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. 5 1, Mississippi v. EPA, 744 F.3d at 1351, 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 the sensitive
population(s), and the kind and degree of uncertainties. The selection of any particular approach
to providing an adequate margin of safety is a policy choice left specifically to the
Administrator's judgment. S qq Lead Industries Association v. EPA, 647 F.2d at 1161-62;
Mississippi v. EPA, 744 F.3d at 1353.
Section 109(d)(1) of the Act requires periodic review and, if appropriate, revision of
existing air quality criteria to reflect advances in scientific knowledge on the effects of the
pollutant on public health and welfare. Under the same provision, the EPA is also to periodically
review and, if appropriate, revise the NAAQS, based on the revised air quality criteria.5
Section 109(d)(2) addresses the appointment and advisory functions of an independent
scientific review committee. Section 109(d)(2)(A) requires the Administrator to appoint this
committee, which is to be composed of "seven members including at least one member of the
National Academy of Sciences, one physician, and one person representing State air pollution
control agencies." Section 109(d)(2)(B) provides that the 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 1980s,
this independent review function has been performed by the Clean Air Scientific Advisory
Committee (CASAC) of the EPA's Science Advisory Board. A number of other advisory
functions are also identified for the committee by section 109(d)(2)(C), which reads:
5 This section of the Act requires the Administrator to complete these reviews and make any revisions that may be
appropriate "at five-year intervals."
1-3
-------
Such committee shall also (i) advise the Administrator of areas in which
additional knowledge is required to appraise the adequacy and basis of existing,
new, or revised national ambient air quality standards, (ii) describe the research
efforts necessary to provide the required information, (iii) advise the
Administrator on the relative contribution to air pollution concentrations of
natural as well as anthropogenic activity, and (iv) advise the Administrator of any
adverse public health, welfare, social, economic, or energy effects which may
result from various strategies for attainment and maintenance of such national
ambient air quality standards.
As previously noted, the Supreme Court has held that section 109(b) "unambiguously bars cost
considerations from the NAAQS-setting process" (Whitman v. Am. Trucking Associations, 531
U.S. 457, 471 [2001]). Accordingly, while some of these issues regarding which Congress has
directed the CASAC to advise the Administrator are ones that are relevant to the standard setting
process, others are not. Issues that are not relevant to standard setting may be relevant to
implementation of the NAAQS once they are established.6
1.2 OVERVIEW OF THE NAAQS REVIEW PROCESS
The process for reviewing the NAAQS has three general phases: (1) planning, (2)
assessment, and (3) decision making. Each of these phases is described in this section. The
Agency maintains a web site on which key documents developed in each phase of each NAAQS
review are made available (https://www.epa.gov/naaqs). This website also makes available
information regarding the process for NAAQS reviews, including the May 2018 memorandum
from the Administrator to Assistant Administrators (Pruitt, 2018) that describes five areas for
emphasis (principles) in the reviews and that builds on prior memoranda concerning the process
for NAAQS reviews (Peacock, 2006; Jackson, 2009).
The planning phase of each NAAQS review begins with a call for information and the
identification of issues and questions to frame the review. Drawing on this information and
issues raised in the last review, a draft IRP is prepared jointly by the EPA's National Center for
Environmental Assessment (NCEA), within the Office of Research and Development (ORD),
6 Some aspects of CASAC advice may not be relevant to EPA's process of setting primary and secondary standards
that are requisite to protect public health and welfare. Indeed, were EPA to consider costs of implementation
when reviewing and revising the standards "it would be grounds for vacating the NAAQS." Whitman, 531 U.S. at
471 n.4. At the same time, the Clean Air Act directs CASAC to provide advice on "any adverse public health,
welfare, social, economic, or energy effects which may result from various strategies for attainment and
maintenance" of the NAAQS to the Administrator under section 109(d)(2)(C)(iv). In Whitman, the Court
clarified that most of that advice would be relevant to implementation but not standard setting, as it "enable[s] the
Administrator to assist the States in carrying out their statutory role as primary implementers of the NAAQS." Id.
at 470 (emphasis in original). However, the Court also noted that CASAC's "advice concerning certain aspects of
'adverse public health ... effects' from various attainment strategies is unquestionably pertinent" to the NAAQS
rulemaking record and relevant to the standard setting process. Id. at 470 n.2.
1-4
-------
and the EPA's Office of Air Quality Planning and Standards (OAQPS), within the Office of Air
and Radiation (OAR). The draft IRP is made available for consultation with the CASAC and for
public comment. The final IRP, prepared in consideration of CASAC and public comments,
presents the current plan, projected timeline, and process for conducting the review, and also
identifies key policy-relevant issues or questions intended to guide the review.
The assessment phase of the review involves assessments of scientific information,
exposure or risk, and policy, which are described in key documents for the review. The
Integrated Science Assessment (ISA), prepared by the NCEA, provides a focused review,
synthesis, and evaluation of the most policy-relevant scientific information, including key
scientific judgments that are important to the design and scope of any exposure and risk
assessments, as well as other aspects of the NAAQS review. The ISA7 provides a comprehensive
assessment of the current scientific literature pertaining to known and anticipated effects on
public health and welfare associated with the presence of the pollutant in the ambient air,
emphasizing information that has become available since the last air quality criteria review in
order to reflect the current state of knowledge. As such, the ISA forms the scientific foundation
for each NAAQS review and is intended to provide information useful in forming policy-
relevant judgments about air quality indicator(s), form(s), averaging time(s) and level(s) for the
NAAQS. Prior to its completion in final form, the ISA, in draft form, is reviewed by the CASAC
and made available for public comment. Chapter 4 below provides a more detailed description of
the planned scope, organization and assessment approach for the ISA and its supporting
materials in this review of the air quality criteria and O3 NAAQS.
Based on the information and conclusions presented in the ISA, the EPA considers the
support provided for the development of quantitative assessments of the risks and/or exposures
for health and/or welfare effects. In so doing, the EPA considers the extent to which newly
available scientific evidence and tools/methodologies may warrant the conduct of new
quantitative risk and exposure assessments for the review.8 Key to the EPA's decision on
exposure or risk analyses that may be appropriate to develop in the review is consideration of the
newly available data, methods and tools in light of areas of uncertainty in the assessments
conducted for the last review and of the potential for new or updated assessments to provide
7 The ISA functions in the current NAAQS review process as the Air Quality Criteria Document (AQCD) did in
reviews completed prior to 2009.
8 In some reviews this consideration, and, as warranted, a general plan, including scope and methods, for conducting
the assessments, have been described in a planning document (e.g., REA Planning Document) that has been
provided to the CASAC for consultation and made available for public comment. The EPA is not planning to
prepare such a separate document in this review of the O3 NAAQS; the EPA's general considerations for
identifying the quantitative air quality exposure and risk analyses to be performed in this review are discussed in
Chapter 5 of this IRP.
1-5
-------
notably different exposure and/or risk estimates with lower associated uncertainty. Any
exposure/risk analyses performed for the review, and/or exposure/risk information developed in
the prior review that remains relevant in the current review, are considered in the policy
assessment (PA) for the review. The details regarding methods, key results, observations, and
related uncertainties are documented in a separate document accompanying the PA9 or in an
appendix to the PA. Chapter 5 includes preliminary consideration of quantitative human health-
and welfare-related assessments for this review.
The PA, prepared by the OAQPS, is a document that provides a transparent analysis
regarding the adequacy of the current standards and, as appropriate, potential alternatives for
Agency consideration prior to the issuance of proposed and final decisions. The PA integrates
and interprets the information from the ISA and from any risk and exposure analyses to frame
policy options for consideration by the Administrator. Such an evaluation of policy implications
is intended to help "bridge the gap" between the Agency's scientific assessments, presented in
the ISA and quantitative analyses, and the judgments required of the EPA Administrator in
determining whether it is appropriate to retain or revise the NAAQS. In so doing, the PA is also
intended to facilitate CASAC advice to the Agency and recommendations to the Administrator
on the adequacy of the existing standards or revisions that may be appropriate to consider, as
provided for in the CAA. In evaluating the adequacy of the current standards and, as appropriate,
a range of alternative standards, the PA considers the available scientific evidence and, as
available, quantitative risk-based analyses, together with related limitations and uncertainties.
The PA focuses on the information that is most pertinent to evaluating the basic elements of
NAAQS: indicator, averaging time, form, and level. The PA, in draft form, is released for
CASAC review and public comment prior to completion of the final PA.
The May 2018 NAAQS process memorandum identified a set of general charge questions
to be posed to the CASAC in the NAAQS review process, while recognizing that these would be
supplemented with more detailed requests as necessary (Pruitt, 2018). The general questions
cited in the May 2018 memo are as follows:
• Are there areas in which additional knowledge is required to appraise the
adequacy and basis of existing, new, or revised NAAQS? Please describe the
research efforts necessary to provide the required information.
• What scientific evidence has been developed since the last review to indicate
if the current primary and/or secondary NAAQS need to be revised or if an
alternative level or form of these standards is needed to protect public health
and/or public welfare? Please recommend to the Administrator any new
NAAQS or revisions of existing criteria and standards as may be appropriate.
9 In reviews conducted since 2008, the separate, stand-alone document presenting these analyses has been termed
the Risk and Exposure Assessment (REA).
1-6
-------
In providing advice, please consider a range of options for standard setting, in
terms of indicators, averaging times, form, and ranges of levels for any
alternative standards, along with a description of the alternative underlying
interpretations of the scientific evidence and risk/exposure information that
might support such alternative standards and that could be considered by the
Administrator in making NAAQS decisions.
• Do key studies, analyses, and assessments which may inform the
Administrator's decision to revise the NAAQS properly address or
characterize uncertainty and causality? Are there appropriate criteria to ensure
transparency in the evaluation, assessment and characterization of key
scientific evidence for this review?
• What is the relative contribution to air pollution concentrations of natural as
well as anthropogenic activity? In providing advice on any recommended
NAAQS levels, please discuss relative proximity to peak background levels.
• Please advise the Administrator of any adverse public health, welfare, social,
economic, or energy effects which may result from various strategies for
attainment and maintenance of such NAAQS.
The memo recognized that the last two charge questions may elicit information which is not
relevant to the standard-setting process under the interpretation of section 109(b) articulated by
the Supreme Court in Whitman, noting that the EPA should consider an appropriate mechanism,
including opportunities after the CASAC has provided its final advice on the standards, to
facilitate robust feedback on these topics (Pruitt, 2018). In order to facilitate meaningful advice
on these questions, the EPA issued a call for information in June 2018 that requested interested
parties to submit information on any adverse public health, welfare, social, economic, or energy
effects which may result from various strategies for attainment and maintenance of existing,
new, or revised NAAQS for consideration by the CASAC (83 FR 29784, June 26, 2018).
Separately, the EPA issued a separate call for scientific and policy-relevant information for the
current O3 NAAQS review, as noted in section 1.3 below (83 FR 29785, June 26, 2018).
Following issuance of the final PA and consideration of conclusions presented therein,
the Agency develops and publishes a notice in the Federal Register that communicates the
Administrator's proposed decisions regarding the review. A draft of this notice may undergo
interagency review involving other federal agencies prior to publication (e.g., in cases when the
proposed decision in a NAAQS review involves revision of a standard).10 Materials upon which
10 Where the proposed or final action involves NAAQS revisions for which implementation would have a large
economic effect (e.g., an annual effect on the economy of $100 million or more), such as by necessitating the
implementation of emissions controls, EPA develops and releases a regulatory impact analysis (RIA) concurrent
with the notice of proposed or final action. This activity is conducted under Executive Order 12866. The RIA is
conducted completely independent of and, by statute, is not considered in decisions regarding the review of the
NAAQS.
1-7
-------
this proposed decision is based, including the documents described above, are made available to
the public in the docket for the review. A public comment period, during which public hearings
are generally held, follows publication of the notice of the proposed action. Taking into account
comments received on the proposed decision,11 the Agency develops a notice of its final action,
which communicates the Administrator's final decisions on the review. As with the notice of
proposed action, a draft of this notice may undergo interagency review prior to publication in the
Federal Register to complete the process. Chapter 6 discusses the development of the PA and
Chapter 7 the anticipated steps for issuing a proposed and then final decision for the review.
1.3 PLANNED PROCESS AND PROJECTED TIMELINE FOR THIS
REVIEW
In May 2018, the Administrator directed his Assistant Administrators to initiate this
review of the O3NAAQS (Pruitt, 2018). In conveying this direction, the Administrator further
directed the EPA staff to expedite the review, implementing an accelerated schedule to ensure
completion of the review in 2020 (Pruitt, 2018). Accordingly, the EPA took immediate steps to
proceed with the review. In June 2018, the EPA's NCEA announced the initiation of the current
periodic review of the air quality criteria for photochemical oxidants and the O3 NAAQS and
issued a call for information in the Federal Register (83 FR 29785, June 26, 2018). Two types of
information were called for: information regarding significant new O3 research to be considered
for the ISA for the review, and policy-relevant issues for consideration in this NAAQS review.
Based in part on the information received in response to the call for information, the EPA
developed the draft IRP which was made available for consultation with the CASAC and for
public comment (83 FR 55163, November 2, 2018; 83 FR 55528, November 6, 2018).
Consultative comments from the CASAC (Cox, 2018), as well as public comments on the draft
document were considered in preparing the final IRP.
Under the plan outlined here, the current review of the O3 NAAQS is progressing on an
accelerated schedule and the EPA is incorporating a number of efficiencies in various aspects of
the review process to ensure completion within the statutorily required period (Pruitt, 2018). For
example, the kick-off workshop has been replaced with the addition of a call for policy-relevant
information coincident with the call for scientific information that traditional initiates a NAAQS
review (83 FR 29785, June 26, 2018). Also coincident with preparation of the IRP, the EPA has
begun review of the literature for consideration in the ISA, as described in Chapter 4 below. The
EPA is not planning to develop a Risk and Exposure Assessment (REA) Planning Document in
11 When issuing the final action, the Agency responds to all significant comments on the proposed decision. Where a
separate Response to Comments document is created for this purpose, it is added to the public docket for the
review, along with any additional materials upon which the final decision is based.
1-8
-------
this review; key considerations with regard to development of quantitative analyses are discussed
in Chapter 5 of this document, which was the subject of a consultation with the CASAC. Further,
the EPA has also considered combining the reviews by the CASAC and the public for some of
the main documents in a review (Pruitt, 2018). As a result, the EPA is planning to incorporate the
REA-related analyses into the PA, combining what had been two documents into a single
document for review by the CASAC and the public. Further, we are striving to ensure that initial
draft documents are sufficiently robust and complete to support a single, full review by the
CASAC and the public. The successfulness of these and other efficiencies implemented in this
review will be considered by the EPA in planning for other future NAAQS reviews (Pruitt,
2018).
The current timeline projects release of a draft ISA for CASAC review and public
comment in the latter half of 2019, with CASAC advice and public comment informing
completion of the final ISA. Comments and recommendations from the CASAC, and public
comment, on the draft PA will inform completion of the final PA, including its presentation of
options appropriate for the Administrator to consider in this review of the O3 NAAQS. The
current timeline also projects a proposed decision in the spring of 2020 and completion of the
review with a final decision in the subsequent winter.
Table 1-1. Projected timeline for completion of the review.
Key Milestones in the Review
May 2018
Administrator's memo directing initiation of the review
June 2018
Announcement and Call for Information in Federal
Register
August 2018
End comment period for Call for Information
October 2018
Draft IRP for CASAC and public comment
November 2018
CASAC consultation on draft IRP
Summer 2019
Final IRP
September 2019
Draft ISA for CASAC review and public comment
October 2019
Draft PA for CASAC review and public comment
November/December
2019
CASAC review meeting on draft ISA and draft PA
Early Spring 2020
Final ISA
Final PA
Late Spring 2020
Proposed decision
Winter 2020/2021
Final decision
1-9
-------
2 BACKGROUND
Air quality criteria were developed for photochemical oxidants in 1970 (U.S. DHEW,
1970; 35 FR 4768, March 19, 1970), and primary and secondary NAAQS were first established
in 1971 (36 FR 8186, April 30, 1971). Based on the scientific information in the 1970 air quality
criteria document (AQCD), the EPA set both primary and secondary standards at 0.08 parts per
million (ppm), as a 1-hour average of total photochemical oxidants, not to be exceeded more
than one hour per year. As summarized in section 2.1, the EPA has reviewed the air quality
criteria and standards a number of times since then, with the most recent review being completed
in 2015. An overview of the requirements for ambient air monitoring and data analysis for the
current standards are summarized in section 2.2 and current ozone air quality is summarized in
section 2.3.
2.1 PRIOR REVIEWS OF AIR QUALITY CRITERIA AND STANDARDS
FOR PHOTOCHEMICAL OXIDANTS INCLUDING 03
The EPA initiated the first periodic review of the NAAQS for photochemical oxidants in
1977. Based on the 1978 AQCD (U.S. EPA, 1978), the EPA published proposed revisions to the
original NAAQS in 1978 (43 FR 26962, June 22, 1978) and final revisions in 1979 (44 FR 8202,
February 8, 1979). At that time, the EPA changed the indicator from photochemical oxidants to
O3, revised the level of the primary and secondary standards from 0.08 to 0.12 ppm and revised
the form of both standards from a deterministic (i.e., not to be exceeded more than one hour per
year) to a statistical form. With these changes, attainment of the standards was defined to occur
when the average number of days per calendar year (across a 3-year period) with maximum
hourly average O3 concentration greater than 0.12 ppm equaled one or less (44 FR 8202,
February 8, 1979; 43 FR 26962, June 22, 1978).
Following the EPA's decision in the 1979 review, several petitioners sought judicial
review. Among those, the city of Houston challenged the Administrator's decision arguing that
the standard was arbitrary and capricious because natural O3 concentrations and other physical
phenomena in the Houston area made the standard unattainable in that area. The U.S. Court of
Appeals for the District of Columbia Circuit (D.C. Circuit) rejected this argument, holding (as
noted in section 1.1 above) that attainability and technological feasibility are not relevant
considerations in the promulgation of the NAAQS {American Petroleum Institute v. Costle, 665
F.2d at 1185). The court also noted that the EPA need not tailor the NAAQS to fit each region or
locale, pointing out that Congress was aware of the difficulty in meeting standards in some
locations and had addressed this difficulty through various compliance related provisions in the
CAA (id. at 1184-86).
2-1
-------
The next periodic reviews of the criteria and standards for O3 and other photochemical
oxidants began in 1982 and 1983, respectively (47 FR 11561, March 17, 1982; 48 FR 38009,
August 22, 1983). The EPA subsequently published the 1986 AQCD (U.S. EPA, 1986) and the
1989 Staff Paper (U.S. EPA, 1989). Following publication of the 1986 AQCD, a number of
scientific abstracts and articles were published that appeared to be of sufficient importance
concerning potential health and welfare effects of O3 to warrant preparation of a supplement to
the 1986 AQCD (U.S. EPA, 1992). In August of 1992, the EPA proposed to retain the existing
primary and secondary standards based on the health and welfare effects information contained
in the 1986 AQCD and its 1992 Supplement (57 FR 35542, August 10, 1992). In March 1993,
the EPA announced its decision to conclude this review by affirming its proposed decision to
retain the standards, without revision (58 FR 13008, March 9, 1993).
In the 1992 notice of its proposed decision in that review, the EPA announced its
intention to proceed as rapidly as possible with the next review of the air quality criteria and
standards for O3 and other photochemical oxidants in light of emerging evidence of health effects
related to 6- to 8-hour O3 exposures (57 FR 35542, August 10, 1992). The EPA subsequently
published the AQCD and Staff Paper for that next review (U.S. EPA, 1996a, b). In December
1996, the EPA proposed revisions to both the primary and secondary standards (61 FR 65716,
December 13, 1996). With regard to the primary standard, the EPA proposed to replace the then-
existing 1-hour primary standard with an 8-hour standard set at a level of 0.08 ppm (equivalent
to 0.084 ppm based on the proposed data handling convention) as a 3-year average of the annual
third-highest daily maximum 8-hour concentration. The EPA proposed to revise the secondary
standard either by setting it identical to the proposed new primary standard or by setting it as a
new seasonal standard using a cumulative form. The EPA completed this review in 1997 by
setting the primary standard at a level of 0.08 ppm, based on the annual fourth-highest daily
maximum 8-hour average concentration, averaged over three years, and setting the secondary
standard identical to the revised primary standard (62 FR 38856, July 18, 1997).
On May 14, 1999, in response to challenges by industry and others to the EPA's 1997
decision, the D.C. Circuit remanded the O3 NAAQS to the EPA, finding that section 109 of the
CAA, as interpreted by the EPA, effected an unconstitutional delegation of legislative authority
(American Trucking Assoc. v. EPA, 175 F.3d 1027, 1034-1040 [D.C. Cir. 1999]). In addition, the
court directed that, in responding to the remand, the EPA should consider the potential beneficial
health effects of O3 pollution in shielding the public from the effects of solar ultraviolet (UV)
radiation, as well as adverse health effects {id. at 1051-53). In 1999, the EPA petitioned for
rehearing en banc on several issues related to that decision. The court granted the request for
rehearing in part and denied it in part, but declined to review its ruling with regard to the
potential beneficial effects of O3 pollution {American Trucking Assoc. v. EPA, 195 F.3d 4, 10
2-2
-------
[D.C Cir., 1999]). On January 27, 2000, the EPA petitioned the U.S. Supreme Court for
certiorari on the constitutional issue (and two other issues), but did not request review of the
ruling regarding the potential beneficial health effects of O3. On February 27, 2001, the U.S.
Supreme Court unanimously reversed the judgment of the D.C. Circuit on the constitutional
issue. Whitman v. American Trucking Assoc., 531 U. S. 457, 472-74 (2001) (holding that section
109 of the CAA does not delegate legislative power to the EPA in contravention of the
Constitution). The Court remanded the case to the D.C. Circuit to consider challenges to the O3
NAAQS that had not been addressed by that court's earlier decisions. On March 26, 2002, the
D.C. Circuit issued its final decision on the remand, finding the 1997 O3 NAAQS to be "neither
arbitrary nor capricious," and so denying the remaining petitions for review. See American
Trucking Associations, Inc. v. EPA, 283 F.3d 355, 379 (D.C Cir. 2002, hereafter referred to as
"ATAIIF).
Specifically, in ATA III, the D.C. Circuit upheld the EPA's decision on the 1997 O3
standard as the product of reasoned decision making. With regard to the primary standard, the
court made clear that the most important support for the EPA's decision to revise the standard
was the health evidence of insufficient protection afforded by the then-existing standard ("the
record [is] replete with references to studies demonstrating the inadequacies of the old one-hour
standard"), as well as extensive information supporting the change to an 8-hour averaging time
{id. at 378). The court further upheld the EPA's decision not to select a more stringent level for
the primary standard noting "the absence of any human clinical studies at ozone concentrations
below 0.08 [ppm]" which supported the EPA's conclusion that "the most serious health effects
of ozone are 'less certain' at low concentrations, providing an eminently rational reason to set the
primary standard at a somewhat higher level, at least until additional studies become available"
{id. at 379, emphasis in original, internal citations omitted). The court also pointed to the
significant weight that the EPA properly placed on the advice it received from the CASAC {id. at
379). In addition, the court noted that "although relative proximity to peak background O3
concentrations did not, in itself, necessitate a level of 0.08 [ppm], EPA could consider that factor
when choosing among the three alternative levels" {id. at 379).
Coincident with the continued litigation of the other issues, the EPA responded to the
court's 1999 remand to consider the potential beneficial health effects of O3 pollution in
shielding the public from effects of UV radiation (66 FR 57268, Nov. 14, 2001; 68 FR 614,
January 6, 2003). The EPA provisionally determined that the information linking changes in
patterns of ground-level O3 concentrations to changes in relevant patterns of exposures to UV
radiation of concern to public health was too uncertain, at that time, to warrant any relaxation in
1997 O3 NAAQS. The EPA also expressed the view that any plausible changes in UV-B
radiation exposures from changes in patterns of ground-level O3 concentrations would likely be
2-3
-------
very small from a public health perspective. In view of these findings, the EPA proposed to leave
the 1997 primary standard unchanged (66 FR 57268, Nov. 14, 2001). After considering public
comment on the proposed decision, the EPA published its final response to this remand in 2003,
re-affirming the 8-hour primary standard set in 1997 (68 FR 614, January 6, 2003).
The EPA initiated the fourth periodic review of the air quality criteria and standards for
O3 and other photochemical oxidants with a call for information in September 2000 (65 FR
57810, September 26, 2000). In 2007, the EPA proposed to revise the level of the primary
standard within a range of 0.075 to 0.070 ppm (72 FR 37818, July 11, 2007). The EPA proposed
to revise the secondary standard either by setting it identical to the proposed new primary
standard or by setting it as a new seasonal standard using a cumulative form. Documents
supporting these proposed decisions included the 2006 AQCD (U.S. EPA, 2006a) and 2007 Staff
Paper (U.S EPA, 2007) and related technical support documents. The EPA completed the review
in March 2008 by revising the levels of both the primary and secondary standards from 0.08 ppm
to 0.075 ppm while retaining the other elements of the prior standards (73 FR 16436, March 27,
2008).
In May 2008, state, public health, environmental, and industry petitioners filed suit
challenging the EPA's final decision on the 2008 O3 standards. On September 16, 2009, the EPA
announced its intention to reconsider the 2008 O3 standards,12 and initiated a rulemaking to do
so. At the EPA's request, the court held the consolidated cases in abeyance pending the EPA's
reconsideration of the 2008 decision.
In January 2010, the EPA issued a notice of proposed rulemaking to reconsider the 2008
final decision (75 FR 2938, January 19, 2010). In that notice, the EPA proposed that further
revisions of the primary and secondary standards were necessary to provide a requisite level of
protection to public health and welfare. The EPA proposed to revise the level of the primary
standard from 0.075 ppm to a level within the range of 0.060 to 0.070 ppm, and to revise the
secondary standard to one with a cumulative, seasonal form. At the EPA's request, the CAS AC
reviewed the proposed rule at a public teleconference on January 25, 2010 and provided
additional advice in early 2011 (Samet, 2010, 2011). In view of the need for further
consideration and the fact that the Agency's next periodic review of the O3NAAQS required
under CAA section 109 had already begun (as announced on September 29, 2008), the EPA
decided to consolidate the reconsideration with its statutorily required periodic review.13
12 The press release of this announcement is available at:
https://archive.epa.gov/epapages/newsroom archive/newsreleases/85f90b77Ilacb0c88525763300617d0d.html.
13 This rulemaking, completed in 2015, concluded the reconsideration process.
2-4
-------
In light of the EPA's decision to consolidate the reconsideration with the current review,
the D.C. Circuit proceeded with the litigation on the 2008 final decision. On July 23, 2013, the
court upheld the EPA's 2008 primary O3 standard, but remanded the 2008 secondary standard to
the EPA (Mississippi v. EPA, 744 F. 3d 1334 [D.C. Cir. 2013]). With respect to the primary
standard, the court first rejected arguments that the EPA should not have lowered the level of the
existing primary standard, holding that the EPA reasonably determined that the existing primary
standard was not requisite to protect public health with an adequate margin of safety, and
consequently required revision. The court went on to reject arguments that the EPA should have
adopted a more stringent primary standard. With respect to the secondary standard, the court held
that the EPA's explanation for the setting of the secondary standard identical to the revised 8-
hour primary standard was inadequate under the CAA because the EPA had not adequately
explained how that standard provided the required public welfare protection.
At the time of the court's decision, the EPA had already completed significant portions of
its next statutorily required periodic review of the O3 NAAQS. This review had been formally
initiated in 2008 with a call for information in the Federal Register (73 FR 56581, September 29,
2008). In late 2014, based on the Integrated Science Assessment (ISA), Risk and Exposure
Assessments (REAs) for health and welfare, and PA14 developed for this review, the EPA
proposed to revise the 2008 primary and secondary standards by reducing the level of both
standards to within the range of 0.070 to 0.065 ppm (79 FR 75234, December 17, 2014).
The EPA's final decision in this review was published in October 2015, establishing the
now-current standards (80 FR 65292, October 26, 2015). In this decision, based on consideration
of the health effects evidence on respiratory effects of O3 in at-risk populations, the EPA revised
the primary standard from a level of 0.075 ppm to a level of 0.070 ppm, while retaining all the
other elements of the standard (80 FR 65292, October 26, 2015). The EPA's decision on the
level for the standard was based on the weight of the scientific evidence and quantitative
exposure/risk information. The level of the secondary standard was also revised from 0.075 ppm
to 0.070 ppm based on the scientific evidence of O3 effects on welfare, particularly the evidence
of O3 impacts on vegetation, and quantitative analyses available in the review.15 The other
elements of the standard were retained. This decision on the secondary standard also
incorporated the EPA's response to the D.C. Circuit's remand of the 2008 secondary standard in
Mississippi v. EPA, 744 F.3d 1344 (D.C. Cir. 2013). The 2015 revisions to the NAAQS were
14 The final versions of these documents, released in August 2014, were developed with consideration of the
comments and recommendations from the CASAC, as well as comments from the public on the draft documents
(U.S. EPA 2014a; U.S. EPA, 2014b; U.S. EPA, 2014c; Frey, 2014a; Frey, 2014b; Frey, 2014c).
15 The standards set in 2015 (generally referred to as the current standards in this document) are specified at 40 CFR
50.19.
2-5
-------
accompanied by revisions to the data handling procedures, and the ambient air monitoring
requirements16 (80 FR 65292, October 26, 2015).17
After publication of the final rule, a number of industry groups, environmental and public
health organizations, and certain states filed petitions for judicial review in the D.C. Circuit. The
industry and state petitioners filed briefs arguing that the revised standards are too stringent,
while the environmental and health petitioners' brief argued that the revised standards are not
stringent enough to protect public health and welfare as the Act requires. On August 23, 2019,
the court issued an opinion that denied all the petitions for review with respect to the 2015
primary standard while also concluding that the EPA had not provided a sufficient rationale for
aspects of its decision on the 2015 secondary standard and remanding that standard to the EPA
(Murray Energy v. EPA, No. 15-1385, Order, Doc. No. 1803352 [D.C. Cir. Aug. 23, 2019]).
2.2 AMBIENT AIR MONITORING AND DATA HANDLING
CONVENTIONS FOR THE CURRENT STANDARDS
2.2.1 Monitoring Requirements and the Current Monitoring Network
State and local environmental agencies operate O3 monitors at state or local air
monitoring stations (SLAMS) as part of the SLAMS network. The requirements for the SLAMS
network depend on the population and most recent O3 design values18 in the area. The minimum
number of O3 monitors required in a metropolitan statistical area (MSA) ranges from zero for
areas with a population less than 350,000 and no recent history of an O3 design value greater
than 85 percent of the level of the standard, to four for areas with a population greater than 10
million and an O3 design value greater than 85 percent of the standard level.19 Within an O3
monitoring network, at least one site for each MSA must be designed to record the maximum
concentration for that particular metropolitan area. Since the highest O3 concentrations tend to be
associated with a particular season for various locations, the EPA requires O3 monitoring during
16 The current federal regulatory measurement methods for O3 are specified in 40 CFR 50, Appendix D and 40 CFR
part 53. Consideration of ambient air measurements with regard to judging attainment of the standards is
specified in 40 CFR 50, Appendix U. The O3 monitoring network requirements are specified in 40 CFR 58.
17 This decision additionally announced revisions to the exceptional events scheduling provisions, as well as changes
to the air quality index and the regulations for the prevention of significant deterioration permitting program.
18 A design value is a statistic that describes the air quality status of a given area relative to the level of the standard,
taking the averaging time and form into account, as well as any data handling requirements (e.g., for the 2015 O3
NAAQS, these requirements are specified in Appendix U to 40 CFR Part 50), Design values are typically used to
classify nonattainment areas as meeting or not meeting the standard, to assess progress towards meeting the
NAAQS, and to develop control strategies.
19 The SLAMS minimum monitoring requirements to meet the O3 design criteria are specified in 40 CFR Part 58,
Appendix D. The minimum O3 monitoring network requirements for urban areas are listed in Table D-2 of
Appendix D to 40 CFR Part 58 (accessible at https://www.ecfr.gov).
2-6
-------
specific O3 monitoring seasons which vary by state from five months (May to September in
Oregon and Washington) to all twelve months (in a number of states).20
Most of the state, local, and tribal air monitoring stations that report data to the EPA use
ultraviolet Federal Equivalent Methods. The Federal Reference Method (FRM) was revised in
2015 to include a new chemiluminescence by nitric oxide (NO-CL) method. The previous
ethylene (ET-CL) method is no longer commonly used due to lack of availability and safety
concerns with ethylene.21 The NO-CL method is beginning to be implemented in the SLAMS
network.
In 2017, there were over 1,300 federal, state, local, and tribal ambient air monitors
reporting O3 concentrations to the EPA. Figure 2-1 shows the locations of such monitoring sites
that reported data to the EPA at any time during the 2015-2017 period. About 80% of this
network are SLAMS monitors operated by state and local governments to meet regulatory
requirements and provide air quality information to public health agencies; these sites are largely
focused on urban and suburban areas.
Two important subsets of SLAMS sites separately make up the National Core (NCore)
multi-pollutant monitoring network and the Photochemical Assessment Monitoring Stations
(PAMS) network. Each state is required to have at least one NCore station, and O3 monitors at
NCore sites are required to operate year-round. At each NCore site located in a CBS A with a
population of 1 million or more (based on the most recent census), a PAMS network site is
required.22 Monitors at PAMS are required to operate during the months of June, July and
August, although monitoring may extend over longer periods of time to improve the usefulness
of data collected during an area's O3 season (U.S. EPA, 2018a).
In addition to reporting O3 concentrations, the NCORE and PAMS networks provide data
on O3 precursor chemicals. The NCore sites feature co-located measurements of chemical
species such as nitrogen oxide and total reactive nitrogen, along with meteorological
measurements. The additional data collected at the PAMS sites include measurements of
nitrogen oxides (NOx), and a target set of volatile organic compounds (VOC). The enhanced
monitoring at sites in these two networks informs our understanding of local O3 formation.
20 The required O3 monitoring seasons for each state are listed in Table D-3 of Appendix D to 40 CFR Part 58.
21 The current FRM for 03 (established in 2015) is a chemiluminescence method. This is an automated method
allowing for the measurement of O3 concentrations in ambient air using continuous (real-time) sampling and
analysis. This method is based on continuous automated measurement of the intensity of the characteristic
chemiluminescence released by the gas phase reaction of O3 in sampled air with either ethylene or nitric oxide
gas. This method is fully described in Appendix D to 40 CFR Part 50.
22 The requirements for PAMS, which were most recently updated in 2015, is fully described in Appendix D to 40
CFR Part 58.
2-7
-------
While the SLAMS network has a largely urban and population-based focus, there are
monitoring sites in other networks that can be used to track compliance with the NAAQS in rural
areas (see Figure 2-1). For example, the Clean Air Status and Trends Network (CASTNET)
monitors which are located in rural areas. There were about 80 CASTNET sites operating in
2017, with most of the sites in the eastern U.S. being operated by the EPA, and most of the sites
in the western U.S. being operated by the National Park Service (NPS). Finally, there are also a
number of Special Purpose Monitoring Stations (SPMs), which are not required but are often
operated by air agencies for short periods of time (less than 3 years) to collect data for human
health and welfare studies, as well as other types of monitoring sites, including monitors
operated by tribes and industrial sources. The SPMs are typically not used to assess compliance
with the NAAQS.23
• %
••
V
*1
• •
• SLAMS • CASTNET • NCORE/PAMS • SPM/OTHER
Figure 2-1. Map of U.S. ambient air O3 monitoring sites reporting data to the EPA during
the 2015-2017 period.
23 However, SPMs that use federal reference or equivalent methods, meet all applicable requirements in 40 CFR Part
58. and operate continuously for at least 3 years may be used to assess compliance with the NAAQS.
2-8
-------
2.2.2 Data Handling Conventions for Comparison to the Standards
To assess whether a monitoring site or geographic area meets or exceeds a NAAQS, the
monitoring data are analyzed consistent with the established regulatory requirements for the
handling of monitoring data for the purposes of deriving a design value. A design value
expresses ambient air concentrations in terms of the averaging time and form for a given
standard such that its comparison to the level of the standard indicates whether the location
meets or exceeds the standard. Consistent with the form and averaging time of the O3 standards,
O3 design values for the standards established in the last review are calculated as the 3-year
average of the annual fourth highest daily maximum 8-hour average O3 concentration.
Hourly average O3 concentrations at the monitoring sites used for assessing compliance
with the NAAQS are required to be reported in ppm to the third decimal place, with additional
digits truncated, consistent with the typical measurement precision associated with most O3
monitoring instruments. The hourly average concentrations are used to compute moving 8-hour
average concentrations for each day, with the daily maximum 8-hour average identified as the
highest of the 17 consecutive, valid24 8-hour averages that begin with the 8-hour period from
7am to 3pm and end with the 8-hour period from 11pm to 7am the subsequent day.25 An O3
monitoring site meets the standard if its design value is less than or equal to the level of the
standard. A geographic area meets the NAAQS if all ambient air monitoring sites in the area
have valid26 design values meeting the standard, and if one or more monitors has a design value
exceeding the standard, then the area exceeds the NAAQS.
2.3 OVERVIEW OF OZONE AIR QUALITY
Ozone is a gas composed of three oxygen atoms (O3). It is naturally present in the Earth's
atmosphere, both in the stratospheric layer occurring roughly 10 to 30 miles above the Earth's
surface as well as in the closer tropospheric layer. The stratosphere contains a large reservoir of
O3 (i.e. the "ozone layer") that results naturally from photochemical reactions between ultraviolet
24 An 8-hour average is considered valid if at least six of the hourly concentrations are available or if substitution of
zero for the missing hourly concentrations yields an 8-hour average above the level of the standard. The 8-hour
averages are required to be reported to three decimal places with additional digits to right of third decimal place
truncated (Appendix U to 40 CFR Part 50).
25 A daily maximum concentration is considered valid if at least 13 of the 17 consecutive 8-hour averages are
available or if the daily maximum based on fewer than 13 is greater than the level of the standard (Appendix U to
40 CFR Part 50).
26 An O3 design value less than or equal to the level of the standard is valid if daily maximum values are available
for at least 90% of the days in the O3 monitoring season on average over the 3 years, with a minimum of 75%
data completeness in any individual year (Appendix U to 40 CFR Part 50). A design value greater than the level
of the standard is always valid.
2-9
-------
light (UV) and molecular oxygen (O2).27 Under specific meteorological conditions, this reservoir
can contribute to O3 concentrations at the Earth's surface (Langford et al., 2017). Ozone is also
produced near the earth's surface due to chemical interactions involving solar radiation and
pollution resulting from human activity. These chemical reactions involve specific O3 precursors,
such as NOx, VOC, and carbon monoxide, which can be emitted from both natural and
anthropogenic sources.28
Global air quality models have estimated that natural sources of O3 precursors, such as
vegetation, lightning, and wildfires, can produce daily 8-hour peak O3 concentrations of 15-35
parts per billion by volume (ppb) across the U.S. during the warm season (2014 PA, section
2.4.1). Human activity from combustion of fossil fuels or biomass and the use of industrial and
consumer chemicals can also lead to emissions of these O3 precursors, which can then yield O3
concentrations substantially above naturally occurring levels. The EPA conducted air quality
modeling analyses in the last review to assess the role of natural sources (i.e., natural
background) and the combined impacts of natural background plus anthropogenic sources
outside of the U.S. (i.e., U.S. background) on O3 concentrations (2014 PA, section 2.4).29 These
2007-based annual modeling analyses (presented in the 2014 PA) estimated that seasonal mean
natural background levels ranged from 15 to 35 ppb over the U.S. This modeling also estimated
that seasonal mean daily maximum 8-hour average concentrations of U.S. background O3 ranged
from 25 to 50 ppb. While the majority of modeled events greater than 70 ppb were primarily
driven by local and regional O3 precursor emissions, there were some events with substantial
U.S. background contributions where O3 concentrations approached or exceeded 75 ppb (80 FR
65300, October 26, 2015).30
As part of the current review, the EPA plans to utilize state-of-the-science air quality
modeling for a more recent time period, 2016, to provide updated estimates of the relative
contributions of natural and anthropogenic sources of O3 in the U.S. Specifically, the EPA
intends to use the Community Multiscale Air Quality (CMAQ) modeling system (Appel et al.,
27 This layer of O3 in the upper atmosphere helps to protect the earth's populations and ecosystems from the
damaging effects of UV radiation (Norval et al., 2011; Bais et al., 2017).
28 Methane emissions can also contribute to O3 formation, but its impacts are more frequently observed at the global
scale over longer time periods (e.g., decadal scale).
29 The difference between natural and U.S. background is that U.S. background also includes, along with
contributions from natural sources, the impacts from anthropogenic emissions outside the U.S.
30 Noting the infrequency of such events, and of the statutory and regulatory provisions that allow for the exclusion
of air quality monitoring data substantially affected by certain background influences (e.g., wildfires or
stratospheric intrusions) from design value calculations when they meet certain criteria, the EPA explained in the
2015 decision that background concentrations of O3 were not expected to preclude attainment of a revised O3
standard with a level of 70 ppb (80 FR 65328, October 26, 2015).
2-10
-------
2017) over a Northern Hemisphere domain to provide boundary conditions for a finer-scale
national application of CMAQ to estimate current levels of background ozone using recently
available emissions estimates and meteorological data.31 Using this model configuration, the
EPA plans to conduct, evaluate,32 and summarize the results of a series of "zero-out" sensitivity
runs33 designed to isolate natural background and U.S. background.34 While the model estimates
of background O3 will be based on a single year (2016), the EPA will also consider the potential
implications of interannual variability on these estimates.
Based on estimates compiled in version 2 of the 2014 National Emissions Inventory
(NEI) (U.S. EPA, 2018b), biogenic and fire emissions comprise 78 percent35 of the total VOC
emissions in the U.S., but only 9% of the NOx emissions.36 Mobile sources, such as on-road
vehicles and non-road equipment, are the largest contributors of NOx emissions. Figures 2-2 and
2-3 show the downward trends in anthropogenic source emissions of NOx and VOC based on
estimates for the last 15 years.37 Emissions of NOx decreased by more than 40% and VOC
emissions by more than 15% since 2002.
31 The modeling analyses conducted in the review completed in 2015 used boundary conditions from the global
GEOS-Chem model (Henderson et al., 2014) as inputs into regional models (e.g., CMAQ) to estimate background
levels (2014 PA, section 2.4).
32 Model performance will be assessed using a variety of 03 measurements, including data from upper atmosphere
monitoring tools (e.g., global ozonesonde networks) and surface air monitoring sites within and outside the U.S.
33 Zero-out sensitivity modeling refers to a commonly used method for isolating the O3 impacts of specific emissions
source categories or sources from specific regions. To accomplish this, O3 concentrations are estimated from
model simulations in which emissions of interest are set to zero. As an example, natural background could be
estimated from a simulation in which all anthropogenic emissions are zeroed out in the simulation.
34 These analyses can be used to facilitate CASAC advice on CAA Section 109(d)(2)(c)(iii) (e.g., as discussed Pruitt
[2018]).
35 In locations near large concentrations of anthropogenic VOC sources (e.g., in certain urban areas or oil and gas
development basins), the relative contribution of anthropogenic sources can be much higher than the national
average.
36 The NEI is updated every three years based on emissions estimate data provided by state, local, and tribal air
agencies for sources in their jurisdiction and supplemented by national data developed by the EPA. The 2014
version of the NEI is the latest currently NEI dataset.
37 The estimates of long-term annual emissions are drawn from
https ://gispub .epa. gov/air/trendsreport/2018/#naaas trends (as of March 2019). The process for deriving these
estimates from the NEI is described at https://www.epa.gov/air-emissions-inventories/air-pollutant-emissions-
trends-data
2-11
-------
Anthropogenic NOx Emissions
30 = stationary fuel combustion
{5 25 #a industrial & other processes
.highway
inonroad
° 15
Sk-
O
<2 10
o
S 5
0
ilill
2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017
Figure 2-2. Trends in anthropogenic emissions of NOx (2002-2017).
Anthropogenic VOC Emissions
30
ra 25
£_ 20
- stationary fuel combustion
a industrial & other processes
fj highway
¦ nonroad
O 15
10
2002 2003 2004 2005 2006 2007 2008 2009 20102011 2012 2013 2014 2015 2016 2017
Figure 2-3. Trends in anthropogenic emissions of VOC (2002-2017).
The chemistry that leads to O3 formation is complex and can vary depending upon the
relative proportions of different types of precursor pollutants as well as external conditions such
as temperature and sunlight. Over most areas of the U.S., daytime O3 production typically
increases as NOx concentrations increase (2013 ISA, section 3.2.4). Formation of O3 in this
regime is described as "NOx-limited." At other times and locations, where NOx concentrations
are higher, O3 formation may be only weakly dependent on NOx emissions, or even inversely
2-12
-------
correlated (i.e., NOx emissions actually deplete O3 locally38). O3 formation in these regimes
increases as VOC concentrations increase and is described as "VOC-limited." Once formed, O3
near the Earth's surface can be transported by the prevailing winds before eventually being
removed from the atmosphere over the course of hours to weeks via chemical reactions or
deposition to surfaces.
As described in section 2.2.1, to assess O3 concentrations across the U.S., state and local
environmental agencies operate O3 monitors at various locations and subsequently submit the
data to the EPA for analyses and storage. As shown in Figure 2-4, several locations across the
U.S. have design values in 2015-2017 that exceeded the standard level of 70 ppb. California
contains numerous monitoring sites where design values exceeded 70 ppb in 2015-2017, but high
O3 was also measured in Texas, the Northeast Corridor, along the Lake Michigan shoreline, and
certain urban areas in the western U.S. These locations include some of the most densely
populated areas in the country that also experience conducive meteorology for O3 formation. The
highest daily peak 8-hour average O3 concentrations most commonly occur during the afternoon
within the warmer months due to higher solar radiation and other conducive meteorological
conditions during these times. However, there can be exceptions such as the Uintah Basin in
Utah where the highest O3 concentrations occur during the winter on sunny days with strong
temperature inversions and ample snow cover.
38 In these cases, NOx generally results in eventual net ozone production downwind of the emissions sources over
longer time scales.
2-13
-------
2015-2017 O3 design values across the U.S.
-o
3?
• 43-60 ppb (187 sites) O 66 - 70 ppb (334 sites) • 76 - 112 ppb (108 sites)
©61-65 ppb (373 sites) © 71-75 ppb (136 sites)
Figure 2-4. 2015-2017 O3 design values across the U.S. Red and orange circles indicate
locations exceeding the standard. Design values available at:
https://www.epa.gov/air-trends/air-qualitv-design-values.
Concentrations of O3 in the U.S. have trended downward over the past several decades
due to reductions in precursor emissions noted above. The average downward trend in annual
fourth highest 8-hour daily maximum O3 concentration has been 17% between 2000 and 2017
(U.S. EPA, 2018c), as shown in Figure 2-5 (based on 809 monitoring sites that operated for the
full 18-year period). Downward trends in this metric have been even more substantial in the
Eastern U.S. and in California. Air quality model simulations estimate that O.3 air quality will
continue to improve over the next decade as additional reductions in O3 precursors from mobile
sources, industrial processes, and other sources are realized as a result of "on-the-books" EPA
regulations (U.S. EPA, 2015b; Collet et al., 2017) and other technological changes. In addition to
being affected by changing emissions, future O3 concentrations may also be affected by climate
change (Nolte et al., 2018) as well as any changes in the amount of O3 transported into the U.S.
from other countries (He et al., 2016).
2-14
-------
o
o
o
3
to o
> 00
X
>
\
\
v -
~
*
s
s
\
v*
N.
\
\
\
¦Vi
v-;
- - -
- - -
"X.
V
\
\
s
s
¦Ni
V
— \
\
V
N
*s
•S.
~
~
-x.
•X.
s
o
o
o
CM
r- CJ
o o
o o
CM Cvl
CO
O
O
CJ
O
o
CM
10
o
o
CM
CD
O
O
CJ
r-
o
o
a
CO
o
o
CJ
(7)
O
O
CJ
o
o
CJ
o
CJ
CM
o
CJ
CO
o
CM
o
CJ
ID
o
CJ
CD
o
CJ
o
CM
Figure 2-5. Trends in annual 4th highest daily maximum 8-hour average 0.< concentrations
and design values at all sites across the U.S. with complete data (2000-2017).
2-15
-------
3 KEY POLICY-RELEVANT ISSUES FOR THE
CURRENT REVIEW
The overarching question in each NAAQS review is:
• Do the currently available scientific evidence and exposure/risk-based information
support or call into question the adequacy of the protection afforded by the current
standard(s)?
As appropriate, a review also addresses a second overarching question:
• What alternative standards, if any, are supported by the currently available
scientific evidence and exposure/risk-based information and are appropriate for
consideration?
In considering these overarching questions, a series of key policy-relevant issues
particular to a given review are addressed.
The policy-relevant issues thus far identified for this review of the O3 standards are
presented in sections 3.1.1 and 3.2.1 below as series of questions intended to frame our approach
to considering the information available in this review of the current primary and secondary
standards for O3. The ISA and PA developed in this new review39 will provide the basis for
addressing these questions and will inform the Administrator's judgment as to whether the
current primary and secondary standards for O3 provide the requisite protection of public health
and public welfare, and his decisions as to whether to retain or revise these standards. These
assessments focus on policy-relevant scientific information and analyses that address key
questions related to the adequacy of the O3 standards.40 In this chapter, the primary standard is
discussed in section 3.1 and the secondary standard in section 3.2.
39 As summarized in sections 1.2 and 1.3 above, stand-alone REA documents will not be developed for this review.
Rather, any exposure and risk analyses performed for this review will be presented in the PA along with any such
information from the last review that remains informative in this review, taking into account the newly available
evidence presented in the ISA and any other technical documents prepared for the review.
40 Several examples of policy-relevant analyses in NAAQS reviews, generally, are noted in Pruitt (2018): "EPA's
Integrated Science Assessments (ISA), Risk and Exposure Assessments (REA), and Policy Assessments (PA)
should focus on policy-relevant science and on studies, causal determinations, or analyses that address key
questions related to the adequacy of primary and secondary NAAQS, including levels near - both above and
below—the current standard(s). Policy-relevant science may also include information that directly relates to the
indicator, averaging time, form and level of a NAAQS as well as alternative policy approaches."; "In developing
additional analyses in the REA or elsewhere, EPA should focus on policy-relevant including consideration of
issues such as thresholds or background levels, as appropriate for context."
3-1
-------
3.1 THE PRIMARY STANDARD
The approach planned for this review of the primary standard is most fundamentally
based on using the Agency's assessment of the current scientific evidence, quantitative
assessments of exposures and/or risks, and other associated analyses (e.g., air quality analyses) to
inform the Administrator's judgments regarding a primary standard that is requisite to protect
public health with an adequate margin of safety. This approach involves translating scientific and
technical information into the basis for addressing a series of key policy-relevant questions using
both evidence- and exposure-/risk-based considerations. This series of key questions related to
the primary standard is presented in section 3.1.1, along with a summary of the general approach
for the review. Additionally, to provide context for this review of the current primary O3
standard, section 3.1.2 summarizes key aspects of the decisions made in the last review,
including the Agency's consideration of important policy judgments concerning the scientific
evidence and exposure/risk information, and associated uncertainties and limitations, as well as
the Administrator's public health policy judgments regarding an adequate margin of safety.
3.1.1 Key Issues Related to the Primary Standard
The approach planned for this review of the primary O3 standard will build on the
substantial body of work developed during the course of the last review, taking into account the
more recent scientific information and air quality data now available to inform our understanding
of the key-policy relevant issues in this review. The ISA, risk and exposure analyses (as
warranted), and PA developed in this review will provide the basis for addressing the key policy-
relevant questions in the review and these documents will inform the Administrator's decisions
as to whether to retain or revise the primary O3 standard. As summarized in section 1.2, and also
described in chapter 6, evaluations in the PA are intended to inform the Administrator's public
health policy judgments and decisions. In so doing, the PA considers the potential implications
of various aspects of the scientific evidence, the exposure/risk-based information, and the
associated uncertainties and limitations.
In building upon the conclusions from the last review, the current review takes into
account the updated evidence and information that has become available since that review. The
Agency's consideration of the full set of evidence and information available in this review will
inform the answer to the following initial overarching question for the review:
• Do the currently available scientific evidence and exposure-/risk-based information
support or call into question the adequacy of the public health protection afforded by
the current primary O3 standard?
In reflecting on this question, we will consider the available body of scientific evidence,
assessed in the ISA, and used as a basis for developing or interpreting risk/exposure analyses,
3-2
-------
including whether it supports or calls into question the scientific conclusions reached in the last
review regarding health effects related to exposure to ambient air-related O3. Information
available in this review that may be informative to public health judgments regarding
significance or adversity of key effects will also be considered. Additionally, the currently
available exposure and risk information, whether newly developed in this review or
predominantly developed in the past and interpreted in light of current information, will be
considered, including with regard to the extent to which it may continue to support judgments
made in the last review. Further, in considering this question with regard to the primary O3
standard, as in all NAAQS reviews, we give particular attention to exposures and health risks to
at-risk populations.41
Evaluation of the available scientific evidence and risk/exposure information with regard
to this consideration of the current standard will focus on key policy-relevant issues by
addressing a series of questions including the following:
• Is there newly available evidence that indicates the importance of photochemical oxidants
other than O3 with regard to abundance in ambient air, and potential for human exposures
and health effects?
• Does the currently available scientific evidence alter our conclusions from the last review
regarding the nature of health effects attributable to human exposure to O3 from ambient air?
• Does the current evidence alter our understanding of populations that are particularly at risk
from O3 exposures?
• Does the current evidence alter our conclusions from the previous review regarding the
exposure duration and concentrations associated with health effects? To what extent does the
currently available scientific evidence indicate health effects attributable to exposures to O3
concentrations lower than previously reported and what are important uncertainties in that
evidence?
• To what extent have previously identified uncertainties in the health effects evidence been
reduced or do important uncertainties remain? Have new ones been identified?
• What are the nature and magnitude of O3 exposures and associated health risks associated
with air quality conditions just meeting the current standard?
• To what extent are the estimates of exposures and risks to at-risk populations associated with
air quality conditions just meeting the current standard reasonably judged important from a
public health perspective?
• What are the important uncertainties associated with any risk/exposure estimates?
41 As used here and similarly throughout this document, the term population refers to persons having a quality or
characteristic in common, such as a specific pre-existing illness or a specific age or life stage. Some populations
may be at increased risk of health effects occurring with exposure to O3 as a result of any of a variety of factors,
including genetic or developmental aspects, disease or smoking status, and factors related to socioeconomic
status, reduced access to health care or increased exposure.
3-3
-------
If the information available in this review suggests that revision of the current primary
standard would be appropriate to consider, the PA will evaluate how the standard might be
revised based on the available scientific information, air quality assessments, and exposure/risk
information, and also considering what the available information indicates as to the health
protection expected to be afforded by the current or potential alternative standards. Such an
evaluation may consider the effect of revision of one or more elements of the standard (indicator,
averaging time, level and form), with the effect being evaluated based on the resulting potential
standard and all of its elements collectively. Based on such evaluations, the PA would then
identify potential alternative standards (specified in terms of indicator, averaging time, level, and
form) intended to reflect a range of alternative policy judgments as to the degree of protection
that is requisite to protect public health with an adequate margin of safety, and options for
standards expected to achieve it. The specific policy-relevant questions that frame such
evaluation of what revision of the standard might be appropriate to consider include:
• Does the currently available information call into question the identification of ozone as the
indicator for photochemical oxidants? Is support provided for considering a different
indicator?
• Does the currently available information call into question the current averaging time? Is
support provided for considering different averaging times for the standard?
• What does the currently available information indicate with regard to a range of levels and
forms of alternative standards that may be supported and what are the uncertainties and
limitations in that information?
• What do the available analyses indicate with regard to exposure and risk associated with
specific alternative standards? What are the associated uncertainties? To what extent might
such alternatives be expected to reduce adverse impacts attributable to O3, and what are the
uncertainties in the estimated reductions?
3-4
-------
The approach to reaching conclusions on the current primary standard and, as
appropriate, on potential alternative standards is summarized in general terms in Figure 3-1.
Adequacy of Current Standard
Evidence-Based Consideraaons
»- Does currency avaSabte evidence and related
unceraanSes strengthen orcal in® quesfcn prior
conclusions?
¦ Evidence of heath effects no! previously
idenifed?
¦ Newly Merited at-risk populates?
¦ Evidence of heath. effecs at lower levels or for
different exposure du radons?
¦ Uncertainties jdenafed in the las review
reduced or new uncertainies emerged?
Exposure- and R&fc-Based Considerations
>Nature, magniude, and importance of
estimated exposures and risks associated
wflti just mee&ig the current standard?
> Uncertainties in the exposure and risk
estimaies?
Consider Retaining \
Does
information call into
question adequacy
of current
standard?
Consider Potential Alternative Standards
Current Standard j
y
t
Demerits of Po:ernaJ Aternafive Standards
> Irvdica&or, Averaging Time, Form, Level
Potential Alternative Standards for Consideration
Figure 3-1. Overview of general approach for review of the primary O3 standard.
3-5
-------
The Agency's approach in reviewing primary standards is consistent with requirements
of the provisions of the CAA related to the review of the NAAQS and with how the EPA and the
courts have historically interpreted the CAA. As discussed in section 1.1 above, these provisions
require the Administrator to establish primary standards that, in the Administrator's judgment,
are requisite (i.e., neither more nor less stringent than necessary) to protect public health with an
adequate margin of safety. The CAA does not require the Administrator to establish a primary
standard at a zero-risk level or at background concentration levels, but rather at a level that
reduces risk sufficiently so as to protect public health with an adequate margin of safety. The
decisions on the adequacy of the current primary standard and, on any alternative standards
considered in a review, are largely public health policy judgments made by the Administrator.
The four basic elements of the NAAQS (i.e., indicator, averaging time, form, and level) are
generally considered collectively in evaluating the health protection afforded by the current
standard, and by any alternatives considered. The Administrator's final decisions in a review
draw upon the scientific evidence for health effects, quantitative analyses of populations
exposures and/or health risks, as available, and judgments about how to consider the
uncertainties and limitations that are inherent in the scientific evidence and quantitative analyses.
3.1.2 Background on the Current Primary Standard (Considerations and Conclusions in
the Last Review)
The 2015 decision to strengthen the primary standard was based on the scientific
evidence and quantitative exposure and risk analyses available at the time of the last review, the
Administrator's judgments regarding the available scientific evidence, the appropriate degree of
public health protection for the revised standard, and the available exposure and risk information
regarding the exposures and risk that may be allowed by such a standard (80 FR 65292, October
26, 2015). With the 2015 decision, the EPA revised the level of the primary standard from 0.075
to 0.070 ppm,42 in conjunction with retaining the then-current indicator (O3), averaging time
(eight hours), and form (fourth-highest daily maximum 8-hour average concentration, averaged
across three consecutive years). The 2015 decision drew upon the available scientific evidence
assessed in the 2013 ISA, the exposure and risk information presented and assessed in the 2014
health REA (HREA), the consideration of that evidence and information in the 2014 PA, the
advice and recommendations of the CASAC, and public comments on the proposed decision (80
FR 65292, October 26, 2015; U.S. EPA., 2015a).
42 Although ppm are the units in which the level of the standard is defined, the units ppb are more commonly used
throughout the next three chapters for greater consistency with their use in the more recent literature. The level of
the current primary and secondary standards, 0.070 ppm, is equivalent to 70 ppb.
3-6
-------
The health effects evidence base available in the 2015 review included extensive
longstanding evidence from previous reviews as well as the evidence that had emerged since the
prior review had been completed in 2008. This evidence base, spanning several decades,
documents the causal relationship between exposure to O3 and a broad range of respiratory
effects (2013 ISA, p. 1-14). Such effects range from small, reversible changes in pulmonary
function and pulmonary inflammation (documented in controlled human exposure studies
involving exposures ranging from 1 to 8 hours) to more serious effects such as emergency
department visits and hospital admissions, which have been associated with ambient air
concentrations of O3 in epidemiologic studies (2013 ISA, section 6.2). In addition to extensive
controlled human exposure and epidemiologic studies, the evidence base includes experimental
animal studies that provide insight into potential modes of action for these effects, contributing
to the coherence and robust nature of the evidence. Based on this evidence base, the 2013 ISA
concluded there to be a causal relationship between short-term O3 exposures and respiratory
effects and a likely causal relationship between longer-term exposure and respiratory effects, and
also between short-term exposure and mortality (2013 ISA, p. 1-14).43
With regard to the short-term respiratory effects, that were the primary focus of the 2015
decision, the controlled human exposure studies were recognized to provide the most certain
evidence indicating the occurrence of health effects in humans following specific O3 exposures
(80 FR 65343, October 26, 2015; 2014 PA, section 3.4). These studies additionally illustrate the
role of ventilation rate in eliciting responses to O3 exposure at the lowest studied concentrations.
The exposure concentrations eliciting a given level of response in subjects at rest are higher than
for subjects exposed while at elevated ventilation, such as while exercising (2013 ISA, section
6.2.1.1).44 Further, while the study subjects in the vast majority of the controlled human
exposure studies (and in all of these studies conducted at the lowest exposures) are healthy
43 The 2013 ISA also concluded that there is likely to be a causal relationship between short-term exposure and
cardiovascular effects, including related mortality, and that the evidence at that time was suggestive of causal
relationships between long-term O3 exposures and total mortality, cardiovascular effects and reproductive and
developmental effects, and between O3 exposure and central nervous system effects (2013 ISA, section 2.5.2).
44 In the controlled human exposure studies, the magnitude of respiratory effects (e.g., size of lung function
decrements and prevalence in symptomatic responses) is influenced by ventilation rate and exposure duration as
well as exposure concentration, with physical activity increasing ventilation and potential for effects. In studies of
healthy young adults exposed while at rest for 2 hours, 500 ppb is the lowest concentration eliciting a statistically
significant Ch-induced group mean lung function decrement, while a much lower concentration produces a
statistically significant response in lung function when the ventilation rate of the group of study subjects is
sufficiently increased with exercise (2013 ISA, section 6.2.1.1). For example, the lowest exposure concentration
examined that elicited a statistically significant Ch-induced group mean lung function decrement in an exposure
of 2 hours or less was 120 ppb in a 1-hour exposure of trained cyclists who maintained a high exertion level
throughout the exposure period (2013 ISA, section 6.2.1.1; Gong et al., 1986).
3-7
-------
adults, the 2013 ISA identified several groups as being at increased risk of Cb-related effects. In
light of this finding with regard to children and adults with asthma, the HREA exposure-based
analyses included these population groups as being among those modeled (2014 HREA, p. 3-14).
The exposure and risk information available in the 2015 review included exposure and
risk estimates for air quality conditions just meeting the then-existing standard, and also for air
quality conditions just meeting potential alternative standards. Estimates were derived for two
exposure-based analyses, the first of which involved comparison of population exposure
estimates at elevated exertion to exposure benchmarks (exposures of concern)45 based on
exposure concentrations from controlled human exposure studies in which lung function changes
and other effects were measured in healthy, young adult volunteers exposed to O3 while
engaging in quasi-continuous moderate physical activity for a defined period (generally 6.6
hours).46 The second exposure-based analysis provided population risk estimates of the
occurrence of days with Cb-attributable lung function decrements of varying magnitudes.47 Risk
estimates were also derived from ambient air concentrations based on concentration-response
functions from epidemiologic studies but were given less weight by the Administrator in her
decision on the standard, given conclusions reached in the PA and the HREA which reflected
lower confidence in these estimates (80 FR 65316-17, October 26, 2015).
The 2014 HREA developed the exposure-based estimates for several population groups
including all children and all adults. The estimates involving comparison of exposures to
benchmarks were also derived for children with asthma and adults with asthma. The estimates of
percentages of children with exposures above benchmarks were virtually indistinguishable from
the corresponding estimates of percentages of children with asthma.48 When considered in terms
of the absolute number of children, the estimates for all children were much higher than those for
children with asthma, with the magnitude of the differences varying based on asthma prevalence
in each study area (2014 HREA, sections 5.3.2, 5.4.1.5 and section 5F-1). The estimates for
percent of children above the benchmarks were higher than percent of adults due to the greater
45 The benchmark concentrations to which exposure concentrations experienced while at moderate or greater
exertion were compared were 60, 70 and 80 ppb. This comparison-to-benchmarks analysis, performed in the 2015
review, is summarized in section 5.1.1.1 below.
46 The studies given primary focus were those for which 03 exposures occurred over the course of 6.6 hours during
which the subjects engaged in six 50-minute exercise periods separated by 10-minute rest periods, with a 35-
minute lunch period occurring after the third hour (e.g., Follinsbee et al., 1988 and Schelegle et al., 2009).
Responses after O3 exposure were compared to those involving filtered air.
47 Both exposure-based analyses are described further in section 5.1 below.
48 This reflects use of the same time-location-activity diary pool to construct each simulated individual's time-
activity series, which is based on the similarities observed in the available diary data with regard to time spent
outdoors and exertion levels (2014 HREA, sections 5.3.2 and 5.4.1.5).
3-8
-------
time that children spend outdoors and engaged in exertion (2014 HREA, section 5.3.2). Thus,
consideration of the exposure-based results in the 2015 decision focused on the results for all
children and children with asthma.
In weighing the 2013 ISA conclusions with regard to the health effects evidence and
making judgments regarding the public health significance of the quantitative estimates of
exposures and risks allowed by the then-existing and alternative standards, as well as judgments
regarding margin of safety, the Administrator considered the currently available information and
commonly accepted guidelines or criteria within the public health community, including the
American Thoracic Society (ATS), an organization of respiratory disease specialists,49 advice
from the CASAC and public comments. In so doing, she recognized that the determination of
what constitutes an adequate margin of safety is expressly left to the judgment of the EPA
Administrator {LeadIndustries Association v. EPA, 647 F.2d at 1161-62; Mississippi, 744 F. 3d
at 1353). In NAAQS reviews generally, evaluations of how particular primary standards address
the requirement to provide an adequate margin of safety include consideration of such factors as
the nature and severity of the health effects, the size of the sensitive population(s) at risk, and the
kind and degree of the uncertainties present. Consistent with past practice and long-standing
judicial precedent, the Administrator took the need for an adequate margin of safety into account
as an integral part of her decision-making.
The Administrator's initial decision in the last review was with regard to the adequacy of
protection provided by the then-existing primary standard. Considerations related to that decision
are summarized in section 3.1.2.1 below. The considerations and decisions on revisions to the
then-existing standard in order to provide the requisite protection under the Act, including an
adequate margin of safety, is summarized in section 3.1.2.2.
3.1.2.1 Considering the Need for Revision
The approach to considering the adequacy of the then-current primary standard in the last
review involved the careful consideration of the available evidence, analyses and conclusions
contained in the 2013 ISA, including information newly available in the review; the quantitative
exposure and risk analyses in the 2014 HREA; the information, evaluations, considerations and
conclusions presented in the 2014 PA; advice from the CASAC; and public comment. Key
considerations informing the Administrator's decision on the need for revision of the then-
current standard are summarized below.
The Administrator gave primary consideration to the evidence of respiratory effects from
controlled human exposure studies, including those newly available in the review, and for which
49 With regard to commonly accepted guidelines or criteria within the public health community, the PA considered
statements issued by the ATS that had also been considered in prior reviews (ATS, 2000; ATS, 1985).
3-9
-------
the exposure concentrations were at the lower end of those studied (80 FR 65343, October 26,
2015). This emphasis was consistent with CAS AC comments on the strength of this evidence
(Frey, 2014, p. 5). In placing weight on these studies, the Administrator took note of the variety
of respiratory effects reported from the studies of healthy adults engaged in six 50-minute
periods of moderate exertion within a 6.6-hour exposure to O3 concentrations of 60 and higher.
The most severe respiratory effects have been reported, and the broadest range of effects have
been studied and reported (lung function decrements, respiratory symptoms, airway
inflammation, airway hyperresponsiveness, and impaired lung host defense) following exposures
to 80 ppb O3 or higher, with most exposure studies conducted at these higher concentrations. The
combination of lung function decrements and respiratory symptoms was reported following
exposures for which the average concentration during the exercise periods was 72 ppb,50 and
lung function decrements and pulmonary inflammation were reported following exposures to O3
concentrations as low as 60 ppb. In considering these findings, the Administrator noted that the
combination of 03-induced lung function decrements and respiratory symptoms meets ATS
criteria for an adverse response.51 She additionally recognized the CASAC comments on this
point and also its caution that these study findings were for healthy adults indicating the potential
for such effects in some people, such as people with asthma, at lower exposures (Frey, 2014c,
pp. 5-6). In light of this, the Administrator concluded that "the controlled human exposure
studies indicate that adverse effects are likely to occur following exposures to O3 concentrations
below the level of the [then-current] standard" (80 FR 65343, October 26, 2015).
The 2013 ISA indicated that the pattern of effects observed across the range of exposures
assessed in the controlled human exposure studies, increasing with severity at higher exposures,
is coherent with (i.e., reasonably related to) the health outcomes reported to be associated with
ambient air concentrations in epidemiologic studies (e.g., respiratory-related hospital admissions,
emergency department visits). With regard to the available epidemiologic studies, the
Administrator noted analyses of O3 air quality in the 2014 PA indicating that, while most O3
epidemiologic studies reported health effect associations with O3 concentrations in ambient air
that violated the then-current standard, a small number of single-city U.S. studies indicate the
occurrence of asthma-related hospital admissions and emergency department visits at ambient air
O3 concentrations below the level of the then-current standard. In particular, the Administrator
took note of a study that reported associations between short-term O3 concentrations and asthma
50 For the 70 ppb target exposure, Schelegle et al. (2009) reported that the mean O3 concentration for the six 50-
minute exercise periods was 72 ppb.
51 The most recent statement from the ATS available at the time of the 2015 decision stated that "[i]n drawing the
distinction between adverse and nonadverse reversible effects, this committee recommended that reversible loss
of lung function in combination with the presence of symptoms should be considered as adverse" (ATS, 2000).
3-10
-------
emergency department visits in children and adults in a U.S. location that would have met the
then-current standard over the entire 5-year study period (80 FR 65344, October 26, 2015; Mar
and Koenig, 2009).52 53 While uncertainties54 limited the extent to which the Administrator based
her conclusions on air quality in locations of multicity epidemiologic studies, she additionally
noted some support from several multicity studies of morbidity or mortality in which the
majority of study locations would have met the then-current standard (80 FR 65344, October 26,
2015; 2014 PA, section 3.1.4.2). Accordingly, looking across the body of epidemiologic
evidence, the Administrator reached the conclusion that analyses of air quality in some study
locations supported the occurrence of adverse 03-associated effects at O3 concentrations in
ambient air that met, or are likely to have met, the then-current standard (80 FR 65344, October
26, 2016). Taken together, the Administrator concluded that the scientific evidence from
controlled human exposure and epidemiologic studies called into question the adequacy of the
public health protection provided by the then-current standard.
In considering the exposure and risk information, the Administrator gave particular
attention to the estimates of exposures of concern, focusing on the estimates for children, in 15
urban areas for air quality conditions just meeting the then-current standard. Consistent with the
finding that larger percentages of children than adults were estimated to experience exposures
above benchmarks, the Administrator focused on the results for all children and for children with
asthma, noting that the results for these two groups, in terms of percent of the population group,
are virtually indistinguishable (2014 HREA, sections 5.3.2, 5.4.1.5 and section 5F-1). In
considering these estimates, she placed greatest weight on estimates of two or more days with
occurrences of exposures above benchmarks, in light of her increased concern about the potential
for adverse responses with repeated occurrences of such exposures. In particular, she noted that
the types of effects shown to occur following exposures to O3 concentrations from 60 ppb to 80
52 The design values in this location over the study period were at or somewhat below 75 ppb (Wells et al., 2012).
53 The Administrator also took note of analyses in the PA for some single-city study locations where the then-current
standard was not met during the study period (i.e., those evaluated in Silverman and Ito, 2010; Strickland et al.,
2010), finding support for the association of hospital admissions and emergency department visits with short-term
O3 on subsets of days with virtually all ambient air O3 concentrations below the level of the then-current standard.
These analyses generally focused on the range of short-term concentrations for which the confidence intervals for
the concentration-response relationship were tightest, finding these to be on many days with O3 concentrations
below the level of the standard (80 FR 65344, October 26, 2015).
54 Compared to the single-city epidemiologic studies the Administrator noted additional uncertainty in interpreting
the relationships between short-term O3 air quality in individual study cities and reported 03 multicity effect
estimates. This uncertainty applied specifically to interpreting air quality analyses within the context of multicity
effect estimates for short-term O3 concentrations, where effect estimates for individual study cities are not
presented (as is the case for the key O3 studies analyzed in the PA, with the exception of the study by Stieb et al.
(2009) where none of the city-specific effect estimates for asthma emergency department visits were statistically
significant) (80 FR 65344; October 26, 2015).
3-11
-------
ppb, such as inflammation, if occurring repeatedly from repeated exposure, could potentially
result in more severe effects based on the ISA conclusions regarding mode of action (80 FR
65343, 65345, October 26, 2015; 2013 ISA, section 6.2.3). While generally placing greatest
weight on estimates of repeated exposures, the Administrator also considered estimates for single
exposures above the higher benchmarks of 70 and 80 ppb (80 FR 65345, October 26, 2015).
With regard to multiple exposures, the HREA found that under conditions just meeting
the then-current standard, fewer than 1% of children in the 15 study areas would be estimated to
experience multiple days in a year with 8-hour exposures at or above 70 ppb while at elevated
ventilation, while the percentage was as high as approximately 2% in the year and location with
the highest exposure estimates (80 FR 65345 and Table 1, October 26, 2015). Although she
expressed less concern with single occurrences, the Administrator noted that the then-current
standard could allow just over 3% of children to experience one or more days, averaged over the
years of analysis, with an 8-hour exposure at or above 70 ppb (while at moderate or greater
exertion), based on the worst-case location, and up to 8% in the worst-case year and location (80
FR 65345, October 26, 2015). She additionally noted that, that in the worst-case year and
location across the 15 study areas, the then-current standard could allow up to about 1% of
children to experience at least one day per year with 8-hour exposures at elevated ventilation at
or above 80 ppb, the highest benchmark evaluated (80 FR 65345, October 26, 2015).55
In considering the HREA estimates of days with exposures at or above 60 ppb, while
expressing less confidence in the adversity of effects observed following exposures as low as 60
ppb, particularly single exposures, she judged the potential for adverse effects to increase with
repeated exposures, as noted above (80 FR 65345, October 26, 2015). In that light, she noted that
the HREA found that under air quality conditions just meeting the then-current standard,
approximately 3 to 8% of children in the 15 urban study areas (including approximately 3 to 8%
of asthmatic children), on average across the years of analysis, were estimated to experience two
or more days per year with 8-hour exposures at or above 60 ppb, while at elevated ventilation (80
FR 65345; October 26, 2015).
In considering these exposure estimates with regard to public health implications, the
Administrator concluded that the exposures and risks projected to remain upon meeting the then-
current standard could reasonably be judged to be important from a public health perspective. In
particular, this conclusion was based on her judgment that it is appropriate to set a standard that
would be expected to eliminate, or almost eliminate, the occurrence of exposures, while at
moderate exertion, at or above 70 and 80 ppb. In addition, given that the average percent of
55 The Administrator additionally noted that the then-current standard could allow up to about 3% of children to
experience one or more days with 8-hour exposures at elevated ventilation at or above 70 ppb, averaged over the
years of analysis across the 15 study areas (80 FR 65313, Table 1, October 26, 2015).
3-12
-------
children estimated to experience two or more days with exposures at or above the 60 ppb
benchmark approaches 10% in some urban study areas (on average across the analysis years), the
Administrator concluded that the then-current standard does not incorporate an adequate margin
of safety against the potentially adverse effects that could occur following repeated exposures at
or above 60 ppb (80 FR 65345-46; October 26, 2015).
With regard to the HREA estimates of lung function risk in terms of decrements in forced
expiratory volume in one second (FEVi), the Administrator also gave greater weight to estimates
of multiple occurrences than to single occurrences, while additionally noting CASAC advice
regarding uses of FEVi decrement estimates as scientifically relevant surrogates for adverse
health outcomes (Frey, 2014c, p. 3). The Administrator noted that, when averaged over the years
of evaluation, the then-current standard was estimated to allow about 1 to 3% of children in the
15 urban study areas to experience two or more Cb-induced lung function decrements >15%, and
to allow about 8 to 12% of children to experience two or more Cb-induced lung function
decrements >10% (80 FR 65346, October 26, 2015). The Administrator concluded that these
HREA estimates for lung function risk, as well as the epidemiologic-study-based risk estimates
(although she recognized increased uncertainty in and placed less weight on both types of
estimates) further support a conclusion that the 03-associated health effects estimated to remain
upon just meeting the then-current standard are an issue of public health importance on a broad
national scale. Thus, she concluded that O3 exposure and risk estimates, when taken together,
support a conclusion that the exposures and health risks associated with just meeting the then-
current standard can reasonably be judged to be of public health significance, such that the then-
current standard was not sufficiently protective and did not incorporate an adequate margin of
safety.
In addition to the evidence and exposure/risk information, the Administrator also took
note of CASAC advice, which included the finding that "the current NAAQS for ozone is not
protective of human health" and the unanimous recommendation "that the Administrator revise
the current primary ozone standard to protect public health" (Frey, 2014c, p. 5). She further
noted similar CASAC advice in the prior 2008 review.56
In consideration of all of the above, the Administrator concluded that the then-current
primary O3 standard was not requisite to protect public health with an adequate margin of safety,
and that it should be revised to provide increased public health protection. This decision was
based on the Administrator's conclusions that the available evidence and exposure and risk
information clearly called into question the adequacy of public health protection provided by the
56 The CASAC O3 Panel for the 2008 review likewise recommended revision of the standard to one with a level
below 75 ppb. This earlier recommendation was based entirely on the evidence and information in the record for
the 2008 decision, which had been expanded in the 2015 review (Samet, 2011; Frey and Samet, 2012).
3-13
-------
then-current primary standard such that it was "not appropriate, within the meaning of section
109(d)(1) of the CAA, to retain the current standard" (80 FR 65346, October 26, 2015).
3.1.2.2 Considering Revisions to the Standard
The following subsections summarize the Administrator's key considerations and
conclusions in considering revisions to the indicator, averaging time, form and level of the
primary standard in the 2015 review.
3.1.2.2.1 Indicator
In considering whether O3 continued to be the most appropriate indicator for a standard
meant to provide protection against photochemical oxidants in ambient air, the Administrator
considered findings and assessments in the 2013 ISA and 2014 PA, as well as advice from the
CAS AC and public comment. The 2013 ISA specifically noted that O3 is the only photochemical
oxidant (other than nitrogen dioxide) that is routinely monitored and for which a comprehensive
database exists (2013 ISA, section 3.6; 80 FR 65347, October 26, 2015). The PA additionally
noted that, since the precursor emissions that lead to the formation of O3 also generally lead to
the formation of other photochemical oxidants, measures leading to reductions in population
exposures to O3 can generally be expected to lead to reductions in other photochemical oxidants.
The CASAC indicated its view that O3 is the appropriate indicator "based on its causal or likely
causal associations with multiple adverse health outcomes and its representation of a class of
pollutants known as photochemical oxidants" (Frey, 2014c, p. ii). Based on all of these
considerations and public comments, the Administrator concluded that O3 remains the most
appropriate indicator for a standard meant to provide protection against photochemical oxidants
in ambient air, and she retained O3 as the indicator for the primary standard (80 FR 65347,
October 26, 2015).
3.1.2.2.2 Averaging time
The 8-hour averaging time for the primary O3 standard was established in 1997 with the
decision to replace the then-existing 1-hour standard with an 8-hour standard (62 FR 38856, July
18, 1997). The decision in that review was based on evidence from numerous controlled human
exposure studies reporting associations between adverse respiratory effects and 6- to 8-hour
exposures, as well as quantitative analyses indicating the control provided by an 8-hour
averaging time of both 8-hour and 1-hour peak exposures and associated health risk (62 FR
38861, July 18, 1997; U.S. EPA, 1996b). The decision at that time was also consistent with
advice from the CASAC (62 FR 38861, July 18, 1997; 61 FR 65727; December 13, 1996). The
EPA reached similar conclusions in the subsequent 2008 review in which the 8-hour averaging
time was retained (73 FR 16436, March 27, 2008).
3-14
-------
In the review completed in 2015, the Administrator considered the averaging time for the
standard in light of both the strong evidence for Cb-associated respiratory effects following
short-term exposures and the available evidence related to effects following longer-term
exposures (80 FR 65347-50, October 26, 2015). In so doing, the Administrator noted the
substantial health effects evidence from controlled human exposure studies that demonstrate that
a wide range of respiratory effects (e.g., pulmonary function decrements, increases in respiratory
symptoms, lung inflammation, lung permeability, decreased lung host defense, and airway
hyperresponsiveness) occur in healthy adults following exposures ranging from 1 to 8 hours (80
FR 65348, October 26, 2015; 2013 ISA, section 6.2.1.1). The Administrator also noted the
strength of evidence from epidemiologic studies that evaluated a wide variety of populations
(e.g., including at-risk lifestages and populations, such as children and people with asthma,
respectively) using a number of different short-term averaging times, including the maximum 1-
hour concentration within a 24-hour period (1-hour max), the maximum 8-hour average
concentration within a 24-hour period (8-hour max), and the 24-hour average (80 FR 65348,
October 26, 2015; 2013 ISA, chapter 6). It was recognized that an 8-hour averaging time is
similar to the exposure periods evaluated in the more recent controlled human exposure studies
conducted at the lowest concentrations, and the Administrator noted that the epidemiologic
evidence alone did not provide a strong basis for distinguishing between the appropriateness of
1-hour, 8-hour and 24-hour averaging times. Thus, in consideration of the then-available health
effects information, the Administrator concluded that an 8-hour averaging time remained
appropriate for addressing health effects associated with short-term exposures to ambient air O3
(80 FR 65348, October 26, 2015).
In considering the evidence related to longer-term exposures, the Administrator initially
considered the extent to which currently available evidence and exposure/risk information
suggested that a standard with an 8-hour averaging time can provide protection against
respiratory effects associated with longer-term exposures to ambient air O3. As in previous
reviews, the review completed in 2015 recognized and further evaluated changes in long-term air
quality patterns in response to attaining an 8-hour standard and the reduction in potential risk of
health effects associated with long-term exposures in areas meeting an 8-hour standard (80 FR
65348, October 26, 2015). Analyses described in detail in the HREA suggested that reductions in
O3 precursors emissions in order to meet a standard with an 8-hour averaging time, coupled with
the appropriate form and level, would be expected to reduce long-term O3 concentrations
reported in epidemiologic studies to be associated with respiratory morbidity and mortality (80
FR 65348, October 26, 2015).
In summary, based on the then-available evidence and information discussed in detail in
the 2013 ISA, 2014 HREA, and 2014 PA, along with CAS AC advice and public comments, the
3-15
-------
Administrator concluded that a standard with an 8-hour averaging time could effectively limit
health effects attributable to both short- and long-term O3 exposures. Furthermore, the
Administrator observed that the CASAC agreed with the choice of averaging time (Frey, 2014c,
p. ii). Therefore, the Administrator concluded it to be appropriate to retain the 8-hour averaging
time and to not set a separate standard with a different averaging time (80 FR 65350, October 26,
2015).
3.1.2.2.3 Form
While giving foremost consideration to the adequacy of public health protection provided
by the combination of all elements of the standard, including the form, the Administrator placed
considerable weight on the findings from prior reviews with regard to the use of the //th-high
metric, as described below (80 FR 65350-65352, October 26, 2015). Based on these findings and
consideration of CASAC advice, the Administrator judged it appropriate to retain the fourth-high
form, more specifically the fourth-highest daily maximum 8-hour O3 average concentration,
averaged over 3 years (80 FR 65352, October 26, 2015).
The concentration-based form was established in the 1997 review when it was recognized
that such a form better reflects the continuum of health effects associated with increasing O3
concentrations than an expected exceedance form, which had been the form of the standard prior
to 1997. Unlike an expected exceedance form, a concentration-based form gives proportionally
more weight to years when 8-hour O3 concentrations are well above the level of the standard
than years when 8-hour O3 concentrations are just above the level of the standard. More weight
was given to high O3 concentrations, in light of the available health evidence that indicated a
continuum of effects associated with exposures to varying concentrations of O3, and because the
extent to which public health is affected by exposure to O3 in ambient air is related to the actual
magnitude of the O3 concentration, not just whether the concentration is above a specified level.
With regard to a specific concentration-based form, the fourth-highest daily maximum was
selected in 1997, recognizing that a less restrictive form (e.g., fifth highest) would allow a larger
percentage of sites to experience O3 peaks above the level of the standard, and would allow more
days on which the level of the standard may be exceeded when the site attains the standard (62
FR 38868-38873, July 18, 1997).
In the subsequent 2008 review, the EPA considered the potential value of a percentile-
based form, recognizing that such a statistic is useful for comparing datasets of varying length
because it samples approximately the same place in the distribution of air quality values, whether
the dataset is several months or several years long (73 FR 16474, March 27, 2008). However, the
EPA concluded that, because of the differing lengths of the monitoring season for O3 across the
U.S., a percentile-based statistic would not be effective in ensuring the same degree of public
3-16
-------
health protection across the country. Specifically, a percentile-based form would allow more
days with higher air quality values (i.e., higher O3 concentrations) in locations with longer O3
seasons relative to locations with shorter O3 seasons. Thus, the EPA concluded in the 2008
review that a form based on the //th-highest maximum O3 concentration would more effectively
ensure that people who live in areas with different length O3 seasons received the same degree of
public health protection (73 FR 16474-75, March 27, 2008). At that time, it was also recognized
that it is important to have a form that provides stability with regard to implementation of the
standard. In the case of O3, for example, it was noted that it was important to have a form that
provides stability and insulation from the impacts of extreme meteorological events that are
conducive to O3 formation. Such events could have the effect of reducing public health
protection, to the extent they result in frequent shifts in and out of attainment due to
meteorological conditions because such frequent shifting could disrupt an area's ongoing
implementation plans and associated control programs (73 FR 16475, March 27, 2008).
In the 2015 review, the Administrator continued to recognize the considerations
supporting the decisions in 1997 and 2008, and additionally noted recent CAS AC advice in
which the CASAC indicated that the O3 standard should be based on the fourth-highest, daily
maximum 8-hour average value (averaged over 3 years), by stating that this form "provides
health protection while allowing for atypical meteorological conditions that can lead to
abnormally high ambient ozone concentrations which, in turn, provides programmatic stability"
(Frey, 2014c, p. 6; 80 FR 65352, October 26, 2015).
3.1.2.2.4 Level
The Administrator's decision to revise the level of the primary O3 standard to 70 ppb
built upon her conclusion (summarized in section 3.1.2.1 above) that the overall body of
scientific evidence and exposure/risk information called into question the adequacy of the public
health protection afforded by the then-current standard, particularly for at-risk populations and
lifestages (80 FR 65362, October 26, 2015). In her decision on level, the Administrator placed
the greatest weight on the results of controlled human exposure studies and on quantitative
analyses based on information from these studies, particularly analyses of O3 exposures of
concern. The Administrator viewed the results of the lung function risk assessment, analyses of
O3 air quality in locations of epidemiologic studies, and epidemiology-based quantitative health
risk assessment as providing information in support of her decision to revise the then-current
standard, but of less utility for selecting a particular standard level among a range of options (80
FR 65362, October 26, 2015). In placing weight on information from controlled human exposure
studies and analyses based on information from these studies, the Administrator noted that
controlled human exposure studies provide the most certain evidence indicating the occurrence
3-17
-------
of health effects in humans following specific O3 exposures, noting in particular that the effects
reported in the controlled human exposure studies are due solely to O3 exposures, and are not
complicated by the presence of co-occurring pollutants or pollutant mixtures (as is the case in
epidemiologic studies). The Administrator's emphasis on the information from the controlled
human exposure studies was consistent with the CASAC's advice and interpretation of the
scientific evidence (80 FR 65362, October 26, 2015; Frey, 2014c). In this regard, the
Administrator recognized that (1) the largest respiratory effects, and the broadest range of
effects, have been studied and reported following exposures to 80 ppb O3 or higher (i.e.,
decreased lung function, increased airway inflammation, increased respiratory symptoms, airway
hyperresponsiveness, and decreased lung host defense); (2) exposures to O3 concentrations
somewhat above 70 ppb have been shown to both decrease lung function and to result in
respiratory symptoms; and (3) exposures to O3 concentrations as low as 60 ppb have been shown
to decrease lung function and to increase airway inflammation (80 FR 65363, October 26, 2015).
The Administrator considered both ATS recommendations and CASAC advice to inform her
judgments on the potential adversity to public health of effects reported in controlled human
exposure studies (80 FR 65363, October 26, 2015). In doing so, the Administrator concluded that
the evidence from controlled human exposure studies provided strong support for the conclusion
that a revised standard with a level of 70 ppb is requisite to protect public health with an
adequate margin of safety. This conclusion was based, in part, on the fact that such a standard
level would be well below the O3 exposure concentration documented to result in the widest
range of respiratory effects (i.e., 80 ppb), and below the lowest O3 exposure concentration shown
to result in the adverse combination of lung function decrements and respiratory symptoms (80
FR 65363, October 26, 2015).
In considering the degree of protection provided by a revised primary O3 standard, the
Administrator considered the extent to which that standard would be expected to limit population
exposures to the broad range of O3 exposures shown to result in health effects (80 FR 65363,
October 26, 2015). In considering the exposure estimates from the HREA, the Administrator
focused on the estimates of two or more exposures of concern in order to provide a health-
protective approach to considering the potential for repeated occurrences of exposures that could
result in adverse effects. In so doing, she placed the most emphasis on setting a standard that
appropriately limits repeated occurrences of exposures while at elevated ventilation at or above
the 70 and 80 ppb benchmarks. She noted that a revised standard with a level of 70 ppb was
estimated to eliminate the occurrence of two or more days with exposures at or above 80 ppb and
to virtually eliminate the occurrence of two or more days with exposures at or above 70 ppb for
3-18
-------
all children and children with asthma, even in the worst-case year and location evaluated.57
Given the considerable protection provided against repeated exposures of concern for all
benchmarks evaluated in the HREA, the Administrator judged that a standard with a level of 70
ppb incorporated a margin of safety against the adverse 03-induced effects shown to occur in the
controlled human exposure studies (80 FR 65364, October 26, 2015).
While she was less confident that adverse effects would occur following exposures to O3
concentrations as low as 60 ppb,58 as discussed above, the Administrator judged it to also be
appropriate to consider estimates of exposures (while at moderate or greater exertion) for the 60
ppb benchmark (80 FR 65363-64, October 26, 2015). In so doing, she recognized that while
CASAC advice regarding the potential adversity of effects observed in studies of 60 ppb was less
definitive than for effects observed at the next higher concentration studied, the CASAC did
clearly advise the EPA to consider the extent to which a revised standard is estimated to limit the
effects observed in studies of 60 ppb exposures (80 FR 65364, October 26, 2015; Frey, 2014c).
The Administrator's consideration of exposures at or above the 60 ppb benchmark was primarily
in the context of considering the extent to which the health protection provided by a revised
standard included a margin of safety against the occurrence of adverse 03-induced effects. In this
context, the Administrator noted that a revised standard with a level of 70 ppb was estimated to
protect the vast majority of children in urban study areas (i.e., about 96% to more than 99% of
children in individual areas) from experiencing two or more days with exposures at or above 60
ppb (while at moderate or greater exertion). Compared to the estimates for the then-current
standard, this represented a reduction of more than 60%. Given the considerable protection
provided against repeated exposures of concern for all of the benchmarks evaluated, including
the 60 ppb benchmark, the Administrator judged that a standard with a level of 70 ppb would
incorporate a margin of safety against the adverse 03-induced effects shown to occur following
exposures (while at moderate or greater exertion) to a somewhat higher concentration. The
Administrator also judged the HREA results for one or more exposures at or above 60 ppb to
provide further support for her somewhat broader conclusion that "a standard with a level of 70
57 Under conditions just meeting an alternative standard with a level of 70 ppb across the 15 urban study areas, the
estimate for two or more days with exposures at or above 70 ppb was 0.4% of children, in the worst year and
worst area (80 FR 65313, Table 1, October 26, 2015).
58 The Administrator was "notably less confident in the adversity to public health of the respiratory effects that have
been observed following exposures to O3 concentrations as low as 60 ppb," based on her consideration of the
ATS recommendation on judging adversity from transient lung function decrements alone, the uncertainty in the
potential for such decrements to increase the risk of other, more serious respiratory effects in a population (per
ATS recommendations on population-level risk), and the less clear CASAC advice regarding potential adversity
of effects at 60 ppb compared to higher concentrations studied (80 FR 65363, October 26, 2015).
3-19
-------
ppb would incorporate an adequate margin of safety against the occurrence of O3 exposures that
can result in effects that are adverse to public health" (80 FR 65364, October 26, 2015).59
While placing limited weight on the lung function risk estimates,60 epidemiologic
evidence61 and quantitative estimates based on information from the epidemiologic studies, the
Administrator additionally considered that information in the context of her consideration of a
standard with a level of 70 ppb. For example, she judged that a standard with a level of 70 ppb
would be expected to result in important reductions in the population-level risk of Cb-induced
lung function decrements in children, including children with asthma (80 FR 65364, October 26,
2015). With regard to the epidemiologic evidence, the Administrator noted that a revised
standard with a level of 70 ppb would provide additional public health protection, beyond that
provided by the then-current standard, against the clearly adverse effects analyzed in
epidemiologic studies (80 FR 65364, October 26, 2015). With regard to the epidemiology-based
risk estimates, the Administrator judged that a revised standard with a level of 70 ppb would
result in meaningful reductions in the mortality and respiratory morbidity risk that is associated
with short- or long-term concentrations of O3 in ambient air (80 FR 65365, October 26, 2015).
In summary, given her consideration of the evidence, exposure and risk information,
advice from the CAS AC, and public comments, the Administrator judged a primary standard of
70 ppb in terms of the 3-year average of fourth-highest daily maximum 8-hour average O3
59 While the Administrator was less concerned about single occurrences of O3 exposures of concern, especially for
the 60 ppb benchmark, she judged that estimates of one or more exposures of concern can provide further insight
into the margin of safety provided by a revised standard. In this regard, she noted that "a standard with a level of
70 ppb is estimated to (1) virtually eliminate all occurrences of exposures of concern at or above 80 ppb; (2)
protect the vast majority of children in urban study areas from experiencing any exposures of concern at or above
70 ppb (i.e., > about 99%, based on mean estimates; Table 1); and (3) to achieve substantial reductions, compared
to the then-current standard, in the occurrence of one or more exposures of concern at or above 60 ppb (i.e., about
a 50% reduction; Table 1)" (80 FR 65364, October 26, 2015).
60 The Administrator noted important uncertainties in using lung function risk estimates as a basis for considering
the occurrence of adverse effects in the population (also recognized in the prior review) that limited her reliance
on these estimates to distinguish between the appropriateness of the health protection afforded by a standard level
of 70 ppb versus lower levels (80 FR 65364, October 26, 2015). These uncertainties related to (1) the ATS
recommendation that "a small, transient loss of lung function, by itself, should not automatically be designated as
adverse" (ATS, 2000); (2) uncertainty in the extent to which a transient population-level decrease in FEVi would
increase the risk of other, more serious respiratory effects in that population (i.e., per ATS recommendations on
population-level risk); and (3) that CAS AC did not advise considering a standard that would be estimated to
eliminate 03-induced lung function decrements >10 or 15% (Frey, 2014c); 80 FR 65364, October 26, 2015).
61 While the Administrator concluded that analyses of air quality in single-city epidemiologic studies support a level
at least as low as 70 ppb, based on a study (Mar and Koenig, 2009) reporting health effect associations in a
location that met the then-current standard over the entire study period but that would have violated a revised
standard with a level of 70 ppb, she further judged that they are of more limited utility for distinguishing between
the appropriateness of the health protection estimated for a standard level of 70 ppb and the protection estimated
for lower levels (80 FR 65364, October 26, 2015).
3-20
-------
concentrations to be requisite to protect public health, including the health of at-risk populations,
with an adequate margin of safety (80 FR 65365, October 26, 2015).
3.2 THE SECONDARY STANDARD
The approach planned for this review of the secondary standard is most fundamentally
based on using the Agency's assessment of the current scientific evidence and associated
quantitative analyses to inform the Administrator's judgments regarding a secondary standard
that is requisite to protect the public welfare from known or anticipated adverse effects. This
approach involves translating scientific and technical information into the basis for addressing a
series of key policy-relevant questions using both evidence- and exposure/risk-based
considerations. This series of key questions related to the secondary standard is presented in
section 3.2.1, along with a summary of the general approach for the review. Additionally, to
provide context for this review of the current secondary standard, section 3.2.2 below
summarizes key aspects of the decisions made in the last review, including the Agency's
consideration of important policy judgments on effects that may be adverse to the public welfare,
as well as uncertainties and limitations in the scientific evidence and in the air quality and
exposure/risk information.
3.2.1 Key Issues Related to the Secondary Standard
The approach planned for this review of the secondary O3 standard will build on the
substantial body of work developed during the course of the last review, taking into account the
more recent scientific information and air quality data now available to inform our understanding
of the key policy-relevant issues in this review. The ISA, risk and exposure analyses (as
warranted), and PA developed in this new review will provide the basis for addressing the key
policy-relevant questions and these documents will inform the Administrator's decisions as to
whether to retain or revise this standard. As summarized in section 1.2, and also described in
chapter 6, evaluations in the PA are intended to inform the Administrator's public welfare policy
judgments and decisions. In so doing, the PA considers the potential implications of various
aspects of the scientific evidence, the exposure/risk-based information, and the associated
uncertainties and limitations.
In building upon the conclusions from the last review, the current review of the
secondary standard, as with the review of the primary standard, takes into account the updated
evidence and information that has become available since the last review. The Agency's
consideration of the full set of evidence and information available in this review will inform the
answer to the following initial overarching question for the review:
3-21
-------
• Do the currently available scientific evidence and exposure-/risk-based information
support or call into question the adequacy of the public welfare protection afforded by
the current secondary O3 standard?
In reflecting on this question, we will consider the available body of scientific evidence,
assessed in the ISA, and considered as a basis for developing or interpreting risk and exposure
analyses, including whether it supports or calls into question the scientific conclusions reached in
the last review regarding welfare effects related to exposure to O3 in ambient air. Information
available in this review that may be informative to public policy judgments regarding
significance or adversity of key effects on the public welfare will also be considered.
Additionally, the currently available exposure and risk information, whether newly developed in
this review or predominantly developed in the past and interpreted in light of current
information, will be considered, including with regard to the extent to which it may continue to
support judgments made in the last review. Further, in considering this question with regard to
the secondary O3 standard, we give particular attention to exposures and risks for effects with the
greatest potential for public welfare significance.
Evaluation of the available scientific evidence and risk/exposure information with regard
to consideration of the current standard will focus on key policy-relevant issues by addressing a
series of questions including the following:
• Is there newly available evidence that indicates the importance of photochemical oxidants
other than O3 with regard to abundance in ambient air, and potential for welfare effects?
• Does the current evidence alter our conclusions from the last review regarding the nature of
welfare effects attributable to O3 in ambient air? Is there new evidence on welfare effects
beyond those identified in the last review?
• What information is newly available in this review relevant to consideration of public
welfare implications? Does it alter our understanding of locations or ecosystems where the
presence of species sensitive to Cb-related effects indicates the potential for effects on the
public welfare?
• Does the current evidence continue to support a cumulative, seasonal exposure index as a
biologically-relevant and appropriate metric for assessment of the evidence or exposure/risk
information for vegetation? Does the W126 index continue to be supported for this purpose?
• To what extent does the available evidence indicate the occurrence of 03-related effects
attributable to cumulative O3 exposures lower than previously established or that might be
expected to occur under the current standard?
• Is there new evidence on factors that influence relationships between O3 concentrations and
vegetation-related or other welfare effects?
• What are important uncertainties in the evidence? To what extent have important
uncertainties in the evidence identified in the last review been reduced and/or have new
uncertainties been recognized?
3-22
-------
• What are the nature and magnitude of exposure- and risk-related estimates for vegetation
associated with conditions just meeting the current standard, and what do they indicate
regarding the potential for Cb-related vegetation impacts?
• Are such exposures and risks reasonably judged important from a public welfare perspective?
• What are the important uncertainties associated with any exposure estimates and associated
characterization of potential for public welfare effects?
If the information available in this review suggests that revision of the current secondary
standard would be appropriate to consider, the PA will include evaluation of how the standard
might be revised, based on the currently available scientific information, air quality assessments
and exposure/risk information, and also considering what the available information indicates as
to public welfare protection expected to be afforded by the current or potential alternative
standards. In such an evaluation, the PA may consider the effect of revision of one or more
elements of the standard (indicator, averaging time, level and form), with the effect being
evaluated based on the resulting potential standard and all of its elements collectively. Based on
such evaluations, the PA would then identify potential alternative standards (in terms of
indicator, averaging time, level, and form) that would reflect a range of alternative policy
judgments as to the degree of protection that is requisite to protect public welfare from known or
anticipated adverse effects, and options for standards expected to achieve it. The specific policy-
relevant questions that frame such evaluation of what revision of the standard might be
appropriate to consider include:
• Does the currently available information call into question the identification of ozone as the
indicator for photochemical oxidants? Is support provided for considering a different
indicator?
• To what extent does the currently available information call into question the current
averaging time? Is support provided for considering different averaging times for the
standard?
• What does the currently available information indicate with regard to a range of levels and
forms of alternative standards that may be supported and what are the uncertainties and
limitations in that information?
• What do the available analyses indicate with regard to exposure and risk associated with
specific alternative standards? What are the associated uncertainties? To what extent might
such alternatives be expected to reduce adverse impacts attributable to O3, and what are the
uncertainties in the estimated reductions?
3-23
-------
The approach to reaching conclusions on the current secondary O3 standard and, as
appropriate, on potential alternative standards, including consideration of the policy-relevant
questions which will frame the current review, is illustrated in Figure 3-2.
Adequacy of Current Standard
Exposure and Risk-Based
Considerations
r Nature, magnitude, and importance of
estimated exposures and risks associaed
wth meeang the current stendard?
Uncertainties in lie exposure and risk
esSmases?
-"Does currency available evidence and related
uncertainties strengthen orcal infoquesSon prior
conclusions?
¦ Evidence of welfare effecs not previously identified?
¦ Evidence defects at iower levels or for different
exposure circumstances? Evidence for vegetaSon
elfeois from cumulate exposures atowed by the
current standard?
• Uncertainties tdenafied in the last review reduced or
new uncertainties emerged?
Evidence-Based Considersrions
Does
information
call into question
adequacy of current
standard?
' Consider retaining
current standard
1
YES
Consider Potential Alternative Standards
'
Elements of Potential Alternative Standards
> Indicator, Averaging Time, Form, Leve)
Potential Alternative Standards for Consideration
Figure 3-2. Overview of general approach for review of the secondary O3 standard.
3-24
-------
The Agency's approach in review secondary standards is consistent with the
requirements of the provisions of the CAA related to the review of NAAQS and with how the
EPA and the courts have historically interpreted the CAA. As discussed in section 1.1 above,
these provisions require the Administrator to establish secondary standards that, in the
Administrator's judgment, are requisite (i.e., neither more nor less stringent than necessary) to
protect the public welfare from known or anticipated adverse effects. The CAA does not require
that standards be set at a zero-risk level, but rather at a level that reduces risk sufficiently so as to
protect the public welfare from known or anticipated adverse effects. The Agency's decisions on
the adequacy of the current secondary standard and, as appropriate, on any potential alternative
standards considered in a review, are largely public welfare policy judgments made by the
Administrator. The four basic elements of the NAAQS (i.e., indicator, averaging time, form, and
level) will be considered collectively in evaluating the protection afforded by the current
standard, or any alternative standards considered. The Administrator's final decisions in a review
draw upon the scientific information and analyses about welfare effects, environmental
exposures and risks, and associated public welfare significance, as well as judgments about how
to consider the range and magnitude of uncertainties that are inherent in the scientific evidence
and analyses.
3.2.2 Background on the Current Secondary Standard (Considerations and Conclusions
in the Last Review)
The 2015 decision to revise the secondary O3 standard was based on the scientific and
technical information available at that time, as well as the Administrator's judgments regarding
the available welfare effects evidence, the appropriate degree of public welfare protection for the
revised standard, and available air quality information on seasonal cumulative exposures that
may be allowed by such a standard (80 FR 65292, October 26, 2015). With the 2015 decision,
the Administrator revised the level of the secondary standard from 0.075 to 0.070 ppm, in
conjunction with retaining the then-current indicator, averaging time (8 hours) and form (fourth-
highest daily maximum 8-hour average concentration, averaged across three years).
The welfare effects evidence base available in the 2015 review includes more than fifty
years of extensive research on Cte's phytotoxic effects, conducted both in and outside of the U.S.
that documents the impacts of O3 on plants and their associated ecosystems (U.S. EPA, 1978,
1986, 1996, 2006, 2013). As was established in prior reviews, O3 can interfere with carbon gain
(photosynthesis) and allocation of carbon within the plant, making fewer carbohydrates available
for plant growth, reproduction, and/or yield. For seed-bearing plants, these reproductive effects
will culminate in reduced seed production or yield (U.S. EPA, 1996, pp. 5-28 and 5-29). The
strongest evidence for effects from O3 exposure on vegetation is from controlled exposure
3-25
-------
studies, which "have clearly shown that exposure to O3 is causally linked to visible foliar injury,
decreased photosynthesis, changes in reproduction, and decreased growth" in many species of
vegetation (2013 ISA, p. 1-15). Such effects at the plant scale can also be linked to an array of
effects at larger spatial scales, with the evidence available in the last review indicating that
"ambient O3 exposures can affect ecosystem productivity, crop yield, water cycling, and
ecosystem community composition" (2013 ISA, p. 1-15, Chapter 9, section 9.4).
In light of this robust evidence base, the 2013 ISA concluded there to be causal
relationships between O3 and visible foliar injury, reduced vegetation growth, reduced
productivity in terrestrial ecosystems, reduced yield and quality of agricultural crops and
alteration of below-ground biogeochemical cycles. The 2013 ISA additionally found there to
likely be a causal relationship between O3 and reduced carbon sequestration in terrestrial
ecosystems, alteration of terrestrial ecosystem water cycling and alteration of terrestrial
community composition (2013 ISA, Table 9-19). Further, based on the then-available evidence
with regard to O3 effects on climate, the 2013 ISA also found there to be a causal relationship
between changes in tropospheric O3 concentrations and radiative forcing, found there likely to be
a causal relationship between tropospheric O3 concentrations and effects on climate as quantified
through surface temperature response, and found the evidence to be inadequate to determine if a
causal relationship exists between tropospheric O3 concentrations and health and welfare effects
related toUV-B shielding (2013 ISA, section 10.5).
The 2015 decision was a public welfare policy judgment made by the Administrator,
which drew upon the available scientific evidence for 03-attributable welfare effects and on
analyses of exposures and public welfare risks based on impacts to vegetation, ecosystems and
their associated services, as well as judgments about the appropriate weight to place on the range
of uncertainties inherent in the evidence and analyses. Such judgments in the context of that
review included judgments on the weight to place on the evidence of specific vegetation-related
effects estimated to result across a range of cumulative seasonal concentration-weighted O3
exposures; on the weight to give associated uncertainties, including those related to the
variability in occurrence of such effects in areas of the U.S., especially areas of particular public
welfare significance; and on the extent to which such effects in such areas may be considered
adverse to public welfare.
The decision was based on a thorough review, in the 2013 ISA, of the scientific
information on 03-induced environmental effects. The decision also took into account: (1) staff
assessments in the 2014 PA of the most policy-relevant information in the 2013 ISA regarding
evidence of adverse effects of O3 to vegetation and ecosystems, information on biologically-
relevant exposure metrics, 2014 welfare REA (WREA) analyses of air quality, exposure, and
ecological risks and associated ecosystem services, and staff analyses of relationships between
3-26
-------
levels of a W126-based exposure index62 and potential alternative standard levels in combination
with the form and averaging time of the then-current standard; (2) additional air quality analyses
of the W126 index and design values based on the form and averaging time of the then-current
standard (3) CASAC advice and recommendations; and (4) public comments received during the
development of these documents and on the proposal notice. In addition to reviewing the most
recent scientific information as required by the CAA, the 2015 rulemaking also incorporated the
EPA's response to the judicial remand of the 2008 secondary O3 standard in Mississippi v. EPA,
744 F.3d 1334 (D.C. Cir. 2013) and, in accordance with the court's decision in that case, fully
explained the Administrator's conclusions as to the level of air quality that provides the requisite
protection of public welfare from known or anticipated adverse effects.
Consistent with the general approach routinely employed in NAAQS reviews, the initial
consideration in the last review of the secondary standard was with regard to the adequacy of
protection provided by the then-existing standard. Key aspects of that consideration are
summarized in section 3.2.2.1 below. The subsequent selection of a standard concluded by the
Administrator to provide the requisite protection under the Act is summarized in section 3.2.2.2.
3.2.2.1 Considering the Need for Revision
The approach to considering the adequacy of the secondary O3 standard in the 2015
review involved the careful consideration of the available evidence, analyses and conclusions
contained in the 2013 ISA, including information newly available in the review; the information,
quantitative assessments, considerations and conclusions presented in the 2014 WREA and 2014
PA; additionally available air quality analyses; the advice and recommendations from the
CASAC; and public comments. The Administrator gave primary consideration to the evidence of
growth effects in well-studied tree species and information on cumulative seasonal O3 exposures
occurring in Class I areas63 when the then-current standard was met (80 FR 65385-65386,
October 26, 2015). The exposure information for Class I areas evaluated in terms of the W126
cumulative seasonal exposure index, an index recognized by the 2013 ISA as a mathematical
approach "for summarizing ambient air quality information in [a] biologically meaningful form[]
for O3 vegetation effects assessment purposes" (2013 ISA, section 9.5.3). The EPA focused on
62 The W126 index is a cumulative seasonal metric described as the sigmoidally weighted sum of all hourly 03
concentrations observed during a specified daily and seasonal time window, where each hourly O3 concentration
is given a weight that increases from zero to one with increasing concentration (80 FR 65373-74, October 26,
2015). Accordingly, W126 index values are in the units of ppm-hours (ppm-hrs).
63 Areas designated as Class I include all international parks, national wilderness areas which exceed 5,000 acres in
size, national memorial parks which exceed 5,000 acres in size, and national parks which exceed six thousand
acres in size, provided the park or wilderness area was in existence on August 7, 1977. Other areas may also be
Class I if designated as Class I consistent with the CAA.
3-27
-------
the W126 index for this purpose consistent with the evidence in the 2013 ISA and advice from
the CASAC (80 FR 65375, October 26, 2015).
In her decision making, the Administrator considered the effects of O3 on tree seedling
growth, as suggested by the CASAC, as a surrogate or proxy for the full array of vegetation-
related effects of O3, ranging from effects on sensitive species to broader ecosystem-level effects
(80 FR 65369, 65406, October 26, 2015). The metric used for quantifying effects on tree
seedling growth in the review was relative biomass loss (RBL), with the evidence base providing
robust and established exposure-response (E-R) functions for seedlings of 11 tree species (80 FR
65391-92, October 26, 2015; 2014 PA, Appendix 5C).64 The Administrator used this proxy in
making her judgments on O3 effects to the public welfare.
In considering the public welfare protection provided by the then-current standard, the
Administrator gave primary consideration to an analysis of cumulative seasonal exposures in or
near Class I areas during periods when the then-current standard was met and the associated
estimates of growth effects, in terms of the O3 attributable reductions in RBL in the median
species for which exposure-response (E-R) functions have been established (80 FR 65389-
65390, October 26, 2015). 65 The Administrator noted the occurrence of exposures for which the
associated estimates of growth effects in the median species extend above a magnitude
considered to be "unacceptably high" by CASAC.66 This analysis estimated such cumulative
exposures occurring under the then-current standard for nearly a dozen areas, distributed across
two NOAA climatic regions of the U.S (80 FR 65385-86, October 26, 2015). The Administrator
gave particular weight to this analysis because of its focus in Class I areas, lands that Congress
set aside for specific uses intended to provide benefits to the public welfare, including lands that
are to be protected so as to conserve the scenic value and the natural vegetation and wildlife
within such areas, and to leave them unimpaired for the enjoyment of future generations. Such an
emphasis on lands afforded special government protections, such as national parks and forests,
64 These functions for RBL estimate the reduction in a year's growth as a percentage of that expected in the absence
of O3 (2013 ISA, section 9.6.2; 2014 WREA, section 6.2).
65 In specifically evaluating exposure levels in terms of the W126 index as to potential for impacts on vegetation, the
Administrator focused on RBL estimates for the median across the eleven tree species for which robust E-R
functions were available. The presentation of robust established E-R functions for growth effects on tree seedlings
(and crops) included estimates of RBL (and RYL) at a range of W126-based exposure levels (2014 PA, Tables
5C-1 and 5C-2). The median tree species RBL or crop RYL was presented for each W126 level (2014 PA, Table
5C-3; 80 FR 65391 [Table 4], October 26, 2015). The Administrator focused on RBL as a surrogate or proxy for
the broader array of vegetation-related effects of potential public welfare significance, which include effects on
growth of individual sensitive species and extend to ecosystem-level effects, such as community composition in
natural forests, particularly in protected public lands, as well as forest productivity (80 FR 65406, October 26,
2015).
66 In the CASAC's consideration of RBL estimates presented in the draft PA, it characterized an estimate of 6%
RBL in the median studied species as "unacceptably high" (Frey, 2014c).
3-28
-------
wildlife refuges, and wilderness areas, some of which are designated Class I areas under the
CAA, was consistent with a similar emphasis in the 2008 review of the standard (73 FR 16485,
March 27, 2008). The Administrator additionally recognized that states, tribes and public interest
groups also set aside areas that are intended to provide similar benefits to the public welfare for
residents on those lands, as well as for visitors to those areas (80 FR 65390, October 26, 2015).
As noted across reviews of O3 secondary standards, the Administrator's judgments
regarding effects that are adverse to public welfare consider the intended use of the ecological
receptors, resources and ecosystems affected (80 FR 65389, October 26, 2015). Thus, in the
2015 review, the Administrator utilized the median RBL estimate for the studied species as a
quantitative tool within a larger framework of considerations pertaining to the public welfare
significance of O3 effects. She recognized such considerations to include effects that are
associated with effects on growth and that the 2013 ISA determined to be causally or likely
causally related to O3 in ambient air, yet for which there are greater uncertainties affecting our
estimates of impacts on public welfare. These other effects included reduced productivity in
terrestrial ecosystems, reduced carbon sequestration in terrestrial ecosystems, alteration of
terrestrial community composition, alteration of below-grown biogeochemical cycles, and
alteration of terrestrial ecosystem water cycles. Thus, in giving attention to the CASAC's
characterization of a 6% estimate for tree seedling RBL in the median studied species as
"unacceptably high", the Administrator, while mindful of uncertainties with regard to the
magnitude of growth impact that might be expected in mature trees, was also mindful of related,
broader, ecosystem-level effects for which the available tools for quantitative estimates are more
uncertain and those for which the policy foundation for consideration of public welfare impacts
is less well established. As a result, the Administrator considered tree growth effects of O3, in
terms of RBL as a surrogate for the broader array of O3 effects at the plant and ecosystem levels
(80 FR 65389, October 26, 2015).
Based on all of these considerations, and taking into consideration CASAC advice, the
Administrator concluded that the protection afforded by the then-current standard was not
sufficient and that the standard needed to be revised to provide additional protection from known
and anticipated adverse effects to public welfare, related to effects on sensitive vegetation and
ecosystems, most particularly those occurring in Class I areas, and also in other areas set aside by
states, tribes and public interest groups to provide similar benefits to the public welfare for
residents on those lands, as well as for visitors to those areas. In so doing, she further noted that a
revised standard would provide increased protection for other growth-related effects, including
for crop yield loss, reduced carbon storage and for areas for which it is more difficult to
determine public welfare significance, as well as for other welfare effects of O3, such as visible
foliar injury (80 FR 65390, October 26, 2015).
3-29
-------
3.2.2.2 Considering Revisions to the Standard
Consistent with the approach employed for considering the adequacy of the then-current
secondary standard, the approach for considering revisions that would result in a standard
providing the requisite protection under the Act also focused on growth-related effects of O3,
using RBL as a surrogate for the broad array of vegetation-related effects and included
judgments on the magnitude of such effects that would contribute to public welfare impacts of
concern. In considering the adequacy of potential alternative standards to provide protection
from such effects, the approach also focused on considering the cumulative seasonal O3
exposures likely to occur with different alternative standards.
In light of the judicial remand of the 2008 secondary O3 standard referenced above, the
2015 decision on selection of a revised secondary standard first considered the available
evidence and quantitative analyses in the context of an approach for considering and identifying
public welfare objectives for such a standard (80 FR 65403-65408, October 26, 2015). The
robust and longstanding evidence of O3 effects on vegetation and associated terrestrial
ecosystems, including evidence newly available in the 2015 review, provided the foundation for
the Administrator's consideration of O3 effects, associated public welfare protection objectives,
and the revisions to the standard needed to achieve those objectives. In light of the extensive
evidence base in this regard, the Administrator focused on protection against adverse public
welfare effects of O3 related effects on vegetation. In so doing, she took note of effects that
compromise plant function and productivity, with associated effects on ecosystems. She had
particular concern about such effects in natural ecosystems, such as those in areas with
protection designated by Congress for current and future generations, as well as areas similarly
set aside by states, tribes and public interest groups with the intention of providing similar
benefits to the public welfare. The Administrator additionally recognized that providing
protection for this purpose will also provide a level of protection for other vegetation that is used
by the public and potentially affected by O3 including timber, produce grown for consumption
and horticultural plants used for landscaping (80 FR 65403, October 26, 2015).
As an initial matter, the Administrator considered the use of a cumulative seasonal
exposure index for purposes of assessing potential public welfare risks, and similarly, for
assessing potential protection achieved against such risks on a national scale. In consideration of
conclusions of the 2013 ISA and 2014 PA, as well as advice from the CAS AC and public
comments, the focus was on a W126 index described as a maximum 3-month, 12-hour index,
defined by the 3-consecutive-month period within the O3 season with the maximum sum of
W126-weighted hourly O3 concentrations during the period from 8:00 a.m. to 8:00 p.m. each day
(80 FR 65404, October 26, 2015). While recognizing that no one definition of an exposure
metric used for the assessment of protection for multiple effects at a national scale will be
3-30
-------
exactly tailored to every species or each vegetation type, ecosystem and region of the country,
the Administrator judged that on balance, a W126 index derived in this way, and averaged over
three years would be appropriate for such purposes. Thus, in considering revisions to the
secondary standard that would specify a level of air quality to provide the necessary public
welfare protection, the Administrator focused on use of a cumulative seasonal concentration-
weighted exposure index, including specifically the W126 index, for assessing exposure, both for
making judgments with regard to the potential harm to public welfare posed by conditions
allowed by various levels of air quality and for making the associated judgments regarding the
appropriate degree of protection against such potential harm (80 FR 65403, October 26, 2015).
Based on a number of considerations, the Administrator recognized greater confidence in
judgments related to public welfare impacts based on a 3-year average metric than a single year
metric, and consequently concluded it to be appropriate to use an index averaged across three
years forjudging public welfare protection afforded by a revised secondary standard (80 FR
65404, October 26, 2015). For example, while recognizing that the scientific evidence
documents the effects on vegetation resulting from individual growing season exposures of
specific magnitude, including those that can affect the vegetation in subsequent years, the
Administrator was also mindful of both the strengths and limitations of the evidence and of the
information on which to base her judgments with regard to adversity of effects on the public
welfare. In this regard, she recognized uncertainties associated with interpretation of the public
welfare significance of effects resulting from a single-year exposure, and that the public welfare
significance of effects associated with multiple years of critical exposures are potentially greater
than those associated with a single year of such exposure. While recognizing the potential for
effects on vegetation associated with a single-year exposure, the Administrator concluded that
use of a 3-year average metric can address the potential for adverse effects to public welfare that
may relate to shorter exposure periods, including a single year (80 FR 65404, October 26,
2015).67
In reaching a conclusion on the amount of public welfare protection from the presence of
O3 in ambient air that is appropriate to be afforded by a revised secondary standard, the
Administrator gave particular consideration to the following: (1) the nature and degree of effects
67 While the Administrator recognized the scientific information and interpretations, as well as CASAC advice, with
regard to a single-year exposure index, she also took note of uncertainties associated with judging the degree of
vegetation impacts for annual effects that would be adverse to public welfare. It was noted that even in the case of
annual crops, the assessment of public welfare significance is unclear due to the role of crop management and
related agricultural practices. The Administrator was also mindful of the variability in ambient air O3
concentrations from year to year, as well as year-to-year variability in environmental factors, including rainfall
and other meteorological factors, that influence the occurrence and magnitude of Ch-related effects in any year,
and contribute uncertainties to interpretation of the potential for harm to public welfare over the longer term (80
FR 65404, October 26, 2015).
3-31
-------
of O3 on vegetation, including her judgments as to what constitutes an adverse effect to the
public welfare; (2) the strengths and limitations of the available and relevant information; (3)
comments from the public on the Administrator's proposed decision, including comments related
to identification of a target level of protection; and (4) CASAC's views regarding the strength of
the evidence and its adequacy to inform judgments on public welfare protection. The
Administrator recognized that such judgments include judgments about the interpretation of the
evidence and other information, such as the quantitative analyses of air quality monitoring,
exposure and risk. She also recognized that such judgments should neither overstate nor
understate the strengths and limitations of the evidence and information nor the appropriate
inferences to be drawn as to risks to public welfare. It was also noted that the CAA does not
require that a secondary standard be protective of all effects associated with a pollutant in the
ambient air but rather those known or anticipated effects judged adverse to the public welfare.
She additionally recognized that the choice of the appropriate level of protection is a public
welfare policy judgment entrusted to the Administrator under the CAA taking into account both
the available evidence and the uncertainties (80 FR 65404-05, October 26, 2015).
With regard to the extensive evidence of welfare effects of O3, including the established
evidence base regarding O3 and visible foliar injury, in addition to the long-standing evidence
base on 03-attributable crop yield loss, the information available for forest tree species was
judged to be more useful in informing judgments regarding the nature and severity of effects
associated with different air quality conditions and associated public welfare significance.
Accordingly, the Administrator gave particular attention to the effects related to native tree
growth and productivity, recognizing their relationship to a range of ecosystem services,
including forest and forest community composition (80 FR 65405-06, October 26, 2015).
In so doing, the Administrator recognized that the robust evidence base documented a
broad array of 03-induced vegetation effects, among which were the occurrence of visible foliar
injury and growth and/or yield loss in 03-sensitive annual and perennial species, including crops
and other commercial species, such as timber, horticultural and landscaping plants, as well as
native species in unmanaged natural areas (80 FR 65405, October 26, 2015). In regard to visible
foliar injury, as stated in the 2013 ISA, "[experimental evidence has clearly established a
consistent association of visible injury with O3 exposure, with greater exposure often resulting in
greater and more prevalent injury" (2013 ISA, p. 9-41). The Administrator recognized the
potential for this effect to affect the public welfare in the context of affecting values pertaining to
natural forests, particularly those afforded special government protection, with the significance
of 03-induced visible foliar injury depending on the extent and severity of the injury (80 FR
65407, October 26, 2015). In so doing, however, the Administrator also took note of limitations
in the available visible foliar injury information, including the lack of robust E-R functions that
3-32
-------
would allow prediction of visible foliar injury severity and incidence under varying air quality
and environmental conditions, a lack of clear quantitative relationships linking visible foliar
injury with other Cb-induced vegetation effects, such as growth or related ecosystem effects, and
a lack of established criteria or objectives that might inform consideration of potential public
welfare impacts related to this vegetation effect (80 FR 65407, October 26, 2015). Similarly,
while 03-related growth effects on agricultural and commodity crops had been extensively
studied and robust E-R functions developed for a number of species, the Administrator found
this information less useful in informing her judgments regarding an appropriate level of public
welfare protection (80 FR 65405, October 26, 2015).68
Thus, and in light of the extensive evidence base in this regard, the Administrator focused
on trees and associated ecosystems in identifying the appropriate level of protection for the
secondary standard. Accordingly, the Administrator found the estimates of tree seedling growth
impacts (in terms of RBL) associated with a range of W126-based index values developed from
the robust E-R functions for 11 tree species to be appropriate and useful for considering the
appropriate public welfare protection objective for a revised standard (80 FR 65391-92, Table 4,
October 26, 2015). The Administrator also incorporated into her considerations the broader
evidence base associated with forest tree seedling biomass loss, including other less quantifiable
effects of potentially greater public welfare significance. That is, in drawing on these RBL
estimates, the Administrator recognized she was not simply making judgments about a specific
magnitude of growth effect in seedlings that would be acceptable or unacceptable in the natural
environment. Rather, though mindful of associated uncertainties, the Administrator used the
RBL estimates as a surrogate or proxy for consideration of the broader array of related
vegetation and ecosystem effects of potential public welfare significance that include effects on
growth of individual sensitive species and extend to ecosystem-level effects, such as community
composition in natural forests, particularly in protected public lands, as well as forest
productivity (80 FR 65406, October 26, 2015).
Thus, the Administrator used the RBL estimates as a proxy for the array of vegetation-
related effects, including those for which public welfare implications are more significant but for
which the tools for quantitative estimates were more uncertain. In so doing, the Administrator
68 With respect to commercial production of commodities, the Administrator noted that judgments about the extent
to which 03-related effects on commercially managed vegetation are adverse from a public welfare perspective
are particularly difficult to reach, given that the extensive management of such vegetation (which, as the CAS AC
noted, may reduce yield variability) may also to some degree mitigate potential 03-related effects. The
management practices used on these lands are highly variable and are designed to achieve optimal yields, taking
into consideration various environmental conditions. In addition, changes in yield of commercial crops and
commercial commodities, such as timber, may affect producers and consumers differently, further complicating
the question of assessing overall public welfare impacts (80 FR 65405, October 26, 2015).
3-33
-------
recognized that the CASAC gave weight to these relationships in formulating its advice and she
took particular note of the characterization by the CASAC of the 6% RBL level in the median
studied species as "unacceptably high," as this comment was provided in the context of the
CASAC's consideration of the significance of effects associated with a range of alternatives for
the secondary standard (Frey, 2014c, pp. iii, 13, 14; 80 FR 65406, October 26, 2015). Moreover,
the range recommended by the CASAC excluded W126 index values for which the median
species was estimated to have a 6% RBL in the draft PA (which was the context for the CASAC
advice) (Frey, 2014c, p. 12-13; 80 FR 65406, October 26, 2015). In consideration of CASAC
advice; strengths, limitations and uncertainties in the evidence; and the linkages of growth effects
to larger population, community and ecosystem impacts, the Administrator considered it
appropriate to focus on a standard that would generally limit cumulative exposures to those for
which the median RBL estimate would be somewhat below 6% (80 FR 65406-07, October 26,
2015).
In focusing on cumulative exposures associated with a median RBL estimate somewhat
below 6%, the Administrator considered the relationships between W126-based exposure and
RBL in the studied species (presented in the final PA and proposal notice), noting that the
median RBL estimate was 6% for a cumulative seasonal W126 exposure index of 19 part per
million-hours (ppm-hrs) (80 FR 65391-92, Table 4, October 26, 2015).69 Given the information
on median RBL at different W126 exposure levels, using a 3-year cumulative exposure index for
assessing vegetation effects, the potential for single-season effects of concern, and CASAC
comments on the appropriateness of a lower value for a 3-year average W126 index, the
Administrator concluded it was appropriate to identify a standard that would restrict cumulative
seasonal exposures to 17 ppm-hrs or lower, in terms of a 3-year W126 index, in nearly all
instances (80 FR 65407, October 26, 2015). Based on such then-current information to inform
consideration of vegetation effects and their potential adversity to public welfare, the
Administrator additionally judged that the RBL estimates associated with marginally higher
exposures in isolated, rare instances are not indicative of effects that would be adverse to the
public welfare, particularly in light of variability in the array of environmental factors that can
influence O3 effects in different systems and uncertainties associated with estimates of effects
associated with this magnitude of cumulative exposure in the natural environment (80 FR 65407,
October 26, 2015).
The Administrator's decisions regarding the revisions to the then-current standard that
would appropriately achieve these public welfare protection objectives were based on extensive
69 The median RBL estimate was 5.7% (which rounds to 6%) for a cumulative seasonal W126 exposure index of 18
ppm-hrs and the median RBL estimate was 5.3% (which rounds to 5%) for 17 ppm-hrs (80 FR 65407, October
26, 2015).
3-34
-------
air quality analyses that extended from the then most recently available data (monitoring year
2013) back more than a decade (80 FR 65408, October 26, 2015; Wells, 2015). These analyses
evaluated the cumulative seasonal exposure levels in locations meeting different alternative
levels for a standard of the then-current form and averaging time, indicating reductions in
cumulative exposures associated with air quality meeting lower levels of a standard of the
existing form and averaging time. Based on these analyses, the Administrator judged that the
desired level of public welfare protection could be achieved with a secondary standard having a
revised level in combination with the existing form and averaging time (80 FR 65408, October
26, 2015).
The air quality analyses described the occurrences of 3-year W126 index values of
various magnitudes at monitor locations where O3 concentrations met potential alternative
standards defined by different levels combined with the current form and averaging time (Wells,
2015). In the then-most recent period, 2011-2013, across the monitor locations meeting the then-
current standard (with a level of 75 ppb), the 3-year W126 index values were above 17 ppm-hrs
in 25 sites distributed across different NOAA climatic regions, and above 19 ppm-hr at nearly
half of these sites, with some well above. In comparison, among sites meeting an alternative
standard of 70 ppb, there were no occurrences of a W126 value above 17 ppm-hrs and fewer than
a handful of occurrences that equaled 17 ppm-hrs.70 For the longer time period (extending back
to 2001), among the nearly 4000 locations meeting a standard level of 70 ppb, there was only a
handful of isolated occurrences of 3-year W126 index values above 17 ppm-hrs, all but one of
which were below 19 ppm-hrs.71 The Administrator concluded that that single higher value of
19.1 ppm-hrs, observed at a monitor for the 3-year period of 2006-2008, was reasonably
regarded as an extremely rare and isolated occurrence, and, as such, it was unclear whether it
would recur, particularly as areas across U.S. took further steps to reduce O3 to meet revised
primary and secondary standards. Further, based on all of the then available information, as
noted above, the Administrator did not judge RBL estimates associated with marginally higher
exposures in isolated, rare instances to be indicative of adverse effects to the public welfare. The
Administrator concluded that a standard with a level of 70 ppb and the current form and
averaging time may be expected to limit cumulative exposures, in terms of a 3-year average
W126 exposure index, to values at or below 17 ppm-hrs, in nearly all instances, and accordingly,
70 The more than 500 monitors that would meet an alternative standard of 70 ppb during the 2011-2013 period were
distributed across all nine NOAA climatic regions and 46 of the 50 states (Wells, 2015 and associated dataset in
the docket [document identifier, EPA-HQ-OAR-2008-0699-4325]).
71 Among sites meeting a level of 65 ppb, there were no occurrences above 11 ppm-hrs, well below the objectives
identified for affording public welfare protection. For this level, the appreciably smaller and less geographically
extensive database contributes uncertainty to conclusions based on such analysis (80 FR 65409, October 26,
2015).
3-35
-------
to eliminate or virtually eliminate cumulative exposures associated with a median RBL of 6% or
greater (80 FR 65409, October 26, 2015). Thus, using RBL as a proxy in judging effects to
public welfare, the Administrator judged that a standard with a level of 70 ppb would provide the
requisite protection from adverse effects to public welfare by limiting cumulative seasonal
exposures to 17 ppm-hrs or lower (in terms of a 3-year W126 index) in nearly all instances.
In summary, the Administrator judged that the revised standard would protect natural
forests in Class I and other similarly protected areas against an array of adverse vegetation
effects, most notably including those related to effects on growth and productivity in sensitive
tree species. The Administrator additionally judged that a revised standard set at a level of 70
ppb, in combination with the then-existing form and averaging time, would be sufficient to
protect public welfare from known or anticipated adverse effects. This judgment by the
Administrator appropriately recognized that the CAA does not require that standards be set at a
zero-risk level, but rather at a level that reduces risk sufficiently so as to protect the public
welfare from known or anticipated adverse effects. Thus, based on the conclusions drawn from
the air quality analyses which demonstrated a strong, positive relationship between the 8-hour
and W126 metrics and the findings that indicated the significant amount of control provided by
the fourth-high metric, the evidence base of O3 effects on vegetation and her public welfare
policy judgments, as well as public comments and CASAC advice, the Administrator decided to
retain the existing form and averaging time and revise the level to 0.070 ppm, judging that such a
standard would provide the requisite protection to the public welfare from any known or
anticipated adverse effects associated with the presence of O3 in ambient air (80 FR 65409-10,
October 26, 2015).
3-36
-------
4 SCIENCE ASSESSMENT
Integrated Science Assessments serve as the scientific foundation of the NAAQS review
process and are developed by the EPA's NCEA. This assessment focuses on reviewing and
updating the air quality criteria associated with primary (health-based) and secondary (welfare-
based72) effects evidence to inform science policy judgments about the primary and secondary
standards for O3 and other photochemical oxidants. This chapter provides an overview of the ISA
development process and discusses key aspects of the EPA's planned approach for the ISA in
this review.
4.1 PURPOSE OF THE ISA
The purpose of the ISA is to draw upon the existing body of evidence to synthesize and
provide a critical evaluation of the current state of scientific knowledge on the most relevant
issues pertinent to the review of the NAAQS for O3 and other photochemical oxidants, to
identify changes in the scientific evidence bases since the previous review, and to describe
remaining or newly identified uncertainties. The ISA will identify, critically evaluate and
synthesize the most policy-relevant current scientific literature (e.g., epidemiology, controlled
human exposure, animal toxicology, atmospheric science, exposure science, ecology and
climate-related science), including key science judgments that are important to inform the
development of risk and exposure analyses (as warranted) and the PA, as well as other aspects of
the NAAQS review process (summarized in section 1.2 above). The ISA will provide a focused
assessment of the scientific evidence to address specific scientific questions (section 4.4) and
inform the overall policy-relevant questions for the PA (as described in Chapter 3).
4.2 ORGANIZATION OF THE ISA
The general organization of the ISA for the current review will be consistent with the 2nd
External Review Draft ISA for Oxides of Nitrogen, Oxides of Sulfur, and Particulate Matter-
Ecological Criteria (U.S. EPA, 2018d). Accordingly, the ISA will begin with a Preface
discussing major legal and historical aspects of prior O3 NAAQS reviews. An executive
summary targeted to a wide range of audiences will succinctly summarize the conclusions of the
ISA. An integrated synthesis will serve as the main body of the ISA and provide a detailed
summary of the key information for each topic area, including background concentrations of O3
72 Under Clean Air Act, section 302(h) (42 U.S.C. § 7602(h)), effects on welfare include, but are not limited to,
"effects on soils, water, crops, vegetation, manmade 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."
4-1
-------
in the U.S., conclusions regarding the nature of health and welfare effects associated with O3
exposure (including causality determinations for relationships between exposure to O3 and
specific types of health and welfare effects), and identification of the human lifestages and
populations at increased risk of the effects of O3. The integrated synthesis will discuss additional
policy-relevant issues, such as the exposure durations, metrics, and concentrations eliciting
health and welfare effects; the concentration-response relationships for specific effects, including
the overall shape and whether or not there is evidence of a discernible threshold below which
effects are not likely to occur; and the public health and welfare impact of effects associated with
exposure to O3. The synthesis will also discuss important issues for different types of studies,
such as the air quality metrics and the lag structure of epidemiologic associations with health
effects. Subsequent appendices will be organized by subject area, with the detailed assessment
of atmospheric science, exposure, health, and welfare evidence presented in separate appendices.
Thus, the focused integrated synthesis will make the ISA more concise than in the past, improve
its clarity and also its emphasis on policy-relevant scientific information and analyses; the ISA
scope, as addressed in section 4.3.2 is also more focused than in past ISAs (e.g., as discussed in
Pruitt [2018]). Each of the appendices will contain an evaluation of results from recent studies
integrated with previous findings (see section 4.4 for specific issues to be addressed).
Appendices for each broad health effect category (e.g., respiratory effects) will conclude with a
causal determination describing the strength of the evidence between exposure to O3 and the
health effect(s) [more detail on the types of causal determinations applied in the ISA is given in
the Preamble to the ISAs (U.S. EPA, 2015c) and in section 4.3.6 and Figure 4-1 of this chapter].
Likewise, the appendices devoted to ecological and climate evidence for welfare effects will
conclude with causality determinations for multiple effects on ecosystems and climate,
respectively.
4.3 ASSESSMENT APPROACH
4.3.1 Introduction
In developing ISAs, the EPA employs systematic review methodologies to identify and
evaluate relevant scientific information and produces summary text and figures to communicate
the state of the science to varied audiences. The process begins with a "Call for Information"
published in the Federal Register that announces the start of a NAAQS review and invites the
public to assist in this process through the submission of research studies in identified subject
areas. For the current O3 NAAQS review, this notice was published on June 26, 2018 (83 FR
29785). The subsequent ISA development steps are generally presented in Figure 4-1 and are
described in greater detail in the Preamble to the Integrated Science Assessments (U.S. EPA,
2015c), which provides a general overview of the ISA development process. The plan for
4-2
-------
developing the ISA for the current review is described in detail in the following sections. The
process for review of the draft ISA is described in Section 4.5.
Literature Search and
Study Selection
*
Evaluation of Individual Study Quality
After study selection, the quality of individual studies is evaluated by EPA or outside experts in the fields of
atmospheric science, exposure assessment, dosimetry, animal toxicology, controlled human exposure studies,
epidemiology, ecology, and otherwelfare effects, considering the design, methods, conduct, and documentation of
each study. Strengths and limitations of individual studies that may affect the interpretation of the study are
considered.
*
Develop Initial Sections
Review and summarize new study results as well
as findings and conclusions from previous
assessments by category of outcome/effect and
by discipline, e.g., toxicological studiesof lung
function.
Peer Input Consultation
Review of initial draft materials by scientists
from both outside and within EPA in public
meeting orpublic teleconference.
I-
Evaluation, Synthesis, and Integration of Evidence
Integrate evidence from scientific disciplines-forexample, toxicological, controlled human exposure, and
epidemiologic study findings for a particular health outcome. Evaluate evidence for related groups of endpoints or
outcomes to draw conclusions regarding health orwelfare effect categories, integrating health orwelfare effects
evidence with information on mode of action and exposure assessment.
Development of Scientific Conclusions and Causal Determinations
Characterize weight of evidence and develop judgments regarding causality for health or welfare effect categories.
Develop conclusions regarding concentration-ordose-response relationships, potentially at-risk populations,
lifestages, orecosystems.
Draft Integrated Science Assessment
Evaluation and integration of newly published studies
after each draft.
Clean Air Scientific Advisory Committee
Indepe nd e nt rev ie w of d raft docume nts fo r scie ntific
quality and sound implementation of causal
framework; anticipated reviewoftwo drafts oflSA in
public meetings.
Public Comments
Comments on draft ISA solicited by EPA
Final Integrated Science Assessment
Source: Modified from Figure II of the Preamble to the Integrated Science Assessments (U.S. EPA, 2015c).
Figure 4-1. General process for development of Integrated Science Assessments.
4-3
-------
The ISA is developed by authors who are EPA scientists in NCEA with extensive
knowledge in their respective fields and extramural scientists who are solicited by the EPA for
their subject matter expertise. The ISA authors apply systematic review methodologies to
identify relevant scientific findings that have emerged since the previous assessment. The
process is further described in sections below, including clear definition of the scope (Section
4.3.2), literature search and identification of relevant studies (Section 4.3.3), evaluation of
individual study quality (Section 4.3.5), evaluation of relevant studies (Section 4.3.6) and
evidence integration and determination of causality (Section 4.3.7).
4.3.2 Scope of the ISA
Through periodic reviews of the available scientific evidence, ISAs build on the data and
conclusions of previous assessments. The previous O3 ISA was published in 2013 (U.S. EPA,
2013) and included peer-reviewed literature published through July 2011. The ISA for the
current review will identify and evaluate studies published since 2011, synthesizing and
integrating the new evidence in the context of the conclusions from the previous review. Key
findings, conclusions, and uncertainties from the 2013 ISA will be briefly summarized at the
beginning of individual sections. Important older studies may be discussed to reinforce key
concepts and conclusions. Older studies also may be the primary focus in some subject areas or
scientific disciplines where research efforts have subsided, and these older studies remain the
definitive works available in the literature.
Scientific information will be identified and evaluated in order to provide a better
understanding of the following issues: (1) the natural and anthropogenic sources of O3 precursors
in the ambient air; (2) formation, transport, and fate of O3 in the environment; (3) measurement
methods and ambient air concentrations of O3; (4) how exposure assessment methods used in
epidemiologic studies can influence inferences drawn about O3 health effects; (5) the
independent effect of O3 exposure on health and welfare; (6) the potential influence of other
factors (e.g., other pollutants in the ambient air, ambient air temperature) shown to be correlated
with O3 and health or welfare effects; (7) the shape of the concentration-response relationship at
O3 concentrations at the low end of the distribution; and (8) populations and lifestages at
increased risk of 03-related health effects. As was the case for previous reviews, the ISA for this
review will focus mainly on the assessment of health and welfare effects resulting from exposure
to surface-level concentrations of tropospheric O3 whereas less emphasis will be accorded to
other photochemical oxidants because the available information is much more limited compared
to that for O3. Ozone is currently the indicator for the current NAAQS for photochemical
4-4
-------
oxidants, and the primary literature evaluating the health and ecological effects of photochemical
oxidants includes O3 almost exclusively as an indicator of photochemical oxidants73.
In the 2013 ISA, evidence from across scientific disciplines for related health and welfare
effects was evaluated, synthesized, and integrated to develop conclusions and causality
determinations. As described in the Preamble to the ISAs (U.S. EPA, 2015c) and in section 4.3.6
and Figure 4-1 of this chapter, the EPA uses a structured framework to provide a consistent and
transparent basis for classifying the weight of available evidence for health and welfare effects
according to a five-level hierarchy: (1) causal relationship; (2) likely to be a causal relationship;
(3) suggestive of, but not sufficient to infer, a causal relationship; (4) inadequate to infer the
presence or absence of a causal relationship; and (5) not likely to be a causal relationship. This
framework will be applied in the ISA for the current review.
In this review, the EPA will fully evaluate the available evidence for those health and
welfare effects for which the evidence in the 2013 ISA was less certain (i.e., effects where the
causality determination was "likely to be causal", "suggestive", or "inadequate" as described in
section 4.4.1) and where there is now a larger body of evidence. In doing so, the EPA aims to
evaluate the available evidence in order to address uncertainties and limitations in the evidence
identified in the prior review.
For those health and welfare effects for which the 2013 ISA concluded that the evidence
was sufficient to infer a causal relationship (i.e., for the health evidence: short-term O3
exposures [i.e., days to weeks] and respiratory effects; and for the welfare evidence: O3
exposures and ecological effects and effects on climate), the ISA for the current review will
integrate and synthesize the new evidence, placing emphasis on policy-relevant considerations,
such as the exposure conditions at which effects are observed, and characterizing the extent to
which new studies address key uncertainties and limitations identified in the previous review or
provide insight on new issues.
The scope of the health and welfare portions of the ISA is explicitly defined by scoping
tools that generally define the relevant Population, Exposure, Comparison, Outcome, and Study
Design (PECOS) (The PECOS tools for each category of information are provided in Section
4.3.3). The PECOS tool characterizes the parameters and provides a framework to aid in
identifying the relevant evidence in the literature to inform the ISA. There are discipline-specific
PECOS tools for experimental studies, epidemiologic studies, ecological studies and for studies
73Ozone is the only photochemical oxidant other than nitrogen dioxide (NO2) that is routinely monitored in ambient
air (i.e., EPA's AQS database; https://www.epa.gov/aqs). Data for other photochemical oxidants (e.g., PAN,
H2O2, etc.) typically have been obtained only as part of special field studies. Consequently, no data on nationwide
patterns of ambient air concentrations are available for these other photochemical oxidants; nor are extensive data
available on the relationships of concentrations and patterns of these photochemical oxidants to those of O3.
4-5
-------
on the effects of tropospheric O3 on climate, which differ depending on the types of questions to
be answered and are influenced by a priori knowledge related to that question. The use of
PECOS tools is a widely accepted and rapidly growing approach to systematic review in risk
assessment, and consistent with recommendations by the National Academy of Sciences for
improving the design of risk assessment through planning, scoping, and problem formulation to
better meet the needs of decision-makers (National Research Council 2009). The PECOS tools
serve as guides for several aspects of the ISA process, including the literature search strategy,
criteria for the inclusion or exclusion of studies in the ISA, the types of data extracted from
studies, and the integration and synthesis of the results.
4.3.3 Literature Search and Identification of Relevant Studies
4.3.3.1 Systematic Literature Search
The EPA uses a structured approach to identify relevant studies for consideration and
inclusion in the ISAs. The search for relevant literature in this review began with publication of
the Federal Register notice announcing the initiation of this O3 review and requesting
information from the public including relevant literature (83 FR 29785, June 26, 2018). In
addition, the EPA identifies publications by conducting a multi-tiered systematic literature search
that includes extensive mining of literature databases on specific topics in a variety of
disciplines. The search strategies are designed a priori to optimize identification of pertinent
published papers. Studies identified in the literature search are documented in the Health and
Environmental Research Online (HERO) database. The HERO project page for this ISA
(https://hero.epa.gov/hero/index.cfm/proiect/page/proiect id/2737) will contain the references
that will be considered for inclusion in the ISA and electronic links to bibliographic information
and abstracts. It is accessible to the public.
For this ISA, discipline-specific approaches will be used to identify literature. In each
case, careful consideration will be given to literature search strategies used in the development of
previous assessments and the methods that resulted in the best precision and recall for each of the
disciplines, including atmospheric science (section 4.3.4.1), exposure assessment (section
4.3.4.2), experimental health studies (section 4.3.4.3), epidemiology (section 4.3.4.4), ecology
(section 4.3.4.5), and climate (section 4.3.4.6). The literature identification approaches include
broad keyword searches in routinely used databases with Automatic Topic Classification, and
citation mapping (see section 4.3.4 for specific approaches used for each discipline).
As has been done for past ISAs, a broad keyword search was developed as a starting
point to capture literature pertinent to the pollutant of interest. In this case, the main keyword
string to be used is "ozone OR 03", which is sufficiently broad to capture 03-relevant literature
in each database (i.e., PubMed, Web of Science, TOXLINE). Following the broad keyword
4-6
-------
search for O3, automatic topic classification will be used to categorize references by discipline
(e.g., epidemiology, toxicology, etc.). This step employs machine learning where positive and
negative seed references74 for a particular discipline are used to train an algorithm to identify
discipline-specific references based on word use and frequency in titles and abstracts. This
method varies in effectiveness across disciplines due to the broad range of topics and variability
in term usage in some evidence bases. However, it is invaluable when effective, and has been
used in several prior ISAs.
Another approach used in past ISAs that will be employed in this review is citation
mapping, or relational reference searching. In this approach, a set of relevant published
references are identified as a seed set and then more recent literature that has cited any of the
references in the seed set are collected. References from the previous ISA for the respective
pollutant comprise the seed set for the new ISA. Because the seed set is highly relevant to the
topic of interest, this targeted approach to reference identification is more precise than keyword
searches, and it further allows for relevance ranking based on the number of references in a
bibliography that match references in the seed set.
References may be identified for inclusion in several additional ways including:
identification of relevant literature by EPA expert scientists; recommendations received in
response to the call for information and the external review process for the ISA; and review of
citations included in previous assessments.
All of these search methods will be used to identify recent research published or accepted
for publication starting January 1, 2011, providing some overlap with the July 2011 cutoff date
from the last review. Although published after the literature cutoff date (March 30, 2018 for this
review), studies published after this date that were identified by comments submitted in response
to the Call for Information will be considered. Further, studies may also be considered in
subsequent phases of the NAAQS review (e.g., studies identified by CASAC members during
review of the draft ISA), particularly to the extent that they provide new information that affects
key scientific conclusions.
4.3.3.2 Initial Screening (Level 1) of Studies from Literature Search
Once studies are identified, ISA authors (EPA staff and extramural scientists) will review
the studies for relevance. For the primary O3 NAAQS, relevant studies include epidemiologic,
toxicological, and controlled human exposure studies, including studies of dosimetry and mode
74 Positive seed references are those that are examples of references that are relevant, i.e., the references would be
selected for full-text screening. Negative seed references are those that are examples of references that are not
relevant, i.e., they would not be selected for full-text screening. For ISAs, the positive seed set includes references
from the prior ISA for the discipline of interest. The negative seed set includes the references from all of the other
disciplines in the prior ISA.
4-7
-------
of action, or those that examine ambient air O3 exposure assessment, atmospheric chemistry,
sources and emissions. For the review of the secondary O3 NAAQS, relevant studies are those
that examine ecological effects and the effects of O3 on climate. Specific information detailing
the scope of the ISA for the current review, and subsequently those studies that will be evaluated
within it, are detailed above in section 4.3.2.
As described above, the literature search methods will be targeted for discipline-relevant
references to the extent possible, and the subsequent screening will result in a further refined list
of references to be included in the ISA. References for each discipline will first undergo title and
abstract screening using SWIFT-ActiveScreener (SWIFT-AS), which is referred to as Level 1
screening. Level 1 screening criteria for inclusion will be broad and err on the side of inclusion.
For each discipline, title and abstracts will be selected for inclusion if there is indication of O3
and a quantifiable effect relevant to that discipline. SWIFT-AS is a software application that
employs machine learning in real-time to identify relevant literature. The machine learning
feature builds a model to predict relevant references based on inclusion/exclusion screening
decisions in real-time as scientists screen each reference. As title/abstract screening is conducted,
references are queued based on the predicted relevance and SWIFT-AS further predicts when a
95% recall threshold has been reached75, a level often used to evaluate the performance of
machine learning applications and considered comparable to human error rates (Cohen et al.
2006, Howard et al. 2016).
The application of SWIFT-AS will be tailored for each discipline. This will include using
a specific seed set of 50-100 relevant references from the 2013 ISA to train the SWIFT-AS
algorithm and developing specific screening questions for each discipline to allow for the
categorization of references based on the information available in the title and abstract.
Understanding the volume and topics of the recent literature on O3 will be important information
to consider in refining the scope of the ISA. Specific details about inclusion/exclusion criteria
and the screening questions for each discipline are described in more detail below.
Following Level 1 screening, references identified for inclusion will be acquired and
compiled in HERO for full-text Level 2 screening conducted by NCEA subject matter experts.
The Level 2 screening decisions for each discipline will be based on the scoping decisions (see
section 4.3.4). References will be tracked for both relevance to the broad ISA and for the defined
scope for each topic area (e.g., outcome category).
75 A 95% recall threshold represents the point at which 95% of the potentially relevant references have been
identified.
4-8
-------
4.3.3.3 Criteria of In-Scope Studies
To be included in the ISA, relevant studies and reports must have undergone scientific
peer review and have been published or accepted for publication before the cutoff date. Some
publications retrieved from the literature search will be excluded as not being relevant in Level 1
screening based on the title/abstract (e.g., not about air pollution, conference abstract, review
articles, commentaries). For other publications, decisions about relevance will be made in Level
2 screening as they require reading beyond the title. These publications will be labeled as
"considered" for inclusion in the ISA. Inclusion and exclusion decisions will be documented in
the HERO database (https://hero.epa.gov/hero/index.cfm/proiect/page/proiect id/2737).
4.3.4 Discipline-Specific Scoping, Searching and Screening
4.3.4.1 Atmospheric Science
4.3.4.1.1 Scope
The ISA will present and evaluate relevant data and summarize the current scientific
understanding, based on evidence available from previous reviews and new evidence that has
emerged since the 2013 ISA concerning the sources and concentrations of O3 in the lower
troposphere and surface boundary layer. Ozone present in the lower troposphere
is predominantly formed through photochemical reaction between oxides of nitrogen (NOx)
and volatile organic precursor gases. This ISA discussion will focus on: O3 that would be present
in the lower atmosphere in the absence of any manmade emissions in the U.S. (i.e., O3 that has
been transported across international boundaries, produced by natural processes such as lightning
or drawn down from the stratosphere, or forms from natural or internationally transported
precursors), referred to as "U.S. background" O3; and ambient air O3 sources, measurements, and
concentration trends.
4.3.4.1.2 Search and Screen
Literature related to atmospheric science topics will be identified by citation mapping
methods that will rely upon references cited in the 2013 ISA. More specifically, references will
be collected from the atmospheric science sections of the 2013 ISA, including sub-topics on
physical and chemical processes, atmospheric modeling, monitoring, and background O3
concentrations. Citation mapping will be conducted in Web of Science. The focus for evaluation
of the recent literature will be on background concentration of O3 in ambient air.
4.3.4.2 Exposure Assessment
4.3.4.2.1 Scope
The ISA will describe the commonly employed exposure assessment methods in the
epidemiologic evidence, including strengths and limitations of the methods, study designs in
4-9
-------
which those methods are used, and how errors and uncertainties inherent in those methods
influence the bias and precision of health effect estimates for short-term and long-term O3
exposure studies. The exposure assessment appendix includes a summary table that describes
each method, how it is used in epidemiologic studies, and how strengths and limitations of each
method may impact interpretation of the epidemiologic results.
4.3.4.2.2 Search and Screen
Exposure literature relevant to O3 will be identified using the broad keyword search
described in Section 4.3.2 and Automatic Topic Classification. Automatic Topic Classification
for exposure references will include a sufficiently large set of positive and negative seeds from
previous ISAs. More specifically, positive seeds will include references from the exposure
chapter from the 2016 NOx ISA76 and the 2013 ISA; the negative seeds will include non-relevant
references (i.e., those from other disciplines in these two ISAs). Following identification and
binning of the literature, SWIFT-AS will be used for Level 1 screening. Positive seeds to train
the SWIFT-AS algorithm will include a subset of the exposure references cited in the 2013 ISA.
Additionally, references will be categorized in Level 1 screening in SWIFT-AS by study type,
study location, and exposure duration. The references identified for inclusion in Level 1 will then
undergo Level 2 full-text screening.
4.3.4.3 Health - Experimental Studies
4.3.4.3.1 Scope
For experimental studies, specifically controlled human or animal exposure studies, the
evaluation will focus on those studies that also address key uncertainties and limitations in the
evidence identified in the previous review. For example, does the new evidence advance
understanding of or call into question prior conclusions regarding the biological mechanisms by
which O3 elicits a health effect or provide coherence for the effects assessed in epidemiologic
studies? The scope of the experimental evidence encompasses studies of short-term (i.e., hours
to weeks) and long-term (i.e., months to years) exposures conducted at concentrations of O3 that
are relevant to the range of human exposures to ambient air (up to 2 ppm, which is one to two
orders of magnitude above ambient air concentrations) (Table 4-1).
76 The 2016 NOx ISA is the most recent ISA that had the appropriate level of detail comparative to what is needed
for this current review.
4-10
-------
Table 4-1. PECOS tool to define the parameters and provide a framework for identifying
relevant experimental studies.
Exposure Duration and
Health Effect
Population, Exposure, Comparison, Outcome, Study Design (PECOS)
Short-term exposure and
respiratory, cardiovascular,
metabolic, nervous system,
reproductive or
developmental effects
Population: study populations of any controlled human exposure or animal toxicological
study of mammals at any lifestage
Exposure: short-term (in the order of minutes to weeks) inhalation exposure to relevant O3
concentrations (i.e., 0.4 ppm or below for humans, 2 ppm or below for other mammals)
Comparison: human subjects that serve as their own controls with an appropriate washout
period or when comparison to a reference population exposed to lower levels is available,
or, in toxicological studies of mammals, an appropriate comparison group that is exposed to
a negative control (i.e., clean air or filtered air control)
Outcome: respiratory, cardiovascular, metabolic, nervous system, reproductive or
developmental effects
Study Design: controlled human exposure (i.e., chamber) studies; In vivo acute, subacute
or repeated-dose toxicity studies in mammals, reproductive toxicity or immunotoxicity
studies
Long-term exposure and
respiratory, cardiovascular,
metabolic, nervous system,
carcinogenic, reproductive or
developmental effects
Population: study population of any animal toxicological study of mammals at any lifestage
Exposure: long-term (in the order of months to years) inhalation exposure to relevant O3
concentrations (i.e., 2 ppm or below)
Comparison: appropriate comparison group exposed to a negative control (i.e., clean air or
filtered air control)
Outcome: respiratory, cardiovascular, metabolic or nervous system, carcinogenic,
reproductive or developmental effects
Study Design: In vivo chronic, subchronic or repeated-dose toxicity studies in mammals;
reproductive toxicity or immunotoxicity studies; genotoxicity/mutagenicity studies
Population: in controlled human exposure studies, aenerallv healthv adults approved for studv participation bv the
appropriate institutional review board or ethics committee; for toxicological studies, well-defined/well-characterized strains of
mammals at any lifestage.
Exposure: O3 concentrations deliberately delivered to subjects for a predefined duration
Comparator: in controlled human exposure studies, subjects serve as their own controls with an appropriate washout period,
or a reference population exposed to lower O3 concentrations, or, in toxicological studies, an appropriate comparison group
that is exposed to a negative control (i.e., clean air or filtered air control)
Outcome: clearlv measurable health endpoint.
Studv desiqn: controlled human exposure (i.e., chamber) studies; In vivo acute, subacute, subchronic, chronic or repeated-
dose toxicity studies in mammals; reproductive toxicity or immunotoxicity studies; genotoxicity/mutagenicity studies
4.3.4.3.2 Search and Screen
Identification of experimental (i.e., controlled human exposure and animal toxicology)
studies examining the health effects of O3 exposure will be identified using the broad keyword
search described in Section 4.3.2 and Automatic Topic Classification. The Automatic Topic
Classification for experimental references will include a sufficiently large set of positive seeds,
including controlled human exposure and animal toxicology references cited in the 2016 NOx
ISA and the 2013 ISA, and a sufficiently large set of negative seeds, including nonexperimental
references cited in these two ISAs. Following identification of the literature, SWIFT-AS will be
used for Level 1 screening. The SWIFT-AS algorithm will be trained using a set of positive seed
references from a selection of controlled human exposure and animal toxicology studies cited in
4-11
-------
the 2013 ISA. Additionally, references will be categorized in Level 1 screening in SWIFT-AS by
health outcome category (e.g., respiratory, cardiovascular, metabolic, etc.), exposure duration
(e.g., short-term, long-term), and study type (e.g., controlled human exposure, animal toxicology,
etc.). The references identified for inclusion at Level 1 will then undergo Level 2 full-text
screening, for each health outcome category, for relevance to the defined scope as described
above.
4.3.4.4 Health - Observational (Epidemiologic) Studies
4.3.4.4.1 Scope
The evaluation of epidemiologic studies will focus on the associations between short- and
long-term exposure to O3 and a range of health effects, including respiratory, cardiovascular,
reproductive and developmental, metabolic, and nervous system outcomes (Table 4-2). In
instances when a "causal" or "likely to be a causal" relationship was concluded in the 2013
Ozone ISA (e.g., short-term O3 exposure and respiratory and cardiovascular effects and total
mortality, and long-term O3 exposure and respiratory effects), the epidemiologic studies
evaluated for those outcomes are more limited in scope (i.e., targeted towards study locations
that include U.S. airsheds or airsheds that are similar to those found in the U.S.), as reflected in
the PECOS tool. For outcomes for which the 2013 Ozone ISA concluded that evidence was
"suggestive of' or "inadequate to infer" a causal relationship, the epidemiologic studies
evaluated are not limited geographically or by airshed characteristics, as reflected in the PECOS
tool. The discussion of epidemiologic results will emphasize the impact of exposure assessment
techniques on associations observed; evaluating potential copollutant confounding; examining
heterogeneity in O3 associations; and the shape of the concentration-response relationship.
4-12
-------
Table 4-2. PECOS tool to define the parameters and provide a framework for identifying
relevant epidemiologic studies.
Exposure Duration and
Health Effect
Population, Exposure, Comparison, Outcome, Study Design (PECOS)
Short-term exposure and
respiratory effects
Population: any U.S. or Canadian population, including populations or lifestages that might be
at increased risk
Exposure: short-term (on the order of one to several days) ambient air concentration of O3
Comparison: per unit increase (in ppb)
Outcome: change in risk (incidence/prevalence) of respiratory effects
Study Design: epidemiologic studies consisting of panel, case-crossover, time-series studies,
and case-control studies; cross-sectional studies with appropriate timing of exposure for the
health endpoint of interest
Short-term exposure and
mortality
Population: any U.S. or Canadian population, including populations or lifestages that might be
at increased risk
Exposure: short-term exposure (on the order of one to several days) to ambient air
concentrations of O3
Comparison: per unit increase (in ppb)
Outcome: change in risk (incidence) of mortality
Study Design: epidemiologic studies consisting of case-crossover or time-series studies with
appropriate timing of exposure for the health endpoint of interest
Long-term exposure and
respiratory effects
Population: any U.S. or Canadian population, including populations or lifestages that might be
at increased risk
Exposure: long-term (on the order of months to years) ambient air concentration of O3
Comparison: per unit increase (in ppb)
Outcome: change in risk (incidence/prevalence) of respiratory effects
Study Design: epidemiologic studies consisting of cohort and case-control studies; time-
series, case-crossover, and cross-sectional studies with appropriate timing of exposure for the
health endpoint of interest
Short-term exposure and
cardiovascular effects
Population: any U.S., Canadian, European or Australian population, including populations or
lifestages that might be at increased risk
Exposure: short-term (on the order of one to several days)_ambient air concentration of O3
Comparison: per unit increase (in ppb)
Outcome: change in risk (incidence/prevalence) of cardiovascular effects
Study Design: epidemiologic studies consisting of panel, case-crossover, time-series studies,
and case-control studies; cross-sectional studies with appropriate timing of exposure for the
health endpoint of interest
Short-term exposure and
nervous system effects
Population: any population, including populations or lifestages that might be at increased risk
Exposure: short-term (on the order of one to several days) ambient air concentration of O3
Comparison: per unit increase (in ppb)
Outcome: change in risk (incidence/prevalence) of a nervous system effect
Study Design: epidemiologic studies consisting of panel, case-crossover, time-series studies,
and case-control studies; cross-sectional studies with appropriate timing of exposure for the
health endpoint of interest
Long-term exposure and
cardiovascular, nervous
system, reproductive or
developmental effects,
cancer, or mortality
Population: any population, including populations or lifestages that might be at increased risk
Exposure: long-term (on the order of months to years) ambient air concentration of O3
Comparison: per unit increase (in ppb)
Outcome: change in risk (incidence/prevalence) of a cardiovascular, nervous system,
reproductive or developmental, cancer or mortality effect
Study Design: epidemiologic studies consisting of cohort and case-control studies; time-
series, case-crossover, and cross-sectional studies with appropriate timing of exposure for the
health endpoint of interest
4-13
-------
Population: the general population, all age groups, living both in urban and in rural areas exposed on a daily basis to O3
through outdoor (ambient) air, and not exclusively in occupational settings or as a result of indoor exposure. Populations and
lifestages at increased risk are included, such as those with specific pre-existing health conditions (e.g. respiratory or
cardiovascular diseases), children, or older adults.
Exposure: ambient air O3 from any source measured as short-term (minutes to weeks) or long-term (months to years).
Comparator: the health effect observed by unit increase in concentration of O3 in the same or in a control population.
Outcome: clearly measurable health endpoint.
Study design: epidemiologic studies on health effects of O3 consisting of cross-sectional, case-control, case-crossover,
cohort, panel and time-series studies.
4.3.4.4.2 Search and Screen
Identification of recent epidemiologic studies examining a health effect and ambient air
exposure to O3 will be identified using the broad keyword search described in Section 4.3.3 and
Automatic Topic Classification. The approach for Automatic Topic Classification to identify
epidemiologic studies from the broad literature search results parallels the approach described in
Section 4.3.4.3.2 for the experimental studies. A sufficiently large set of seed references cited in
the 2016 NOx and 2013 ISAs will be used, with positive seeds comprised of epidemiologic
references in those ISAs and negative seeds comprised of all references other than epidemiologic
references. Following identification of the literature, SWIFT-AS will be used for Level 1
screening. Positive seeds will also be used to train the SWIFT-AS algorithm and will include
select epidemiologic references cited in the 2013 ISA. Additionally, references will be
categorized in Level 1 SWIFT-AS screening by health outcome category (e.g., mortality,
respiratory, cardiovascular, etc.), exposure duration (e.g., short-term, long-term), and study
location (e.g., U.S., Canada, Europe, etc.). The references identified for inclusion in Level 1
screening will then undergo Level 2 full-text screening, for each health effect category, for
relevance to the defined scope.
4.3.4.5 Welfare Effects - Ecological Studies
4.3.4.5.1 Scope
With respect to ecological effects, this ISA will build on information available during the
last review describing the effect of O3 exposure on vegetation and ecosystems. For research
evaluating ecological effects, emphasis will be placed on recent studies that: (1) evaluate effects
of exposures resulting from O3 concentrations comparable to those occurring in North American
airsheds and (2) investigate effects on any individual, population (in the sense of a group of
individuals of the same species), community, or ecosystem in North America (Table 4-3). In
instances when a "causal relationship" was concluded in the 2013 ISA (i.e., visible foliar injury,
vegetation growth, reduced yield/quality of agricultural crops, reduced productivity, alteration of
belowground biogeochemical cycles) the current review will only evaluate studies conducted in
North America. For all other ecological endpoints in Table 4-3 (terrestrial water cycling, carbon
4-14
-------
sequestration, terrestrial community composition, plant reproduction, phenology, or mortality,
insects, other wildlife, plant-animal signaling) there are no geographic constraints and all
available evidence will be considered.
4.3.4.5.2 Search and Screen
Studies relevant to the ecological effects of O3 exposure will be identified by citation
mapping. The broad keyword searches and Automatic Topic Classification have not resulted in a
well-targeted set of references for Level 1 screening in past ISAs for ecological endpoints.
Citation mapping in Web of Science based on ecological studies cited in the 2013 ISA is
expected to yield a more refined set of references. Following citation mapping, Level 1 screening
of the identified references will be conducted in SWIFT-AS, including the use of a seed set of
ecological references from the 2013 ISA. Screening questions to facilitate organization of the
literature will include effect category (e.g., foliar injury, plant growth, biodiversity, etc.),
exposure conditions, location, and ecosystem type (e.g., wetland, crop, etc.). As will be the case
for the other disciplines, Level 2 full-text screening will be conducted for references included in
Level 1 screening, and full-text inclusion criteria will be defined by the scope.
4-15
-------
Table 4-3. PECOS tool to define the parameters and provide a framework for identifying
relevant ecological studies.
Ecological Endpoint
Population, Exposure, Comparison, Outcome, Study Design (PECOS)
Visible foliar injury, vegetation
growth, yield/quality of agricultural
crops, productivity, belowground
biogeochemical cycling
Population: For any species, an individual, population (in the sense of a group of
individuals of the same species), community, or ecosystem in North America
Exposure: Concentrations occurring in the environment or experimental O3
concentrations within an order of magnitude of recent concentrations (as described in
Appendix 1)
Comparison: Relevant control sites, treatments, or parameters
Outcome: Visible foliar injury, alteration of vegetative growth, yield/quality of agricultural
crops, productivity, belowground biogeochemical cycles
Study Design: Laboratory, greenhouse, OTC, FACE, field, gradient, or modeling studies
Terrestrial water cycling; carbon
sequestration; terrestrial
community composition; plant
reproduction, phenology, or
mortality; insects, other wildlife,
plant-animal signaling
Population: For any species, an individual, population (in the sense of a group of
individuals of the same species), community, or ecosystem in any continent1
Exposure: Concentrations occurring in the environment or experimental O3
concentrations within an order of magnitude of recent concentrations (as described in
Appendix 1)
Comparison: Relevant control sites, treatments, or parameters
Outcome: Alteration of: terrestrial water cycling; carbon sequestration; terrestrial
community composition; plant reproduction, phenology, mortality; growth reproduction
and survival of insects and other wildlife; plant-animal signaling
Study Design: Laboratory, greenhouse, OTC, FACE, field, gradient, or modeling studies
PoDulation = unit of studv;
Exposure = environmental variable to which population is exposed;
Comparator = chanae in endpoint observed bv unit increase in concentration of O3 in the same or in a control population;
Outcome = measurable endpoint resultina from exposure;
Studv desiqn = laboratory, field, aradient, open top chamber (OTC), Free-Air Carbon Dioxide Enrichment (FACE),
greenhouse, and modeling studies.
Notes: This definition of population is for the purpose of applying PECOS to ecology. Ecological populations are defined as a
group of individuals of the same species.
1 In cases where a comprehensive list of affected species was available, non-agricultural North American species were
separated out from the larger datasets and the evidence was evaluated (e.g. foliar injury, biomass)
4.3.4.6 Welfare - Effects on Climate
4.3.4.6.1 Scope
For effects on climate, the ISA will focus on effects of tropospheric O3 on climate,
consistent with the inclusion of "climate" in the list of effects on welfare in section 302(h) of the
Clean Air Act. The ISA will not focus on downstream ecosystem effects, human health effects,
or future air quality projections resulting from changes in climate. Studies that inform the
independent role of O3 in climate forcing as well as effects on U.S. national and regional climate
are within the scope of the literature to be considered in the review (Table 4-4). In addition, the
ISA will assess available evidence on the effects of tropospheric O3 as an absorber of UV-B
radiation in the troposphere, though a PECOS tool is no anticipated to be necessary to assist in
narrowing and scoping the consideration of the limited available evidence.
4-16
-------
Table 4-4. PECOS tool to define the parameters and provide a framework for identifying
relevant studies on the effects of tropospheric O3 on climate.
Effect on Climate
Population, Exposure, Comparison, Outcome, Study Design (PECOS)
Changes in radiative forcing (RF)
Population/Geographical scope: evaluations of radiative forcing at the regional,
continental, and/or global scale
Exposure: tropospheric O3 concentration distributions in 3D (observed/modeled)
Comparison: relevant baseline or unperturbed scenarios/conditions
Outcome: changes in RF resulting from change in tropospheric O3
Study Design: observations or modeling studies
Changes in climate (e.g., surface
temperature, hydrological cycle)
Population/Geographical scope: evaluations of climate effects at the regional,
continental, and/or global scale
Exposure: tropospheric O3 concentration distributions in 3D (observed/modeled)
Comparison: relevant baseline or unperturbed scenarios/conditions
Outcome: subsequent climate effects (via radiative forcing) (e.g., global surface
temperature) resulting from change in tropospheric O3
Study Design: observational or modelling studies
PoDulation/GeoaraDhical scope: spatial extent of study
Exposure: environmental variable (tropospheric O3 concentrations)
Comparator: radiative forcina or climate effects observed from unit chanae in tropospheric O3 concentration.
Outcome: relevant radiative forcina or climate outcomes resultina from chanae in tropospheric O3.
Study desiqn: observations/satellite, modellina
4.3.4.6.2 Search and Screen
Studies examining the effect of tropospheric O3 on climate will be identified in two ways.
First, references will be identified by citation mapping in Web of Science using references cited
in the 2013 ISA. In addition, relevant references will be identified from recent national and
international climate assessments, such as the National Climate Assessment (USGCRP, 2017)
and Intergovernmental Panel on Climate Change (IPCC, 2013), and other recent, more focused
reports relevant to O3 climate forcing. Level 1 screening of the identified references will be
conducted in SWIFT-AS aided by a seed set of select references from the climate section of the
2013 ISA and screening questions to facilitate organization of the literature. The screening
questions will pertain to the following topics: radiative forcing, climate impacts, precursor and
copollutant effects, and factors and feedbacks. Level 2 screening will be conducted for
references included in Level 1 and full-text inclusion criteria will be defined by the scope.
4.3.5 Identification of Policy-Relevant Studies
From the group of "considered" references (see section 4.3.4), studies and reports will be
selected for inclusion in the ISA based on review of the full text. The selection process will be
based on the extent to which the study is potentially policy-relevant and informative. Potentially
policy-relevant and informative studies will include those that provide a basis for or describe the
relationship between exposure to O3 and effects, particularly, those studies that reduce
uncertainty or address limitations of critical issues. Also pertinent are studies that offer
4-17
-------
innovation in method or design or present novel information on effects or issues previously not
identified. Uncertainty can be addressed to some extent, for example, by analyses informing the
independent effect of O3 on health and welfare effects, analyses of potential confounding or
effect modification by co-pollutants or other factors, analyses of concentration-response or dose-
response relationships, or analyses related to time between exposure and response. In keeping
with the ISA's intent to accurately reflect the latest scientific knowledge, the focus of the
discussion in the ISA will be on studies published since July 2011 (i.e., the literature cutoff date
for the 2013 ISA). Building on the last review, the EPA plans to evaluate the recent evidence in
the context of the conclusions from the 2013 ISA. In some cases, evidence from older studies
may be the key policy-relevant information in a particular subject area or scientific discipline and
will be included. Analyses conducted by the EPA using publicly available data—for example, air
quality and emissions data—will also be considered for inclusion in the ISA. Informative studies
will not be limited to specific study designs, model systems, or outcomes.
While study quality is important, it is not the sole criteria for study inclusion. The
combination of approaches described above are intended to produce a comprehensive collection
of pertinent studies needed to address the key scientific issues that form the basis of the ISA.
References for the included studies will be cited in the ISA with a hyperlink to the HERO
database.
4.3.6 Evaluation of Individual Study Quality
After selecting studies for inclusion, individual study quality is evaluated by considering
the design, methods, conduct, and documentation of each study, but not the study results. In the
ISA for the current review, conclusions about the strength of inference from study results will be
made by independently evaluating the overall quality of each study (U.S. EPA, 2015c). This
uniform approach aims to consider the strengths, limitations, and possible roles of chance,
confounding, and other biases that may affect the interpretation of individual studies and the
strength of inference from the results of the study.
More specifically, NCEA will employ a structured, narrative approach to evaluate a sub-
set of health studies (i.e., animal toxicology, controlled human exposure, and epidemiology
studies) using specific study domains, including study design, study population, exposure,
outcome assessment, potential confounding, and statistical analysis. For a subset of studies that
are the most policy relevant, the evaluation will be documented in a narrative format to
transparently convey the overall conclusion on study quality that determines if the study should
be included in the ISA. These narrative study quality evaluations will document study details for
specific study domains (e.g., study population, study design) and will record expert judgments as
4-18
-------
well. The study quality evaluations will be publicly available and accessible via the HERO
website.
In general, in assessing the scientific quality of studies on health and welfare effects, the
following questions are considered.
• Were the study design, study groups, methods, data, and results clearly presented in
relation to the study objectives to allow for study evaluation? Were limitations and any
underlying assumptions of the design and other aspects of the study stated?
• Were the ecosystems, study site(s), study populations, subjects, or organism models
adequately selected, and are they sufficiently well-defined to allow for meaningful
comparisons between study or exposure groups?
• Are the air quality, exposure, or dose metrics of adequate quality and are they sufficiently
representative of or pertinent to ambient air?
• Are the health or welfare effect measurements meaningful, valid, and reliable?
• Were likely covariates or modifying factors adequately controlled or taken into account in
the study design and statistical analysis?
• Do the analytical methods provide adequate sensitivity and precision to support
conclusions?
• Were the statistical analyses appropriate, properly performed, and properly interpreted?
Additional considerations in evaluating individual study quality specific to particular
scientific disciplines are discussed in detail in the Preamble to the ISAs (U.S. EPA, 2015c) and
will be further described in Appendix 10 of the ISA.
4.3.7 Integration of Evidence and Determination of Causality
As described in the Preamble to the ISAs (U.S. EPA, 2015c), the EPA uses a structured
framework to provide a consistent and transparent basis for classifying the weight of available
evidence for health and welfare effects according to a five-level hierarchy: (1) causal
relationship; (2) likely to be a causal relationship; (3) suggestive of, but not sufficient to infer, a
causal relationship; (4) inadequate to infer a causal relationship; and (5) not likely to be a causal
relationship (Table 4-5).
4-19
-------
Table 4-5. Weight of evidence determinations.
Health Effects
Welfare Effects
Causal Evidence is sufficient to conclude that there is a causal
relationship relationship with relevant pollutant exposures
(e.g., doses or exposures generally within one to two
orders of magnitude of recent concentrations). That is,
the pollutant has been shown to result in health effects
in studies in which chance, confounding, and other
biases could be ruled out with reasonable confidence.
For example: (1) controlled human exposure studies
that demonstrate consistent effects, or
(2) observational studies that cannot be explained by
plausible alternatives or that are supported by other
lines of evidence (e.g., animal studies or mode of
action information). Generally, the determination is
based on multiple high-quality studies conducted by
multiple research groups.
Evidence is sufficient to conclude that there is a
causal relationship with relevant pollutant exposures.
That is, the pollutant has been shown to result in
effects in studies in which chance, confounding, and
other biases could be ruled out with reasonable
confidence. Controlled exposure studies (laboratory
or small- to medium-scale field studies) provide the
strongest evidence for causality, but the scope of
inference may be limited. Generally, the
determination is based on multiple studies conducted
by multiple research groups, and evidence that is
considered sufficient to infer a causal relationship is
usually obtained from the joint consideration of many
lines of evidence that reinforce each other.
Likely to be a Evidence is sufficient to conclude that a causal
causal relationship is likely to exist with relevant pollutant
relationship exposures. That is, the pollutant has been shown to
result in health effects in studies where results are not
explained by chance, confounding, and other biases,
but uncertainties remain in the evidence overall. For
example: (1) observational studies show an
association, but copollutant exposures are difficult to
address and/or other lines of evidence (controlled
human exposure, animal, or mode of action
information) are limited or inconsistent, or (2) animal
toxicological evidence from multiple studies from
different laboratories demonstrate effects but limited or
no human data are available. Generally, the
determination is based on multiple high-quality studies.
Evidence is sufficient to conclude that there is a likely
causal association with relevant pollutant exposures.
That is, an association has been observed between
the pollutant and the outcome in studies in which
chance, confounding, and other biases are
minimized but uncertainties remain. For example,
field studies show a relationship, but suspected
interacting factors cannot be controlled, and other
lines of evidence are limited or inconsistent.
Generally, the determination is based on multiple
studies by multiple research groups.
Suggestive of, Evidence is suggestive of a causal relationship with
but not relevant pollutant exposures but is limited, and chance,
sufficient to confounding, and other biases cannot be ruled out. For
infer, a causal example: (1) when the body of evidence is relatively
relationship small, at least one high-quality epidemiologic study
shows an association with a given health outcome
and/or at least one high-quality toxicological study
shows effects relevant to humans in animal species, or
(2) when the body of evidence is relatively large,
evidence from studies of varying quality is generally
supportive but not entirely consistent, and there may
be coherence across lines of evidence (e.g., animal
studies or mode of action information) to support the
determination.
Evidence is suggestive of a causal relationship with
relevant pollutant exposures, but chance,
confounding, and other biases cannot be ruled out.
For example, at least one high-quality study shows
an effect, but the results of other studies are
inconsistent.
Inadequate to Evidence is inadequate to determine that a causal
infer a causal relationship exists with relevant pollutant exposures,
relationship The available studies are of insufficient quantity,
quality, consistency, or statistical power to permit a
conclusion regarding the presence or absence of an
effect.
Evidence is inadequate to determine that a causal
relationship exists with relevant pollutant exposures.
The available studies are of insufficient quality,
consistency, or statistical power to permit a
conclusion regarding the presence or absence of an
effect.
4-20
-------
Health Effects Welfare Effects
Not likely to be Evidence indicates there is no causal relationship with Evidence indicates there is no causal relationship
a causal relevant pollutant exposures. Several adequate with relevant pollutant exposures. Several adequate
relationship studies, covering the full range of levels of exposure studies examining relationships with relevant
that human beings are known to encounter and exposures are consistent in failing to show an effect
considering at-risk populations and lifestages, are at any level of exposure,
mutually consistent in not showing an effect at any
level of exposure.
Source: U.S. EPA (2015c)
Determination of causality involves evaluating and integrating evidence for different
types of health or welfare effects associated with short- and long-term exposure periods. Key
considerations in drawing conclusions about causality include consistency of findings for an
endpoint across studies, coherence of the evidence across disciplines and across related
endpoints, and biological plausibility. As judged by these parameters, studies in which chance,
confounding, and other biases could be ruled out with reasonable confidence are sufficient to
infer a causal relationship. Increasing uncertainty due to limited available information,
inconsistency across the body of evidence, and/or limited coherence and biological plausibility
may lead to conclusions lower in the causality hierarchy. Causality determinations are based on
the confidence in the integrated body of evidence, considering study design and quality and
strengths and weaknesses in the overall collection of previous and recent studies across
disciplines. In discussing each determination of causality, the EPA characterizes the evidence
upon which the judgment is based, including the extent of and weight of evidence for individual
endpoints within the health or welfare effect category or group of related endpoints.
For evaluation of human health effects, determinations of causality are made for major
health effect categories or groups of related endpoints (e.g., respiratory effects) and for the range
of exposure concentrations of O3 defined to be relevant to ambient air concentrations (e.g., up to
2 ppm). The main lines of evidence for use in causality determinations for human health are
controlled human exposure, epidemiologic, and animal toxicological studies. Evidence is
integrated from previous and recent studies. Other information including mechanistic evidence,
toxicokinetics, and exposure assessment may be drawn upon if relevant to the evaluation of
health effects and if of sufficient importance to affect the overall evaluation. The relative
importance of different sources of evidence to the conclusions varies by pollutant or assessment,
as does the availability of different sources of evidence when making a causality determination.
In forming judgments of causality, NCEA scientists will also evaluate uncertainty in the
scientific evidence, considering issues such as generalizing results from a small number of
controlled human exposure subjects to the larger population; extrapolations of observed
pollutant-induced pathophysiological alterations from laboratory animals to humans;
4-21
-------
confounding by co-exposure to other ambient air pollutants, meteorological factors, or other
factors; the potential for effects to be due to exposure to air pollution mixtures; and the influence
of exposure measurement error on epidemiologic study findings. Judgments of causality also are
informed by the extent to which uncertainty in one line of evidence (e.g., potential copollutant
confounding in epidemiologic results) is addressed by another line of evidence (e.g., coherence
of effects observed in epidemiologic studies with experimental findings, mode of action
information). Thus, evidence integration is not a unidirectional process but occurs iteratively
within and across scientific disciplines and related outcomes.
A similar process is used for the integration of evidence and determination of causality
for welfare-related effects. For ecological effects this includes evaluating evidence relevant to
quantitative relationships between pollutant exposures and ecological effects. This also includes
reviewing concentration-response relationships and, to the extent possible, drawing conclusions
on the levels at which effects are observed. Also evaluated are O3 effects on biological levels of
organization from species to populations to biological communities and ecosystems. Both
laboratory and field studies (including field experiments and observational studies) can provide
useful data for causality determination. Integration of evidence for effects on climate draws upon
modeling and monitoring data as well as experimental approaches designed to characterize the
role of O3 in atmospheric processes. Generally, a causality determination is made based on many
lines of evidence that reinforce each other and are based on integrating evidence from both
previous and recent studies.
4.3.8 Quality Management
Within the EPA, Quality Management Plans (QMP) are developed to ensure that all
Agency materials meet a high standard for quality. NCEA participates in the Agency-wide
Quality Management System, which requires the development of a QMP. Implementation of the
NCEA QMP ensures that all data generated or used by NCEA scientists are "of the type and
quality needed and expected for their intended use" and that all information disseminated by
NCEA adheres to a high standard for quality including objectivity, utility, and integrity. Quality
assurance (QA) measures detailed in the QMP will be employed for the development of the ISA.
NCEA QA staff will be responsible for the review and approval of quality-related
documentation. NCEA scientists will be responsible for the evaluation of all inputs to the ISA,
including primary (new) and secondary (existing) data, to ensure their quality is appropriate for
their intended purpose. NCEA adheres to Data Quality Objectives, which identify the most
appropriate inputs to the science assessment and provide QA instruction for researchers citing
secondary information. The approaches utilized to search the literature and criteria applied to
select and evaluate studies were detailed in the two preceding subsections. Generally, NCEA
4-22
-------
scientists rely on scientific information found in peer-reviewed journal articles, books, and
government reports. The ISA also can include information that is integrated or summarized from
multiple sources to create new figures, tables, or summation, which is subject to rigorous quality
assurance measures to ensure their accuracy.
4.4 SPECIFIC SCIENCE ISSUES TO BE ADDRESSED IN THE ISA
The ISA will provide the scientific foundation for this NAAQS review process and
inform the consideration of whether it is appropriate to retain or revise the current primary and
secondary O3 NAAQS. Decisions on the specific content of the ISA will be guided by policy-
relevant questions that frame the entire NAAQS review as outlined in Chapter 3. Policy-relevant
questions for the ISA are related to two overarching issues: (1) the adequacy of the standard to
protect public health, and (2) reductions in uncertainties identified in the previous review or new
sources of uncertainties. The initial overarching policy-relevant question for the primary and
secondary standards concerns the adequacy of public health or public welfare protection afforded
by the standard. In considering this overarching question, the PA addresses a series of more
specific questions (sections 3.1.2 and 3.2.2). The more specific questions relate to the nature of
health and welfare effects attributable to O3; the populations, ecosystems or species particularly
at risk of such effects and the exposure concentrations of O3 associated with health and welfare
effects. Another question concerns whether uncertainties from the last review have been reduced
and/or whether new uncertainties have emerged. In the integrated synthesis and each of the
health and welfare effects appendices, the current ISA will evaluate uncertainties and limitations
in the scientific data, as described below.
In order to evaluate potential confounding by other ambient air pollutants in
epidemiologic studies, the ISA will examine whether epidemiologic associations with O3 are
observed in copollutant models. Copollutant models are the predominant method used in air
pollution epidemiology to estimate the effect of one pollutant controlling for a given
concentration of a copollutant. The ISA also will evaluate whether O3 has either interactions with
copollutants or joint effects in associations with health outcomes. The assessment of potential
confounding, interactions, or joint effects will draw upon results from health effects studies,
available information on copollutant interactions in the atmosphere that influence the spatial
distributions of O3 and copollutants, as well as information from experimental studies that
examine the health effects of O3 exposures alone and O3 in combination with other pollutants. In
the absence of these methods, the ISA will examine whether single-pollutant epidemiologic
associations with health effects in a given study differ between O3 and copollutants, and if
insights regarding potential copollutant confounding can be gained by examining the magnitude
of correlation between pollutants.
4-23
-------
The ISA will consider the strengths and limitations of various exposure assessment
methods. Monitoring data and model output will be used to characterize ambient air O3
concentrations used as surrogates for human exposures. Additionally, the ISA will evaluate the
strength of inference in epidemiologic studies by considering information such as the exposure
duration being examined, the extent of temporal and/or spatial variability in O3 in the study area,
the distribution of monitoring sites in the study area, the performance of exposure models used,
and time-activity patterns of the study population. The adequacy of exposure assessment in
epidemiologic studies will be considered in weighing the quality of evidence, and in turn,
forming causality determinations.
Epidemiologic evidence is unlikely to completely address the uncertainties mentioned
above. Any individual study is unlikely to evaluate all potentially correlated copollutants, and the
limitations of epidemiologic methods in separating effects of highly correlated pollutants or
separating the effects of more than two pollutants in the same model are well recognized. Thus,
coherence with other lines of evidence may strengthen inferences when there are uncertainties in
epidemiologic evidence due to copollutant confounding. Controlled human exposure and
toxicological studies that demonstrate similar effects at relevant O3 exposures may demonstrate
an independent effect of O3 exposure, provide coherence with epidemiologic evidence. Further,
experimental results may provide biological plausibility.
In the previous O3 review, a number of uncertainties were identified with respect to
quantitative relationships between O3 and effects on public welfare. Variation in O3 effects on
vegetation arises from the influence of co-occurring environmental stressors (e.g., drought,
nitrogen deposition), as well as from variation in O3 sensitivity at different vegetative growth
stages or between genotypes. The 2013 ISA identified uncertainties in the magnitude of O3
effects on climate, including the net radiative forcing due to changes in O3 concentrations and the
resulting surface temperature response. The ISA will evaluate the status of these uncertainties
and limitations in each of the welfare effects sections and this information will be used in the
development of causality determinations.
The ISA also will address a set of more specific policy-relevant questions related to the
available scientific evidence, as described in the following sections. These questions were
derived from the last O3 NAAQS review.
4.4.1 Causality Determinations from 2013 ISA
The causality determinations in the 2013 ISA, based on the causal framework and
integration of available evidence from previous and recent studies, were presented with a
summary of the available evidence at the end of the sections for each broad health and welfare
4-24
-------
effect category and in the integrative synthesis chapter at the beginning of the ISA (U.S. EPA,
2013).
In the 2013 ISA, for human health effects, the EPA concluded that the findings of
epidemiologic, controlled human exposure, and animal toxicological studies collectively
provided evidence of a "causal relationship" for short-term O3 exposures and respiratory effects.
In evaluating a broader range of health effects for O3, the 2013 ISA concluded there was
evidence of a "likely to be causal relationship" for long-term O3 exposures and respiratory
effects and for short-term O3 exposures and cardiovascular effects and mortality. Additionally,
there was evidence "suggestive of a causal relationship" for O3 exposures and other health
effects, including developmental and reproductive effects (e.g., low birth weight, infant
mortality) and central nervous system effects (e.g., cognitive development).
In the 2013 ISA, for welfare effects, the evidence indicated a "causal relationship"
between O3 exposure and visible foliar injury effects on vegetation, reduced vegetation growth,
reduced productivity in terrestrial ecosystems, reduced yield and quality of agricultural crops,
and alteration of below-ground biogeochemical cycles. The evidence indicated a "likely to be
causal relationship" for reduced carbon sequestration in terrestrial ecosystems, alteration of
terrestrial ecosystem water cycling and alteration of terrestrial community composition. For
climate there was a causal relationship between changes in tropospheric O3 concentration and
radiative forcing and likely to be a causal relationship between changes in tropospheric O3
concentration and effects on climate.
In the current review, specific science questions related to the causality determinations
that we plan to address include:
• Does the evidence base from recent studies contain new information to support or call into
question the causality determinations made for relationships between O3 exposure and
various health and welfare effects in the 2013 ISA?
• Is there new information to extend causality determinations to other ecological endpoints?
• Does new evidence confirm, extend, or call into question prior conclusions on the
biological plausibility for specific 03-related health effects?
• What is the strength of inference from epidemiologic studies based on the extent to which
they have:
o Examined exposure metrics that capture the spatial and/or temporal pattern of O3
in the study area?
o Assessed potential confounding by other pollutants and factors?
• What does the available information indicate with regard to changes in population health
status that may be associated with a decrease in ambient air O3 concentrations that might
inform causality determinations?
4-25
-------
4.4.2 Ambient Air Concentrations of O3
The ISA will present and evaluate relevant data, and summarize the current scientific
understanding concerning the sources and ambient air concentrations of O3 in the U.S. lower
troposphere and surface boundary layer. Ozone present in the lower troposphere is
predominantly formed through photochemical reaction involving reactive volatile organic
compounds and/or NOx as precursor gases. The discussion divides atmospheric O3 into two
classes: U.S. background O3 and non-background O3 (see section 4.3.4.1.1). Specific science
questions that we plan to address in the ISA include:
• What are the origins of U.S. background O3 concentrations, especially related to
international transport into the U.S., stratospheric exchange, and natural emissions from
biogenic sources, wildfires, and lightning? How well quantified are contributions from
these sources on overall tropospheric O3 concentrations?
• What modeling strategies have been used to estimate U.S. background O3 concentrations?
What are the sources of bias and uncertainty associated with the models used to estimate
U.S. background O3 concentrations? What observations or alternative estimates are
available that quantify U.S. background O3 concentrations and characterize its
spatiotemporal patterns?
• What data are available to characterize precursor emissions of non-background O3? How
does recent evidence contribute to what is known about the photochemical production of
non-background O3? How has modeling non-background O3 evolved since the last ISA?
Are there new models, or recent studies that have evaluated the validity of existing
models?
• Have methods for measuring non-background O3 substantively changed since the last
ISA? What are recent O3 concentrations and longitudinal trends in O3 concentrations?
4.4.3 Human Exposure
The ISA will evaluate methods for estimating exposure to ambient air O3, as well as the
ability to make inferences about personal exposure to ambient O3 when extrapolating from
ambient air concentration data, particularly in the context of interpreting results from
epidemiologic studies. The issues surrounding the ability to make inferences about personal
exposure differ by the exposure period of interest. Short-term exposure studies (i.e., exposures
ranging from hours up to weeks) examine how temporal variation in exposure is associated with
temporal variation in a health outcome while long-term exposure studies (i.e., exposures ranging
from months to years) typically examine how spatial variability of exposure is associated with
spatial variation in a health outcome averaged over time. Specific science questions related to
human exposure that we plan to address in the ISA include:
• What new developments have occurred with respect to chemical transport modeling of
short-term and long-term O3 concentrations for use in exposure assessment? How might
4-26
-------
modeling and satellite data supplement monitoring data for understanding human
exposures? What are the limitations of using modeling or satellite data in lieu of
monitoring data? What advancements have been made with respect to techniques for
fusing modeling, monitoring, and/or satellite data for assessing exposures to ambient air
O3? What are the uncertainties in data from chemical transport models and satellites at
the extremes of the concentration distribution, such as in high and low concentration
areas (e.g., near roadways, rural areas) and times?
• What are the errors and uncertainties associated with extrapolating from stationary O3
monitoring instruments to personal exposure to O3 of ambient air origin? Issues may arise
from instrument error in outdoor ambient air monitors, the use of fixed-site monitors for
estimating community concentrations across different spatial scales (e.g., neighborhood
scale, urban scale), spatial misalignment from using fixed-site monitors as a surrogate for
personal exposure to O3 of ambient air origin, and uncertainty in the time-activity
patterns of exposed individuals whose exposure is represented by fixed-site monitors.
• What new developments have been made in assessing and/or correcting the influence of
exposure measurement error on health effect estimates for epidemiologic studies of short-
term and long-term exposure? How do these methods reduce the uncertainty and/or bias
in the health effect estimates for O3 exposure?
4.4.4 Health Effects
In the 2013 ISA, the health effects evidence indicated that a "causal relationship exists"
for short-term exposures to O3 and respiratory effects, and a "likely to be causal relationship
exists" for long-term O3 exposures and respiratory effects and short-term O3 exposures and
cardiovascular effects and mortality. More limited evidence with a larger degree of uncertainty
formed the basis for the determinations for other health effects. The EPA will build on the
conclusions of the 2013 ISA by evaluating the newly available literature related to O3 exposures
and health effects, including, but not limited to respiratory, cardiovascular, nervous system,
reproductive and developmental effects, mortality, and cancer. Depending on data availability
and resources, other health effects may be evaluated.
The ISA will evaluate health effects that occur following both short- and long-term
exposures as examined in epidemiologic, controlled human exposure, and animal toxicological
studies. Efforts will be directed towards identifying the concentrations at which effects are
observed, particularly in potential at-risk lifestages and populations, and assessing the role of O3
within the broader mixture of ambient air pollutants. The discussion of health effects will be
integrated with relevant information on exposure, dosimetry and biological plausibility.
In the current review, specific science questions that we plan to address in consideration
of health effects associated with short- and long-term exposure to O3, include the following:
Short-Term Exposure
• What recent evidence is available to inform policy-relevant considerations of the O3
NAAQS (summarized in Chapter 3) for short-term O3 exposures and respiratory effects?
4-27
-------
Do recent controlled human exposure and toxicological studies continue to provide support
for or call into question prior conclusions on relationships between short-term O3
exposures and respiratory health effects? Do recent studies report O3-attributable effects at
lower O3 exposure concentrations or for different durations or patterns of exposure than
indicated by studies available in the last review?
How do results of recent studies expand understanding of the relationship between short-
term exposure to O3 and cardiovascular effects, such as ischemic heart disease, heart
failure, or vascular effects? Does recent evidence improve coherence across disciplines for
heart rate variability, blood pressure, and outcomes such as cardiovascular hospital
admissions or emergency department visits?
To what extent is short-term exposure to O3 related to or associated with the progression of
diabetes, other metabolic diseases, and/or to other endocrine system effects? To what
extent does the newly available evidence identify health outcomes related to or associated
with O3 exposures that were not previously identified?
Across the evaluated health effects, what new evidence is available on effects occurring
from exposures of different durations than indicated by the previously available evidence?
Long-Term Exposure
What new evidence is available to inform policy-relevant considerations of the O3 NAAQS
(summarized in Chapter 3) for long-term O3 exposures and respiratory effects? Do new
epidemiologic and toxicological studies continue to provide support for biologically
plausible relationships between long-term O3 exposures and respiratory health effects? Do
new studies report 03-attributable effects at lower O3 concentrations than indicated by
studies available in the last review?
To what extent do recent studies improve understanding of the relationships between long-
term O3 exposure and the development of asthma or to the impairment of lung
development? Do recent studies improve coherence across disciplines for respiratory
disease incidence, pulmonary inflammation and oxidative stress, and allergic responses?
To what extent do recent studies improve understanding of the relationship between O3
exposure and reproductive and developmental health outcomes, such as adverse birth
outcomes, fertility and pregnancy outcomes (e.g., infertility, sperm quality, preeclampsia,
gestational hypertension), or developmental outcomes (e.g., neurocognitive effects)? Are
there new studies linking exposures during critical windows of development to increased
risk of 03-related health effects later in life?
To what extent does new literature support or call into question the existence of a
biologically plausible relationship between long-term O3 exposures and nervous system
effects (e.g., cognitive decline and autism)?
How do results of recent studies expand our understanding of the relationship between
long-term O3 exposure and mortality? To what extent does the evidence indicate that long-
term exposure to O3 can increase the risk of respiratory-related mortality or other cause-
specific mortality?
To what extent is long-term exposure to O3 related to or associated with the development
of diabetes and other metabolic diseases, as well as to health effects in the endocrine
4-28
-------
system or other organ systems? To what extent are new health outcomes related to or
associated with O3 exposures?
Additional Science Considerations
• Do epidemiologic studies of mortality, hospital admissions, or emergency department
visits provide new information to improve our understanding of the potential heterogeneity
in effects assessed in U.S. multicity studies?
• How do the results of recent studies inform the shape of the concentration-response
relationship for O3 and various health outcomes (e.g., mortality, hospital admissions, etc.),
especially for exposures relevant to O3 concentrations near the current O3 NAAQS?
• What new evidence adds to the understanding of which lifestages and populations are at
increased risk of 03-related health effects?
• What new evidence informs conclusions regarding inter-individual variability in response
to O3 exposures?
• What is the relationship between short- and long-term exposures and 03-related health
effects? More specifically, across health effects, what new information is available to
delineate the effects of chronic exposure to lower concentrations versus acute, repeated
exposures to higher concentrations of O3?
• What is the nature of health effects in persons exposed to multipollutant mixtures that
contain O3 in comparison to exposure to O3 alone?
4.4.5 At-Risk Lifestages and Populations and Public Health Impact
The NAAQS are intended to protect public health with an adequate margin of safety,
including protection for the populations or lifestages potentially at increased risk for 03-related
health effects. Thus, the ISA will evaluate evidence for an array of factors that may contribute to
increased risk of 03-related health effects for various lifestages or populations (e.g., populations
with preexisting disease). The evaluation of recent evidence will build on the conclusions from
the 2013 ISA, where application of the at-risk framework77 to classify evidence demonstrated
that there was adequate evidence that children, older adults, people with pre-existing asthma,
people with certain genetic variants, people with nutritional deficiencies, and outdoor workers
are at increased risk of 03-related health effects. The ISA will evaluate recent evidence that
informs the identification of at-risk factors (e.g., lifestage, preexisting disease) in each of the
health appendices. Key considerations in characterizing the evidence include consistency of
findings for a factor within a discipline and, where available, coherence of the evidence across
77 In recent reviews, the term "at-risk" has been used to define populations and lifestages potentially at increased risk
of an air pollutant-related health effect (e.g., see 2013 O3 ISA and 2016 NOx ISA; U.S. EPA, 2013; U.S. EPA,
2016). At-risk populations can include those with intrinsic factors that make them more susceptible to pollutant-
related effects (e.g., pre-existing disease, genetic characteristics) or that increase pollutant dose (e.g., breathing
patterns), and extrinsic factors that could increase pollutant exposures (e.g., personal activity patterns) (U.S. EPA,
2016, pp. lxiiito lxiv).
4-29
-------
disciplines as well as biological plausibility. When evaluating evidence to inform the
identification of at-risk lifestages or populations, emphasis will be placed on the health effects
for which there is a causal or likely to be a causal relationship with exposure to O3. Specific
questions we plan to address include:
• What new evidence is available to further support or call into question the at-risk
determination made for lifestages or populations in the 2013 ISA?
• What new evidence is available regarding additional lifestages or populations (e.g., pre-
existing diseases such as diabetes) potentially at increased risk of an Cb-related health
effect?
• Is there new information that identifies a combination of factors (i.e., co-occurring) that
can lead to one lifestage or population being at greater risk compared to another?
4.4.6 Welfare Effects
In the 2013 ISA, the welfare effects evidence for O3 focused on effects on vegetation and
ecosystems, and the role of tropospheric O3 in climate change and supplemental shielding of
UV-B radiation. The EPA will build on the 2013 ISA by evaluating the newly available literature
related to O3 exposures and these welfare effects.
4.4.6.1 Ecological Effects
The ISA will evaluate the literature related to O3 exposures at levels of biological
organization from the organism to the ecosystem. Evidence from experimental (e.g. laboratory,
greenhouse, OTC, FACE) and field, gradient or modeling studies that address effects of O3 on
ecological endpoints will be considered to identify concentrations at which effects are observed
(Table 4-5). The focus will be on information necessary for interpretation of effects and on
newly available information since the last ISA.
4.4.6.1.1 Plant-level Effects
Ambient air O3 concentrations have long been known to cause foliar injury and decreased
growth and biomass accumulation in annual, perennial and woody plants, including agronomic
crops, annuals, shrubs grasses, and trees. In the 2013 ISA the evidence was sufficient to infer a
causal relationship between O3 exposure and endpoints on vegetation including, visible foliar
injury, reduced growth, and reduced yield and quality from individual plants that are agricultural
crop species. Evidence for foliar injury includes data from field, lab and chamber studies dating
back to the 1960's. Decreased growth at the plant scale has been well established for several
decades and may translate to damages at the stand and then ecosystem scales. In the current
review specific policy-relevant questions related to O3 effects on plant-level effects include the
following:
4-30
-------
• Is there any additional information on foliar injury or biomass growth in U.S. species
attributable to O3 in ambient air?
• Is there additional information on the factors influencing the relationship between O3 and
visible foliar injury?
• Is there additional information regarding a relationship between visible foliar injury and
growth?
• Is there any additional information on interspecies differences in responses to O3?
4.4.6.1.2 Ecosystem-level Effects
Effects at the individual plant level can result in changes in ecosystems such as
productivity, below-ground processes, carbon storage, water cycling and nutrient cycling. The
2013 ISA determined there was a causal relationship between O3 exposure and reduced
productivity. Results of long-term experiments provided evidence of the association of O3
exposure and reduced productivity at the ecosystem level of organization which were supported
by decreased plant growth and modeling studies. The 2013 ISA also determined there was a
causal relationship between O3 exposure and alteration of below-ground biogeochemical cycles
including altered carbon allocation to below-ground tissues; and altered rates of leaf and root
production, turnover, and decomposition. These shifts can affect overall carbon loss and nitrogen
loss from the ecosystem. Studies from the leaf and plant level provided biologically plausible
mechanisms and results from experimental studies consistently showed responses of below-
ground processes to O3 exposure. The 2013 ISA determined there was a likely causal
relationship between O3 exposure and reduced carbon sequestration. Evidence for that
conclusion was primarily from global and regional modeling simulations. The 2013 ISA
determined there was a likely causal relationship between O3 and alteration of terrestrial water
cycling. Alteration of stomatal functioning may affect water use in leaves, whole plants, and at
the watershed level based on field and modeling studies. In the current review specific policy-
relevant questions related to O3 effects on ecosystem processes include the following:
• What new information is available, including that for 03-related effects on ecosystem
services, on alteration of below-ground biogeochemical cycles, decreased productivity,
reduced carbon sequestration, and alteration of terrestrial ecosystem water cycling?
• Are there newly identified ecological endpoints or processes affected by O3?
4.4.6.1.3 Community Composition
Ozone exposure can lead to loss of sensitive species and alter community composition of
plants and microorganisms in some ecosystems. In the 2013 ISA the evidence was sufficient to
infer a likely causal relationship between O3 and alteration of terrestrial community composition.
Studies of the impact of O3 on species competition and community composition showed declines
in community composition of above-ground and below-ground communities. In the current
4-31
-------
review specific policy-relevant questions related to O3 effects on ecosystems include the
following:
• Is there additional evidence with respect to O3 effects on ecosystem structure and
terrestrial community composition
• Is there additional evidence with respect to O3 effects on other organisms such as insects
or other wildlife?
4.4.6.1.4 Air Quality Indices and Exposure-Response Relationships
Exposure indices are metrics that quantify exposure as it relates to measured plant
response (e.g., reduced growth). In the 2013 ISA, exposure indices that cumulated and
differentially weighted the higher hourly average concentrations and included the mid-level
values offered the most reliable approach for use in developing response functions and
comparing studies, as well as for defining future indices for vegetation protection. Exposure-
response relationships were available for several tree and crop species from a variety of
experiments. In the current review specific policy-relevant questions related to air quality indices
and exposure-response include the following:
• Are there new U.S. studies which use various O3 metrics to further characterize O3 effects
on plant foliar injury and/or growth?
• Are there new studies which improve the characterization of O3 exposure-response at the
local, regional and/or national scale for the effects determined to be causal or likely
causal? Which are the relevant exposure indices for such relationships?
4.4.6.2 Effects on Climate and UV-B Shielding Effects
The ISA will present information on how changes in tropospheric O3 might affect
radiative forcing, subsequent effects on climate endpoints such as surface air temperature, and
UV-B shielding. The focus will be on information necessary for interpretation of effects and on
newly available information since the last ISA. Specific questions include:
• What new information is available to decrease uncertainties in the magnitude of the
radiative forcing and climate response attributed to tropospheric O3?
• What new information is available on tropospheric O3 as an absorber of UV-B radiation?
• To what extent do we understand the independent effects of O3 on climate in the broader
context of other climate forcers, including copollutants and O3 precursors?
• What feedbacks affect the climate response to radiative perturbations from tropospheric O3
concentration changes?
• What recent advancements have been made in understanding O3 effects on regional climate
in the U.S.?
4-32
-------
4.5 SCIENTIFIC AND PUBLIC REVIEW
4.5.1 Peer Input Workshop
As an early step in development of the draft ISA, the EPA has held a preliminary peer-
input meeting. This meeting brought together subject matter experts from a variety of disciplines
to review initial draft materials for the ISA. This workshop spanned multiple days (October 29
and 31, November 1 and 5, 2018), covering a different topic area each day. This workshop
occurred prior to the integration of evidence across scientific disciplines and the consideration of
the collective body of evidence for the purposes of making causality determinations. Therefore,
the peer input review is different than what will be provided by the CASAC and the public
following the release of the completed draft ISA. During the peer input meeting, expert panelists
were asked to address the following overarching questions:
• Do the initial draft materials capture the key new studies from the peer-reviewed literature
that have been published since the completion of the 2013 O3 ISA? Are there additional
studies published since the 2013 O3 ISA that should be included?
• Are there specific issues that should be considered or highlighted that will be important
for integrating evidence across disciplines?
4.5.2 Peer Review
The EPA's Peer Review Handbook dictates the process for scientific peer review of all
EPA products (U.S. EPA, 2015d). Accordingly, a draft of the ISA will be made available for
review by the CASAC, as well as by the public. Availability of the draft document will be
announced in the Federal Register. The CASAC will review the draft ISA at a public meeting
that will be announced in the Federal Register. The EPA will consider comments, advice, and
recommendations received from the CASAC and from the public in revising the draft ISA
document. The EPA has established a public docket for the development of the ISA.78 After
appropriate revision based on comments received from the CASAC and the public, the final
document will be made available on the EPA website. A notice announcing the availability of the
final ISA will be published in the Federal Register.
78 The ISA docket can be accessed at www.regulations.gov using Docket ID number EPA-HQ-ORD-2018-0274.
4-33
-------
5 QUANTITATIVE RISK AND EXPOSURE
ASSESSMENTS
In NAAQS reviews, quantitative REAs79 are generally designed to assess human
exposure and health risk, as well as ecological exposures and risks to public welfare, for air
quality conditions associated with the existing standards, and as appropriate, for conditions
associated with potential alternative standards. The objective for such assessments is generally to
provide quantitative estimates of impacts that inform judgments on the public health and public
welfare significance of exposures likely to occur under air quality conditions reflective of the
current NAAQS, and, as appropriate, any alternative standards under consideration. Accordingly,
the assessments are also intended to provide a basis for judgments as to the extent of public
health and public welfare protection afforded by such standards.
In developing REAs in each NAAQS review, we draw upon the currently available health
effects evidence that is characterized in the ISA. This includes information on atmospheric
chemistry, air quality, human and environmental exposures, dosimetry and mode of action, and
information on health and welfare effects associated with exposures considered likely to occur
because of pollutant concentrations in ambient air. We additionally employ current methods and
tools to support the quantitative modeling and assessment.
The REAs commonly rely on a case study approach which involves quantitative analyses
focused on populations and pollutant concentrations in one or more specific geographic areas
under air quality conditions that just meet the existing standards (and alternatives as appropriate).
Reliance on this approach is intended to provide assessments of the air quality scenario(s) of
interest for a set of study areas and associated exposed at-risk populations and ecosystems that
will be informative to the EPA's consideration of potential exposures and risks that may be
associated with the stated air quality conditions. For example, we are interested in the exposure
and risk associated with air quality conditions that just meet the current standard(s); such
information is useful in interpreting the degree of protectiveness given by the current standard(s),
the adequacy of such standard(s), and the need to consider alternatives. Further, the REA
analyses employ a case study approach that addresses practical considerations, such as
employing a tractable scale and considering resource constraints, while providing estimates for
populations and geographic areas of interest and also having broader applicability (e.g., offering
risk perspective for similar study areas that were not assessed). Thus, REA analyses are not
79 While the term REA has in the past several NAAQS reviews referred to assessments presented in a stand-alone
REA document, in this review, we are also using this term, or the phrase "REA analyses" to simply refer to the
analyses which we intend to present in appendices or as supplemental materials to the PA.
5-1
-------
generally intended to provide a comprehensive national assessment of such conditions, nor are
they necessarily intended to provide such an assessment of existing air quality. Rather, the
purpose is to assess population exposure and risk for particular air quality conditions based on
currently available scientific information, modeling tools, and other technical information. As a
result, the REA can provide extended perspective on potential exposures and risks in geographic
areas across the U.S. not analyzed but with similarity in the attributes that primarily influence
exposures and risks, such as ambient air concentrations, population demographics, and the
degree of correlation in their spatial distributions.
In planning any REA analyses that may be appropriate for a new NAAQS review, we
first consider the analyses conducted in the last review and the extent to which they provided
important insights that were informative to the Agency's decision on the current standard.
Conclusions in this regard are generally influenced by an assessment of the uncertainties
associated with each type of analysis and the corresponding consideration of each type's relative
strength, as documented in the notice of the decision for the prior review and associated
assessment documents such as the PA and REA. In considering whether new analyses are
warranted for particular types of assessments, we evaluate the availability of new scientific
evidence and technical information in this review, as well as improved methods and tools, that
may provide support for conducting updates to address key limitations or uncertainties in
analyses from the last review, or to provide additional insight beyond those provided by the prior
REA. Thus, we focus on identifying the new analyses that are warranted in consideration of
factors such as those raised here, while also bearing in mind practical and logistical
considerations such as available resources and timeline for the review.
The purpose of this chapter is to briefly summarize the comprehensive, complex, and
resource-intensive quantitative health and welfare assessments completed in the last review of
the O3 NAAQS, giving attention to those analyses concluded to be most informative to the
decisions reached on the standards in that review. In considering the issues raised above, we
additionally summarize key uncertainties and limitations of the analyses conducted for the last
review and consider the extent to which newly available information, tools or methodologies
might address those areas. For example, the scope of any analyses for this review would be
informed by the new scientific information characterized in the upcoming ISA; recent air quality
data; the availability of improved data, methods, tools, and models that can be used to address
limitations and uncertainties from the last review; and any constraints on resources and the
review timeline. The goal is to focus on those analyses that may be particularly policy relevant
and informative to decision-making in this review and to identify the types of analyses for which
updates are warranted and will be conducted in this review (in contrast to, for example, other
5-2
-------
types of analyses for which the assessments presented in the 2014 REAs may remain
appropriately informative).
We are planning that the quantitative exposure and risk analyses newly developed in this
review will be presented in the draft PA, and to consider them along with any previously
conducted analyses that remain pertinent and informative to consideration of the adequacy of the
current standards (and alternative standards, as appropriate). We intend to provide associated
technical details for any new exposure and risk analyses in appendices or supplemental materials
for the PA, while analyses from the last review are documented in the 2014 REAs, 2014 PA, and
technical memos available in the O3 docket for the last review. Any quantitative assessments
newly developed in this review would then be made available for public comment and reviewed
by the CASAC in the context of the draft PA. Public comments and CASAC advice on such
REA-related analyses in the draft PA would be considered in finalizing analyses for presentation
in the final PA.
In this chapter, quantitative exposure and risk assessments for informing the primary
standard are discussed in section 5.1 and those pertaining to the secondary standard are discussed
in section 5.2. Both of those sections present overviews of the types of analyses performed in the
last review and highlight some considerations for analyses in this review.
5.1 ASSESSMENTS INFORMING REVIEW OF THE PRIMARY
STANDARD
In reviews of primary NAAQS, quantitative exposure and health risk assessments are
generally intended to inform consideration of key policy relevant questions (see section 3.1),
such as the following:
• What are the nature and magnitude of exposures and health risks associated with air
quality conditions just meeting the current standard?
• To what extent are the estimates of exposures and risks to at-risk populations associated
with air quality conditions just meeting the current standard reasonably judged important
from a public health perspective?
In considering exposure and risk estimates in this context, an accompanying consideration is:
• What are the important uncertainties associated with any risk/exposure estimates?
The types of analyses performed generally reflect the nature and strength of the evidence
in various aspects. For example, for the health effects pertaining to exposures associated with the
presence of the pollutant in ambient air, the availability and type of information from the health
effects literature on relationships between internal dose, exposure, or ambient air concentration
and health response influences the types of exposure assessment and risk characterization that
are performed. The health assessments focus on exposure metrics that are appropriate for effects
5-3
-------
of concern for the subject pollutant, and along with available ambient air concentration
measurements and model estimates, where appropriate, are used to generate estimates of
exposure. Consistent with the health risk approaches that have been used in NAAQS reviews
(illustrated in Figure 5-1), assessments of ambient air Cb-related health risks have been
conducted in past reviews (including the last review) based on two different types of risk
approaches. The first approach is based on relating areawide average ambient air concentrations
to results from air quality epidemiologic studies by linking ambient air quality concentrations
with concentration-response functions. The second approach is based on relating population
exposure estimates to results from controlled human exposure studies and employing either a
benchmark concentration or exposure-response (E-R) function-based approach to estimate risk.
Risk Assessment/
Characterization
Exposure-response and/or
comparison to health effect
benchmark concentrations
(e.g.. O3, NOj, SOj)
Ambient air concentration-
response and/or
comparison to health effect
benchmark concentrations
(e.g.. Oj, NO2, S02, PM)
Internal concentration-
response
(e.g., CO. Pb)
Air Quality Monitoring/
Modeling
(Estimates of ambient air
concentrations)
Dosimetry Modeling
(Estimates of internal
biomarker concentrations)
Exposure Modeling
(Estimates of inhalation
[and as relevant, other
routes] exposure
concentrations)
Figure 5-1. Summary of health risk assessment approaches that have been employed in
NAAQS reviews.
In the review of the primary O3 standard completed in 2015, the different types of
analyses that were performed varied in the extent to which they informed consideration of the
policy-relevant questions posed above. Accordingly, they also varied in the extent to which they
informed conclusions and judgments related to revision of the then-existing primary O3 standard.
For example, the EPA generally expressed higher confidence in the 2014 HREA results for
exposure-based analyses, which were based on evidence from controlled human exposure
studies, as compared to ITREA estimates derived from the ambient air concentrations and
epidemiologic study associations (2014 HREA, section 9.6; 80 FR 65316).80 These two types of
80 The 2015 decision notice recognized key uncertainties in utilizing the estimated air concentrations and
epidemiologic study relationships (often called epidemiologic-based risk estimates) with potentially important
implications for the Administrator's consideration of epidemiology-based risk estimates (80 FR 65316; 79 FR
75277-75279; 2014 HREA, sections 3.2.3.2 and 9.6). These included the heterogeneity in effect estimates
between locations, the potential for exposure measurement errors, and uncertainty in the interpretation of the
shape of concentration-response functions at lower 0$ concentrations, as well as uncertainties related to the public
5-4
-------
analyses are described below in sections 5.1.1.1 and 5.1.1.2, respectively. The roles of the
analyses in conclusions reached and judgments made in the 2015 O3 NAAQS review are
summarized in section 5.1.2, as are key uncertainties and limitations of the analyses, along with
considerations related to the availability of information, methods or tools in this review that may
address them.
5.1.1 Overview of Assessments in Last Review
The HREA completed for the last review included two types of analyses. The first type
was based on assessment of population exposure using exposure modeling (section 5.1.1.1),
while the second relied on relating ambient air concentrations to adverse health outcomes using
ambient air concentration-response functions drawn from epidemiologic studies (section 5.1.1.2).
Figure 5-2 illustrates the conceptual model for these types of assessments in the framework of
the traditional source to dose to health effects model.
health importance of increases in relatively low O3 concentrations following air quality adjustment. Additionally,
as noted in section 5.1.1.2 below, lower confidence was placed in the results of the epidemiologic-based
assessment of respiratory mortality risks associated with long-term O3 exposures in consideration of several
factors.
5-5
-------
, w
< UJ
CD H
0£ -
10%, 15%, 20%)
Incidence of Respiratory
Health Outcomes:
(e.g., hospital emergency
department [ED] visits,
hospital admissions [HA],
mortality)
Figure 5-2. Conceptual model for 2014 O3 health risk assessment. Solid lines indicate
processes included in the 2014 assessment.
The long-standing evidence base for Cb-related adverse health effects is built from a large
assemblage of controlled human exposure studies, laboratory animal research studies, and air
quality epidemiologic studies. Together, these health effect studies lead to the strongly supported
conclusion that Cb-related exposure causes respiratory effects (2013 ISA, section 6.2.9; 80 FR
65302). The controlled human exposure studies document the occurrence of an array of
respiratory effects in humans in a variety of exposure circumstances, and additionally, in
combination with the laboratory animal research studies, inform our understanding of the mode
of action for Cb-attributable effects. The air quality epidemiologic studies provide additional
5-6
-------
support for the causal conclusion regarding effects of O3 in ambient air (2013 ISA, section
6.2.9).
The quantitative characterizations of health risk or potential risk for which the support in
the evidence has been strongest are those based on the exposure-based risk analyses, including
the analysis used in the last three O3 NAAQS reviews that involves the comparison of estimated
population-based O3 exposures experienced while at elevated exertion81 to benchmark
concentrations drawn from the controlled human exposure studies. A second set of exposure-
based risk analyses performed for the last three O3 reviews, has been those that employ a lung
function risk estimation approach that also draws on results of the controlled human exposure
studies. Another type of analysis that has been used is a risk approach based on ambient air
concentration-response functions from air quality epidemiologic studies. This approach was also
employed in the last two O3 NAAQS reviews (e.g., to estimate risk for various health outcomes,
such as hospital admissions), with a recognition of the uncertainties associated with the
quantitative concentration-response functions used in that approach. In initial planning for the
current review, we consider support for both types of health risk approaches (i.e., exposure-based
and air quality epidemiologic-based), evaluating the extent to which the information newly
available in this review provides support for developing updated or enhanced analyses that
would substantially improve the utility of risk estimates for informing the current review.
In the 2014 HREA, the two exposure-based risk analyses were performed in a set of 15
urban study areas and the air quality epidemiologic-based risk analyses were performed for a
subset of those areas.82 Both approaches were performed for five different air quality scenarios:
unadjusted air quality conditions, air quality adjusted to just meet the then-existing standard (75
ppb O3 as a 3-year average of annual fourth highest daily maximum 8-hour average
concentrations), and air quality adjusted to just meet potential alternative standards with levels of
70, 65 and 60 ppb.83 The scenarios were based on air quality representing two 3-year periods:
2006-2008 and 2008-2010.
81 As summarized in section 3.1 above, the focus on exposures while at elevated exertion reflects the evidence from
controlled human exposure studies in which exposures to O3 concentrations of a magnitude relevant to those
occurring in ambient air have only been shown to result in respiratory effects if the ventilation rates of people in
the exposed populations are raised to a sufficient degree, such as through physical exertion (2013 ISA, section
6.2.1.1).
82 The 15 urban study areas assessed were Atlanta, Baltimore, Boston, Chicago, Cleveland, Dallas, Denver, Detroit,
Houston, Los Angeles, New York, Philadelphia, Sacramento, St. Louis, and Washington, DC. The three not
included in the epidemiologic-based assessment were Chicago, Dallas, and Washington, DC.
83 These scenarios reflect air quality with design values - 8-hour values using the existing form of the NAAQS -
that meet the level of the current or potential alternative standards. These simulations are illustrative and do not
reflect any consideration of specific control programs designed to meet the specified standards. Further, these
simulations do not represent predictions of when, whether, or how areas might meet the specified standards.
5-7
-------
For the air quality scenarios that used adjusted air quality, ambient air O3 concentrations
that would just meet the then-current and potential alternative standards were estimated using a
photochemical model-based adjustment approach (2014 HREA, Chapter 4). This approach
employed the Community Multiscale Air Quality Model version 4.7.1 (CMAQv4.7.1)
instrumented with the higher order decoupled direct method (CMAQ-HDDM).84 The CMAQ-
HDDM was used to estimate sensitivities85 of O3 concentrations to changes in precursor
emissions; using this approach, we estimated hourly O3 concentrations at each monitor location
resulting from reductions in U.S. anthropogenic precursor emissions (i.e., NOx, VOC).86 This
approach to adjusting air quality reflects the physical and chemical atmospheric processes that
influence O3 concentrations in ambient air (2014 HREA, Chapter 4).87'88 For the exposure-based
analyses, the adjusted air quality for census tracts comprising each study area was derived from
the adjusted estimates at the ambient air monitor locations using the Voronoi Neighbor
Averaging (VNA) spatial interpolation technique (2014 HREA, Chapter 4). For the air quality
epidemiologic-based analyses, areawide average concentrations were developed from the
adjusted concentrations at the ambient air monitoring sites in each study area.
84 Details on model set-up, configuration, and input data are provided in 2014 HREA, Appendix 4B.
85 Sensitivities of O3 refer to predicted incremental changes in O3 concentrations in response to incremental changes
in emissions. The "higher order" aspect of the HDDM tool refers to the capability of capturing nonlinear response
curves.
86 Exposure and risk analyses for most of the urban study areas focus on reducing U.S. anthropogenic NOx
emissions alone. The exceptions are Chicago and Denver. Exposure and risk analyses for Chicago and Denver are
based on reductions in emissions of both NOx and VOC (2014 HREA, section 4.3.3.1; Appendix 4D).
87 Compared to the statistical approaches that have been used in the past (e.g., a quadratic equation used in the 2007
REA to adjust high concentrations downwards at a greater rate than lower concentrations), the photochemical
model adjustment approach provides more realistic estimates of the spatial and temporal responses of O3 to
reductions in precursor emissions. Because NOx in ambient air can contribute to both the formation and the
destruction of O3 (2014 HREA, Chapter 4), the response of ambient air O3 concentrations to reductions in NOx
emissions is more variable than indicated by the previously used quadratic adjustment. This improved approach to
adjusting O3 air quality is consistent with recommendations from the National Research Council of the National
Academies of Sciences (NRC, 2008). In addition, the CASAC strongly supported the new approach as an
improvement and endorsed the way it was utilized in the HREA, stating that "the quadratic rollback approach has
been replaced by a scientifically more valid Higher-order Decoupled Direct Method (HDDM)" and that "[t]he
replacement of the quadratic rollback procedure by the HDDM procedure is important and supported by the
CASAC" (Frey, 2014a, pp. 1 and 3).
88 Within urban study areas, the model-based air quality adjustments show reductions in the 03 levels at the upper
ends of ambient air concentrations and increases in the O3 levels at the lower ends of those distributions (2014
HREA, section 4.3.3.2, Figures 4-9 and 4-10). It is important to note that sensitivity analyses in the HREA
indicate that the increases in low O3 concentrations are smaller when NOx and VOC emissions are reduced
together than when only NOx emissions are reduced (2014 HREA, Appendix 4-D, section 4.7). Seasonal means
of daily O3 concentrations generally exhibit only modest changes upon model adjustment, reflecting the seasonal
balance between daily decreases in relatively higher concentrations and increases in relatively lower
concentrations (2014 HREA, Figures 4-9 and 4-10).
5-8
-------
5.1.1.1 Exposure-based Risk Analyses
As noted above, two exposure-based risk analyses were performed for the 2014 HREA in
the last review: one involving comparison of population exposures, while at elevated exertion, to
benchmark concentrations, and the second involving estimated population occurrences of
ambient air Cb-related lung function decrements (Figure 5-3). The exposure-to-benchmark
comparison characterizes the extent to which individuals in at-risk populations could experience
exposures of concern (i.e., concentrations at or above specific benchmarks while at moderate or
greater exertion levels) while engaging in their daily activities in study areas with air quality
adjusted to just meet the current and alternative O3 standards. The lung function risk analysis
provides estimates of the extent to which populations in such areas could experience decrements
in lung function. For the former, results were characterized using three benchmark
concentrations (60, 70, and 80 ppb O3), exposures to which in controlled human exposure studies
yielded different occurrences and severity of respiratory effects in the human subjects (2014
HREA, section 3.2). Similarly, based on the range of health effects considered clinically relevant
and the potential for varied responses in healthy individuals versus people with asthma, the lung
function risk analysis reported estimates for risk of lung function decrement at or above three
different magnitudes, i.e., FEVi reductions of at least 10%, 15%, and 20% (2014 HREA, section
6.2.1).
The risk analysis involving comparison of 8-hour average exposure concentrations that
coincide with an 8-hour average elevated ventilation rates to benchmark concentrations (section
5.1.1.1.1) provides perspective on the extent to which air quality adjusted to just meet different
standards could be associated with discrete exposures to O3 concentrations reported to result in
respiratory effects. For example, estimates of such exposures can provide a sense of the potential
for 03-related effects in the exposed population, including effects for which we do not have E-R
functions that could be used in quantitative risk analyses (e.g., airway inflammation). The
exposure benchmark analysis differs from the second exposure-based risk analysis which
estimates the population incidence of days with lung function decrements of magnitudes of
interest. In the lung function risk analysis (section 5.1.1.1.2), the time-series of exposures and
ventilation rates (rather than 8-hour average exposures and 8-hour average ventilation rates) for
each modeled individual is used to estimate the associated occurrence of lung function
decrements in that simulated individual.
5-9
-------
Hourly concentrations at monitoring sites
Adjustment factors
Hourly concentrations at monitoring site locations for different
AQ scenarios (just meeting existing and alternative standards)
a
I Voronoi Neighbor Averaging (VNA) Interpolation I
Hourly concentrations at census tracts
Population counts
of 8-hour daily
maximum 03
exposures at
elevated ventilation
co
Time series of03
exposure events
(concentrations and
ventilation rates) for
each individual
Population counts of 8-
hour daily maxinum 0:
exposures atelevated
ventilation
Q.
CO
Exposure-
Response
(E-R)
Function
MSS-FE^
Lung Function
Risk Model
Photochemical
Air Quality
Modeling
Health-Based
Benchmark
Concentrations
Ambient Air
Monitoring Data (hourly concentrations)
Controlled Human
Exposure Data
(exposures involving
moderate/greater
exertion)
Exposure Modeling (APEX)
(exposure concentrations and ventilation rate for each individual's exposure events)
Output: Number and percent of simulated at-risk
population groups estimated to experience 1 or
more days with lung function responses
(FEV1 >10%, 15% and 20%)
Lung Function Risk
Output: Number and percent of at-risk populations
at moderate or greater exertion estimated to be
exposed to daily maximum 03 concentrations that
exceed benchmark concentrations
Exposures (atelevated exertion)
at or above Benchmarks
Figure 5-3. Analytical approach for exposure-based risk analyses. Dashed lines and gray
box indicate the sole lung function risk approach used prior to 2014 HREA.
The 2014 HREA derived results for both types of exposure-based analysis for a set of
populations in the 15 study areas under the specified conditions for each of the air quality
scenarios. Population-based exposures used for analyses in the 2014 HREA were estimated using
the Air Pollutants Exposure (APEX) model.89 The APEX model is a probabilistic model that
89 Exposure modeling has been employed in the past several reviews of the O3 NAAQS, as well as reviews of the
primary NAAQS for sulfur oxides, oxides of nitrogen, and carbon monoxide (U.S. EPA, 2008, 2009, 2010,
2018e). In the absence of large-scale exposure studies that encompass the general population, as well as at-risk
5-10
-------
simulates a large number of randomly sampled individuals residing within a given study area
(i.e., 50,000 to 200,000 people, depending on the simulated study group). U.S. Census
demographic data are used by APEX, typically at a census tract level, to weight the population
distribution within the geographic area and best represent area-wide population exposures. The
APEX model simulates the movement of individuals through time and space by accounting for
the places they may visit and the activities they may perform, and then estimates their time-series
of O3 exposures occurring within indoor, outdoor, and in-vehicle microenvironments (2014
HREA, section 5.1.3). By incorporating individual activity patterns, the model estimates physical
exertion associated with each exposure event.90 This aspect of the exposure modeling is critical
in assessing exposure, ventilation rate, intake dose, and estimated health risk for ambient air
concentrations of O3.
The APEX model accounts for the most important factors that contribute to human
exposure to O3 from ambient air, including the temporal and spatial distributions of people and
ambient air O3 concentrations throughout a study area, the variation of ambient air-related O3
concentrations within various microenvironments in which people conduct their daily activities,
and the effects of activities involving different levels of exertion on breathing rate (or ventilation
rate) for the exposed individuals of different sex, age, and body mass in the study area (2014
HREA, section 5.1.3). To the extent spatial and/or temporal patterns of ambient air O3
concentrations are modified by the air quality adjustment as discussed above, exposure estimates
reflect population exposures to those modified patterns of ambient air O3 concentrations.
To represent personal time-location-activity patterns of simulated individuals, the APEX
model draws from the consolidated human activity database (CHAD) developed and maintained
by the EPA (McCurdy et al., 2000; U.S. EPA, 2017).91 The activity patterns of individuals are an
important determinant of their exposure due to the influence of exposure concentration, event
duration, and ventilation rate (2013 ISA, section 4.4.1). Because of variation in O3
concentrations among the various microenvironments in which individuals are active, the amount
of time spent in each location, as well as the exertion level of the activity performed, will
populations, modeling is the preferred approach to estimating exposures to O3. Additional information on APEX
can be found at: https://www.epa.gov/fera/human-exposure-modeling-air-pollutants-exposure-model.
90 An exposure event occurs when a simulated individual inhabits a microenvironment for a specified time, while
engaged at a constant exertion level and experiencing a particular pollutant concentration. If the
microenvironmental concentration and/or activity /activity level changes, a new exposure event occurs (McCurdy
and Graham, 2003).
91 The CHAD is comprised of data from several surveys that collected activity pattern data at city, state, and national
levels. Included are personal attributes of survey participants (e.g., age, sex), the locations visited and activities
performed by survey participants throughout a day, and the time-of-day activities occurred and their duration.
Additional information is available at: https://www.epa.gov/healthresearch/consolidated-human-activity-
database-chad-use-human-exposure-and-health-studies-and
5-11
-------
influence an individual's exposure to O3 from ambient air and potential for adverse health
effects. Activity patterns vary both among and within individuals, resulting in corresponding
variations in exposure across a population and over time (2013 ISA, section 4.4.1). For each
exposure event, APEX tracks activity, ventilation rate, exposure concentration, and duration. The
time-series of exposure events serve as the basis for exposure metrics of interest, such as the
daily maximum 8-hour exposure. Development of the two exposure-based metrics derived for
the 2014 HREA (comparison to benchmarks and lung function risk) is summarized in the
subsections below.
5.1.1.1.1 Benchmark Comparison
In the comparison-to-benchmarks analysis for the last review, the percent and number of
individuals in the study area populations expected to experience one or more days with an
exposure at or above benchmark concentrations, while at specified exertion levels, were
estimated (2014 HREA, chapter 5). As summarized in section 3.1 above, the benchmark
concentrations for this analysis (60, 70, and 80 ppb O3) were established based on a set of
controlled human exposure studies of healthy adults engaged in moderate or greater exertion,
while exposed to those concentrations (2013 ISA, section 6.2; 2014 PA, section 3.1.2.1). These
studies employed a 6.6-hour quasi-continuous exposure during which subjects participated in
five 50-minute exercise periods, each followed by 10-minute rest periods, with a 35-minute
lunch period after the third hour (e.g., Folinsbee et al., 1988 and Schelegle et al., 2009). The
lowest benchmark, 60 ppb, represents the lowest O3 exposure concentration, as a time-weighted
average, for which these controlled human exposure studies have reported respiratory effects. At
this concentration, there is evidence of a statistically significant decrease in lung function and
increase in airway inflammation (Brown et al., 2008; Adams, 2006). Exposure to approximately
70 ppb92 averaged over a similar time resulted in larger lung function decrements (and greater
prevalence of decrements) than was observed for 60 ppb, as well as an increase in prevalence of
respiratory symptoms. In such studies, exposures of 80 ppb O3, as a time-weighted average,
resulted in larger lung function decrements than following exposures to 60 or 70 ppb, in addition
to an increase in airway inflammation, increased respiratory symptoms, increased airway
responsiveness, and decreased resistance to other respiratory effects (section 3.1.2.1, above).
For the 2014 REA, population exposures were estimated for four study groups: all
school-age children (ages 5 to 18), school-age children with asthma, adults with asthma (ages 19
to 95), and all older adults (ages 65 to 95) in each of the 15 urban study areas (2014 HREA,
section 5.2.5). The results given primary attention in the review were those for school-age
92 The study on which the 70 ppb benchmark concentration is based, Schelegle et al. (2009), reported that the actual
mean exposure concentration was 72 ppb.
5-12
-------
children (ages 5-18), including school-age children with asthma,93 both of which were identified
as key at-risk populations in the ISA (2014 PA, section 3.1.5). The percentages of children
estimated to experience exposures at or above benchmarks are considerably larger than the
percentages estimated for adult populations (2014 HREA, section 5.3.2 and Figures 5-5 to 5-8).
The larger benchmark exposure estimates for children are due primarily to the larger percentage
of children estimated to spend an extended period of time being physically active outdoors
during times of day when O3 concentrations are highest compared to other population study
groups (2014 HREA, sections 5.3.2 and 5.4.1).
In estimating the exposures used for comparison to benchmark concentrations, the APEX
model averages exposures over a duration of interest. In addition, the model averages the
ventilation rate (Ve) for the exposed individual (based on the activities performed) over that
exact same period. This can be done because APEX simultaneously estimates Ve and exposure
concentration for every individual's time-series of exposure events. For the exposure duration of
interest (e.g., 5 minutes, 1 hour, or 8 hours), the model then derives and outputs the daily
maximum average Ve (and hence an equivalent ventilation rate or EVR)94 and exposure
concentration for the specified duration for each simulated individual. The model produces
summary tables based on comparison to the specified benchmark concentrations. The averaging
time and EVR used in the 2014 HREA - 8-hour average and 13 L/min-m2 - reflect parameter
values for the exposure assessments performed for the last three O3 NAAQs reviews (2014
HREA; U.S. EPA, 2007; Whitfield, 1996). Additional details on this analysis are provided in
Chapter 5 and the associated appendices of the 2014 HREA.
5.1.1.1.2 Lung Function Risk Assessment
In the 2014 HREA, risk of lung function decrements in terms of FEVi reductions of at
least 10%, 15% and 20% was estimated using two different approaches.95 The primary estimates
were based on a new approach that estimates FEVi responses for simulated individuals
associated with short-term exposures to O3 (McDonnell, Stewart, and Smith, 2007, 2010;
93 In terms of the percentage of the exposed population experiencing days at or above the benchmark concentrations,
the estimates for all children and children with asthma are virtually indistinguishable (2014 HREA, Chapter 5).
This is because HREA analyses indicate that activity data (i.e., time spent outdoors, exertion level) for people
with asthma are generally similar to people not having asthma (2014 HREA, Appendix 5G, Tables 5G-2 to 5G-5).
94 To reasonably extrapolate the ventilation rate of the controlled human study subjects (i.e., adults having a
specified body size and related lung capacity), who were engaging in quasi-continuous exercise during the study
period, to individuals having varying body sizes (e.g., children with smaller size and related lung capacity), an
equivalent ventilation rate (EVR) was calculated by normalizing the ventilation rate (L/min) by body surface area
(m2).
95 Both approaches to estimating lung function risk have been implemented in the air pollution exposure model
APEX (U.S. EPA, 2012a,b).
5-13
-------
McDonnell et al., 2012). This approach (termed here, the McDonnell-Stewart-Smith [MSS]-
FEVi model) uses the time-series of O3 exposure, corresponding ventilation rates, and a few
other influential personal attributes (e.g., age, body surface area) for each APEX simulated
individual to estimate their personal time-series of ambient air Cb-related FEVi reductions,
effectively utilizing an individual-based approach to estimate lung function risk. When selecting
for the daily maximum FEVi reduction for each person and aggregating across individuals,
APEX estimates the percent and number of people at risk, i.e., those experiencing FEVi
reductions of at least 10%, 15% and 20%, in a study area.
The 2014 HREA also provided lung function risk estimates following the methodology
used in the previous reviews which employs a simpler, population-based E-R function approach
to estimate the percent and number of people at risk in a study area (Whitfield et al, 1996; U.S.
EPA, 2007; U.S. EPA, 2014).96 This approach uses a Bayesian Markov Chain Monte Carlo
approach to develop probabilistic E-R functions to estimate the probability of Cb-related lung
function decrements (U.S. EPA, 2007). These E-R functions were then combined with the APEX
estimated population distribution of 8-hr maximum exposures for people at or above moderate
exertion (>13 L/min-m2 body surface area) to estimate the number of people expected to
experience lung function decrements. A key difference between the population-based E-R
function approach and the MSS-FEVi model is that the previous method estimates a population
distribution of FEVi reductions by using the population-based distribution of daily maximum 8-
hour average exposures while at moderate or greater exertion, where the MSS-FEVi model
estimates maximum FEVi reductions at the individual level using their continuous time-series of
exposures and concomitant breathing rates. The lung function risk estimates from the MSS-FEVi
model for simulated individuals are then aggregated to a population level (2014 HREA, section
6.2.2).
The MSS-FEVi model was used in the 2014 HREA to estimate exposure-based lung
function risk for three population groups: school-age children (5-18 years), young adults (19-35
years), and adults (aged 36-55 years) in all 15 urban study areas (2014 HREA, section 6.3). This
model (along with an age adjustment term) was developed based on data from controlled human
exposure study subjects aged 18 to 35 years and was used in the 2014 HREA to estimate lung
function risk for individuals as young as 5 years and as old as 55 years based on the 2013 ISA's
96 The 2014 HREA referred to this approach as the "population E-R model", the same general form of which was
used in the previous reviews.
5-14
-------
interpretation of the available information for these age groups (2014 HREA, section 6.2.4 and
Appendix 6E). 97'98
Additional details on this analysis are provided in Chapter 6 and the associated
appendices of the 2014 HREA.
5.1.1.2 Air Quality Epidemiologic Study-based Risk Analyses
Ozone-associated risk of various respiratory health outcomes and mortality were
estimated in twelve urban study areas using concentration-response (C-R) functions drawn from
the epidemiologic studies and "area-wide" average O3 concentrations, primarily in terms of
several daily air quality metrics (HREA, Table 7-2, Appendix 7A).99
The health outcomes for which Cb-associated risk was estimated using the daily air
quality metrics were: hospital admissions (HAs) for any respiratory outcome (Katsouyanni et al.,
2009; Linn et al., 2000); HAs for chronic lung disease, except asthma (Medina-Ramon et al.,
2006); emergency department (ED) visits for any respiratory outcome (Strickland et al., 2010,
Tolbert et al., 2017, Darrow et al., 2011); ED visits for asthma (Ito et al., 2007), incidence of
97 Assumptions made for extending the MSS-FEVi model to children younger than 18 years old were in part based
on a McDonnell et al. (1985) study of children aged 8 to 11 years old who experienced FEVi responses similar to
those observed in adults aged 18 to 35 years old when both groups were exposed to 120 ppb O3 at an EVR of 32-
35 L/min/m2. In addition, summer camp studies of school-aged children exposed outdoors in the Northeast also
showed 03-induced lung function changes similar in magnitude to those observed in controlled human exposure
studies using adults (e.g., Spektor et al., 1988; Spektor and Lippmann, 1991; see ISA section 6.2.1.2). Thus, for
children younger than 18 years old, we set the MSS-FEVi model age term to its highest value, the value used for
age 18.
98 Assumptions made for extending the MSS-FEVi model to adults older than 35 years old were based on evidence
indicating lung function responses to O3 exposure for adults older than 18 decrease with age until around age 55,
when responses are minimal. "Children, adolescents, and young adults appear, on average, to have nearly
equivalent spirometric responses to O3, but have greater responses than middle-aged and older adults when
similarly exposed to O3" (2013 ISA p. 6-21). "In healthy individuals, the fastest rate of decline in 03
responsiveness appears between the ages of 18 and 35 years (Passannante et al., 1998; Seal et al., 1996), more so
for females than males (Hazucha et al., 2003). During the middle age period (35-55 years), O3 sensitivity
continues to decline, but at a much lower rate. Beyond this age (>55 years), acute O3 exposure elicits minimal
spirometric changes" (2013 ISA p. 6-23). Based on the effect age has on responses observed for middle aged
adults, the model was set with a linearly decreasing response with increasing age for individuals between ages 36
to 55. For adults older than 55 years, the MSS-FEVi model age term was nullified (2014 HREA, sections 6.2.3.1
and 6.2.4; 2013 ISA, pp. 6-21 and 6-23). Simulations were still performed for adults older than 55 years;
however, there was minimal 03-induced lung function risk estimated for any of the air quality scenarios (HREA,
section 6.6).
99 The air quality metrics analyzed in the epidemiologic studies from which concentration-response functions were
taken are daily maximum 1-hour, daily maximum 8-hour average and daily 24-hour average concentrations, each
averaged across multiple monitors within study areas (2014 HREA, Appendix 7A, Table 7-2). The epidemiologic
studies use these ambient air quality metrics as surrogates for the spatial and temporal patterns of exposures in
study populations. Accordingly, the HREA applied the C-R functions obtained from the epidemiologic studies to
O3 concentrations in terms of these same ambient air metrics, as averaged across ambient air monitor locations in
each study area (2014 HREA, section 4.3.2.2). In the last review, we referred to these area-averaged
concentrations as "composite monitor" or "area-wide" O3 concentrations (e.g., 2014 PA, section 3.1.4; 2014
HREA, section4.3.2.2).
5-15
-------
asthma exacerbation-related chest tightness, shortness of breath or wheeze (Gent et al., 2003);
and mortality (Smith et al., 2009; Zanobetti and Schwartz, 2008; Jerrett et al., 2009100). Risk
estimates were derived for each health outcome for 12 urban study areas,101 or a subset thereof,
depending on the array of study areas included in the epidemiologic studies from which each C-
R function was drawn (HREA, Table 3-1).
These risk estimates were derived for air quality scenarios involving unadjusted air
quality from five years encompassing two 3-year periods (2006-2008, 2008-2010), model-
adjusted air quality just meeting the then-current standard (75 ppb), and three potential
alternative standards with alternative levels of 70, 65 and 60 ppb (2014 HREA, section 7.1.1).
The risk estimates were derived using the EPA's Environmental Benefits Mapping and Analysis
Program (BenMAP, version 4.0)102 for the specified health outcomes and locations with the C-R
function information from the studies cited for those outcomes and other relevant information for
the analysis. In presenting the results for the two 3-year periods assessed for each air quality
scenario, the HREA presented the annual risk estimates for one year with generally higher O3
concentrations (2007) and one year with generally lower O3 concentrations (2009). Additional
detail on these analyses is provided in section 3.7, Chapter 7 and the associated appendices of the
2014 HREA.
100 The C-R functions from Jerrett et al. (2009) related 03-associated respiratory mortality to seasonal averages of
daily max l-hour03.
101 The 12 urban areas were Atlanta, Baltimore, Boston, Cleveland, Denver, Detroit, Houston, Los Angeles, New
York, Philadelphia, Sacramento, and St. Louis.
102 BenMAP is a GIS-based computer program that draws upon a database of population, baseline
incidence/prevalence rates and effect coefficients to automate the calculation of health impacts (2014 HREA,
Chapter 7; U.S. EPA, 2013b). Additional information available at: https://www.epa.gov/benmap.
5-16
-------
CD
=3
CO
o
Q.
X
CO
=3
a
U)
h:
Hourly concentrations at monitoring sites
l<- Adjustment factors
Hourly concentrations at monitoring sites for different AQ
scenarios (just meeting existing and alternative standards)
Area-averaged daily 03 ambient air concentration metrics for
each study area and the different AQ scenarios
(just meeting existing and alternative standards)
Ambient Air
Monitoring Data (hourly concentrations)
Environmental Benefits Mapping and Analysis
Program (BenMAP)
Output: Incidence of hospital admissions,
emergency room visits, mortality
Photochemical
Air Quality
Modeling
Figure 5-4. Analytical approach for epidemiologic-based analyses.
5.1.2 Assessments for this Review
In the preceding section we have briefly summarized air quality, exposure and risk
analyses developed in the last review, noting key uncertainties or limitations associated with the
various assessments. The two sections below briefly summary key considerations in our planning
for assessments in the current review and our initial plans for such analyses.
5.1.2.1 Key Considerations
Our planning for assessments in this review will consider the uncertainties and limitations
that were highlighted during the last review to direct new analyses (if any) toward reducing such
uncertainties. This approach could potentially improve the utility of risk estimates in informing
the current review. As a first step, we consolidated the previously identified and characterized
uncertainties in the 2014 HREA,103 along with integrating any related discussions found in the
1113 The 2014 HREA (sections 5.5, 6.5.7, and 7.4.2) included a characterization of uncertainty in which elements
were judged regarding the potential for associated uncertainty to influence the risk estimates, with attention given
to those elements described to have the potential for a "moderate" or greater influence on risk estimates.
5-17
-------
2014 PA, the proposed and final rulemaking notices in the last review and consideration of
public comments in the 2015 response to comments document. Then, as in any review, we
considered the availability of new information, models, and tools since completion of the prior
assessment that have potential to better characterize key areas of uncertainty. Further, we
identified model/assessment aspects for which updates may reduce uncertainty or address
limitations, thus improving appropriateness of model outputs for their intended purposes. And
finally, following the review of this information, consideration was also given to new
uncertainties and limitations relevant for this review that were not explicitly identified in the
prior review documents. Detailed results of this characterization are given in Appendix 5A and
are summarized below.
Regarding the exposure-based analyses (as summarized in section 5.1.1.1 above), several
important uncertainties were identified in the last review, largely related to estimating ambient
air concentrations, estimating exposure concentrations (and concomitant exertion levels), and
modeling lung function decrements. In this review, there are newly available ambient air quality
data that better reflect concentrations at or near the current standard, updated emissions data and
air quality models, and updates to the exposure model to better estimate exposure-based risk
(Appendix 5A). Regarding the epidemiologic-based risk approach (as summarized in section
5.1.1.2 above), there were also several important uncertainties identified in the 2014 HREA.
However, it is expected that, for most if not all the recognized uncertainties except for those
related to the estimation of ambient air quality, there is unlikely to be newly available
information, models, or tools that would result in substantially improved risk estimates with
appreciably less uncertainty than those in the 2014 HREA (Appendix 5 A).104
5.1.2.2 Initial Plans for the Current Review
Based upon the findings presented in Appendix 5 A, we expect that any new quantitative
analyses in this review would focus on exposure-based analyses that can benefit from updated
information, models, and tools, ensuring that the new exposure and risk estimates are both
improved and appropriately targeted. Estimates from the exposure-based analyses, particularly
the comparison of maximum exposures to benchmark concentrations, were most informative to
the Administrator's decision in the last review (as summarized in section 3.1.2 above). This
largely reflected the EPA conclusion that "controlled human exposure studies provide the most
104 There are several important uncertainties associated with aspects of the 03 epidemiologic study-based approach
used in the last review for which information available in this review is not expected to appreciably affect, such
that they are expected to still have a moderate or greater impact on risk estimates. Such uncertainties include
those involving the correlation of population O3 exposures and ambient air monitor concentrations (including the
use of area wide average O3 concentrations) and uncertainties in the derived concentration-response functions
(e.g., the shape of concentration response curves at the lowest O3 concentrations). See Appendix 5A for details.
5-18
-------
certain evidence indicating the occurrence of health effects in humans following specific O3
exposures," and recognition that "effects reported in controlled human exposure studies are due
solely to O3 exposures, and interpretation of study results is not complicated by the presence of
co-occurring pollutants or pollutant mixtures (as is the case in epidemiologic studies)" (80 FR
65343, October 26, 2015). In the last review, the Administrator placed relatively less weight on
the air quality epidemiologic-based risk estimates, in recognition of an array of uncertainties,
including, for example, those related to exposure measurement error (80 FR 65346, October 26,
2015).
Therefore, based on preliminary consideration of the information cited here, including
consideration of the complex and extensive exposure and risk analyses performed for the 2014
REA, and given the expedited nature of this review, we are planning to focus new analyses in
this review on exposure-based risk analyses. This would reflect the emphasis given to these types
of analyses and the characterization of their uncertainties in the last review, along with the
expectation of having newly available information, models, and tools that could address such
uncertainties (Appendix 5A). Briefly, updates to these new analyses would build upon the
combined ambient monitor data, air quality modeling, and exposure/risk modeling approach used
in the last review, as follows.
Air quality updates
• Use recent ambient air monitoring data (e.g., 2015-2017) from US EPA's Air Quality
System (AQS) having unadjusted concentrations at or near the current standard. The prior
assessment used 2006-2010 air quality conditions, that in many study areas, had
unadjusted ambient monitor design values that were well above (>10 ppb) the then-
existing standard (HREA 2014, section 4.3.1.1, Table 4-1)
• Use photochemical modeling (CAMx version 6.5)105 to adjust ambient air concentrations
to just meet the air quality scenarios to be assessed
- Most recent CAMx model contains updated chemical mechanisms reflecting
understanding of important chemical pathways for ozone formation and
destruction
- Use recent year modeling inputs that reflect emissions, meteorology and
international transport (e.g., 2016). For example, on-road/non-road emissions
estimates have been substantially improved via use of the recent MOVES model
(2014) versus that previously used (MOBILE6)106
APEX exposure modeling updates
• Model input data
105 CAMx is the Comprehensive Air Quality Model with Extensions. Additional information and model download
can be found at http://www.camx.com/.
106 MOVES is the Motor Vehicle Emissions Simulator (see http://www.epa.gov/otaa/models/moves/index.htm') and
MOBILE6 is the Mobile Source Emission Factor Model, version 6 (see http://www.epa.gov/otaq/m6.htm)
5-19
-------
- Use most recent U.S. Census demographics and commuting data (i.e., 2010).
- Use meteorological data to reflect the assessment years studied (e.g., 2015-2017).
- Update estimated asthma prevalence for all census tracts in all study areas (e.g.,
2014-2017). Compared to prevalence used in the prior review (2006-2010),
asthma prevalence shows an increasing trend for children aged 10-17 years
(Akinbami et al., 2016) and adults through 2013-2014 (CDC, 2016)107
• Model algorithms, tools, and approaches
- Updated equations to estimate resting metabolic rate (RMR) and associated
ventilation rate (Ve). Compared to the equations previously used, the overall
statistical model fit and predictability has been improved (U.S. EPA, 2018
Appendix H)
- Improve matching of controlled human exposure study duration (6.6-hour) and
target Ve (EVR) to that estimated for simulated individuals and used for
benchmark comparisons and population-based E-R lung function risk estimates
(2014 HREA, section 5.2.8). Addressing this limitation would more appropriately
identify when simulated individuals experience benchmark concentrations of
interest.
- Use a probit link to fit the population-based E-R function used to estimate lung
function risk (rather than the combined logistic/linear model fit used previously).
Note that using logistic fit in E-R functions may overestimate the contribution of
risk attributed to low O3 exposure levels (U.S. EPA, 2018, section 4.6.2)
- Use new MSS-FEVi model (McDonnell et al., 2013) to estimate individual-based
lung function risk. In comparison to their previous model (McDonnell et al. 2012)
that was used in the prior review, McDonnell et al (2013) indicates that when
accounting for intra-subject variability in their new model yields an improved
model fit, however it is uncertain as to how that new model might affect risk
estimates.
Given the rapid timeline for this review, we would expect to focus on a streamlined set of
study areas and air quality scenarios compared to the expansive set assessed in the last review.
As in prior NAAQS reviews, a collection of study areas will be used to estimate population
exposures and risks, primarily considering those areas having ambient air O3 concentrations at or
near the current standard and comprising a large population (e.g., consolidated statistical areas
that include urban and suburban populations). In addition to consideration of the above
assessment updates, we plan to also update our characterization of uncertainties in the exposure-
based risk analyses (2014 HREA sections 5.4, 5.5, 6.4, and 6.5), largely informed by input data
evaluations, sensitivity analyses, and model performance evaluations, where possible.
We expect to consider in the PA other types of analyses from the last review that we do
not update in this review but that remain informative to this review when viewed in the context
107 https://www.cdc.gov/nchs/products/databriefs/db239.htm
5-20
-------
of the currently available evidence as characterized in the ISA and of updated air quality and
other analyses performed for this review. Accordingly, the PA for this review will describe and
discuss in detail, all risk and exposure analyses considered informative to this review. This
would include risk and exposure analyses newly developed in this review, as well as analyses
performed for the last review for which updated analyses were not performed. The draft PA will
be released for public comment and provided to the CASAC for their review. Advice and
comments received on this information will be considered in completing any updated risk and
exposure analyses and drawing on them in the policy evaluations presented in the final PA.
5.2 ASSESSMENTS INFORMING REVIEW OF THE SECONDARY
STANDARD
In reviews of secondary standards, quantitative exposure and risk assessments for welfare
effects are generally intended to inform consideration of key policy relevant questions (see
section 3.2), such as the following:
• What are the nature and magnitude of exposure- and risk-related estimates for welfare
effects associated with air quality conditions just meeting the current standard?
• To what extent are the estimates of exposures and risks associated with air quality
conditions just meeting the current standard reasonably judged important from a public
welfare perspective?
In considering exposure and risk estimates in this context, an accompanying important
consideration is:
• What are the important uncertainties associated with any risk/exposure estimates?
The types of analyses performed generally reflect the nature and strength of the evidence
in various aspects. For example, for the welfare effects pertaining to exposures associated with
the presence of the pollutant in ambient air, the availability of concentration-response, exposure-
response, or dose-response data from the welfare or ecological effects literature influences the
types of exposure and risk assessments that are performed. The assessments focus on exposure
metrics that are appropriate for effects of concern for the subject pollutant, with available
measurements and model estimates, where appropriate, used to generate estimates of exposure.
Several different exposure and risk analyses were conducted in the last review of the
secondary O3 standard. They included extensive air quality-based analyses, E-R function-based
risk analyses and some monitoring-based analyses in the 2014 WREA, as well as monitoring-
based analyses in the 2014 PA and in technical memoranda developed for the rulemaking
notices. Some types of these quantitative analyses were more informative to the 2015 decision on
the standard than others. Regarding the key policy-relevant questions above, the uncertainties
associated with results for some analyses limited their use in the Administrator's decision-
5-21
-------
making, while uncertainties regarding public welfare significance of the findings for other
analyses also limited such use of those analyses. In general, decision-making in the last review
placed greatest weight on estimates of cumulative exposures to vegetation based on ambient air
monitoring data and consideration of those estimates in light of E-R relationships for Cb-related
reduction in tree growth (summarized in section 3.2 above). These analyses supported the
Administrator's consideration of the potential for O3 effects on tree growth and productivity, as
well as its associated impacts on a range of ecosystem services, including forest ecosystem
productivity and community composition (80 FR 65292, October 26, 2015).
In the first section below (section 5.2.1), we provide an overview of the set of
assessments performed in the 2015 review. In the subsequent section (section 5.2.2), the relative
roles of the analyses in judgments made and conclusions reached in the 2015 review are
indicated, along with some key uncertainties and limitations of the analyses. In this section we
additionally consider information, methods or tools that may be newly available in this review
and that may address these uncertainties or limitations and thus provide for the development of
appreciably improved analyses that might be considered in this review, in combination with the
comprehensive analyses developed in the last review that remain informative to this review.
5.2.1 Overview of Assessments in Last Review
Quantitative analyses performed in the last review included both the extensive analyses
presented in the 2014 WREA and a smaller set of additional analyses, which were presented in
the 2014 PA or in technical memoranda to the rulemaking docket and that were described in the
notices of proposed and final rulemaking for the 2015 decision.
The full set of analyses presented in the 2014 WREA were generally related to two types
of effects on vegetation: (1) reduced growth in both trees (relative biomass loss or RBL) and
agricultural crops (relative yield loss or RYL), and (2) visible foliar injury (2014 PA; 2014
WREA; 80 FR 65374-65376, October 26, 2015; 79 FR 75324-75329, December 17, 2014).
Estimates of 03-related reduced growth in native trees and crops were based on combining E-R
functions described in the 2013 ISA for a set of tree and crop species with estimates of O3
exposures. These risk estimates were developed nationally, as well as in a small set of study
areas. The foliar injury related analyses were based on information from the U.S. Forest Service
(USFS) that included estimates of W126-based cumulative exposure108 and foliar injury scores at
108 The W126 index is described above in section 3.2.2.
5-22
-------
established biosites109 in 41 states in the contiguous U.S.110 Analyses of reduced growth, in both
trees and agricultural crops, are described in section 5.2.1.1 and assessments regarding visible
foliar injury are described in section 5.2.1.2.111 The additional analyses, which, in combination
with E-R functions described in the 2014 WREA and summarized in the 2014 PA, proved to be
more informative to the 2015 decision than the WREA analyses, are summarized in section
5.2.1.3 (2014 WREA, section 6.2; 2014 PA, section 5.2.1; 80 FR 65382-65410, October 26,
2015).
5.2.1.1 Growth-related Assessments
The growth-related assessments performed in the WREA included national-scale
analyses of tree growth (in terms of RBL) and crop yield (in terms of RYL), and also estimation,
at national or smaller scales, of associated changes in related ecosystem services, including
pollution removal, carbon sequestration or storage, and hydrology, as well as impacts on the
forestry and agriculture sectors of the economy. These assessments were conducted for several
air quality scenarios developed by adjusting air quality data using factors derived from regional
photochemical modeling to achieve reduced concentrations of O3 that just met the different
scenario objectives.
The air quality scenarios included one in which the then-current standard was just met
and additional scenarios in which the maximum 3-year W126-based exposure equaled 15, 11,
and 7 ppm-hrs.112 These scenarios were developed from ambient air monitoring data for 2006 to
2008 and adjustments based on model-predicted relationships between the response of O3
concentrations at each monitor location to reductions in NOx emissions for the associated NOAA
climate region. The adjustments were applied independently for each of the nine NOAA climate
regions in the U.S., such that the highest monitor in the region was adjusted to just meet the
109 Sampling sites in the National Forest Service's Forest Inventory and Analysis Forest Health Monitoring O3
biomonitoring program, called "biosites", are plots of land on which data are collected regarding the incidence
and severity of visible foliar injury on a variety of Ch-sensitive plant species. Biosite index scores are derived
from these data (2014 WREA, section 7.2.1).
110 Data were not available for several western states (Montana, Idaho, Wyoming, Nevada, Utah, Colorado, Arizona,
New Mexico, Oklahoma, and portions of Texas).
111 The 2014 WREA also presented several more descriptive exposure analyses where W126-based cumulative O3
exposure was estimated for different modeled air quality scenarios in areas of high fire or beetle infestation threat
(2014 WREA, sections 5.2.3 and 5.4).
112 The target for each scenario was judged to have been met when the 03 concentrations at the monitor location
with the highest concentrations equaled the target. For example, for the then-current standard scenario, the highest
monitor location had a fourth highest daily maximum 8-hour O3 concentration averaged over three years equal to
75 ppb. For the W126 scenario of 15 ppm-hrs, the target was met when the 3-year average W126 index value at
the monitor with the highest 3-year W126 value equaled 15 ppm-hrs. The development of the air quality scenarios
is further summarized in the final decision notice (80 FR 65374-65375, October 26, 2015) and Table 5-4 of the
2014 PA, and described in detail in Chapter 4 and Appendix 4A of the 2014 WREA.
5-23
-------
target for the air quality scenario, and other monitor sites not already at/below the target for the
scenario were adjusted by the same factor.113 Based on the adjusted concentrations at all monitor
sites, concentrations were derived for each 12 km by 12 km grid cell in a national-scale spatial
surface by applying the Voronoi Neighbor Averaging (VNA) spatial interpolation technique to
the monitor-location values. This step resulted in further reduction of the highest values in each
modeling region.114
Because the W126 estimates generated for the different air quality scenarios assessed are
inputs to the vegetation risk analyses for tree biomass and crop yield loss, and also used in some
components of the visible foliar injury assessments, limitations and uncertainties in the air
quality analyses, which are discussed in detail in the WREA and some of which are mentioned
here, were propagated into those analyses (2014 WREA, chapters 4 and 8, including section 8.5,
and Table 4-5). An important uncertainty in the analyses is the application of adjustments at the
regional-scale based on modeled emissions reductions in NOx that characterize only one
potential distribution of air quality across a region for situations when all monitor locations in a
region meet the then-current standard or the W126 cumulative exposure targets (2014 WREA,
section 4.3.4.2). The impact of the approach's broad regional reductions on O3 concentrations at
monitor locations that were already well below the target indicated an uncertainty with regard to
air quality expected from specific control strategies that might be implemented to meet a
particular target level (80 FR 65375, October 26, 2015).
An additional uncertainty related to the W126 index estimates in the national surfaces for
each air quality scenario, and to the estimates for the single-year surfaces used in the visible
foliar injury cumulative analysis, is associated with the creation of the national-scale spatial
surfaces of grid cells from the monitor-location O3 data.115 In general, spatial interpolation
techniques perform better in areas where the O3 monitoring network is denser. Therefore, the
113 The adjustment was based on the minimum percentage reduction in NOx emissions necessary to reduce O3
concentrations at all monitors within a region sufficiently to meet the target. This adjustment results in broad
regional reductions in O3 and includes reductions in O3 at some monitors that were already at or below the target
level (2014 WREA, sections 4.3.4.2 and 4.4).
114 This is seen when comparing the W126 index values from before and after the application of the VNA approach
to the then-existing standard scenario. After the adjustment of the monitor location concentrations such that the
highest location in each NOAA region just met the then-existing standard (using the model-based relationships),
the maximum 3-year average W126 values in the nine regions ranged from 18.9 ppm-hrs in the West region to
2.6 ppm-hrs in the Northeast region (2014 WREA, Table 4-3). After application of the VNA technique, however,
the highest 3-year average W126 values across the national surface grid cells, which were in the Southwest
region, were below 15 ppm-hrs (2014 WREA, Figure 4-7). Thus, using VNA interpolated W126 index values at
the centroid of every 12 km x 12 km grid cell compared to using W126 index values only at each monitor
location results in a lowering of the highest values in each region (80 FR 65374, October 26, 2018).
115 Some uncertainty is inherent in any approach to characterizing O3 air quality over broad geographic areas based
on concentrations at monitor locations.
5-24
-------
W126 index values estimated using this technique in those rural areas within the West,
Northwest, Southwest, and West North Central regions where there are few or no monitors (2014
WREA, Figure 2-1) are more uncertain than those estimated for areas with denser monitoring.
Further, as noted above, this interpolation method may underpredict the highest W126 exposure
index values in a region. Due to the important influence of higher exposures in determining risks
to plants and the potential for the interpolation step to dampen peak W126 index values, some
risk underestimation could have resulted.
The assessments related to tree growth relied on the species-specific E-R functions
referenced in section 3.2.2 above. For the air quality scenarios assessed, the species-specific E-R
functions were used to develop estimates of Cb-related RBL and associated effects on
productivity, carbon storage and associated ecosystem services (2014 WREA, Chapter 6). More
specifically, the WREA derived species-specific and weighted RBL estimates for grid cells
across the continental U.S. and summarized the estimates by counties, regions and Class I areas
and national parks (2014 WREA, section 6.2.1 and 6.8). Potential impacts on commercial timber
were also estimated (2014 WREA, section 6.3). Additional case study analyses estimated
impacts on carbon removal and pollutant removal in selected urban areas (2014 WREA, sections
6.6.2 and 6.7).
Relative biomass loss nationally (across all air quality surface grid cells) was estimated
for each of eleven studied species116 using the composite E-R functions for each species and
information on the distribution of those species across the U.S. (2014 WREA, section 6.2.1.3 and
Appendix 6A). These analyses provided estimates of per-species RBL, as well as median and
total RBL across resident species in the different air quality scenarios. The WREA also used the
E-R functions to estimate RBL across tree lifespans and the resulting changes in consumer and
producer/farmer economic surplus in the forestry and agriculture sectors of the economy. Case
studies in five urban areas provided comparisons across air quality scenarios of estimates for
urban tree pollutant removal and carbon storage or sequestration (2014 WREA, sections 6.6.2
and 6.7). The array of uncertainties associated with estimates from these tree RBL analyses,
including those associated with the air quality adjustment approach which contributed to a
potential for the air quality scenarios to underestimate the higher W126 index values and the
associated implications for the RBL estimates, are summarized in section 5.2.2 below.
The assessments of O3 impacts on agricultural crops relied on the robust E-R functions
established prior to the last review. For the different air quality scenarios, the WREA applied the
116 In consideration of CASAC advice regarding uncertainties associated with the E-R function derived for a twelfth
species, the eastern cottonwood, the WREA derived RBL and weighted RBL estimates separately, both with and
without the eastern cottonwood, with primary focus given to analyses that excluded cottonwood (Frey, 2014c, p.
10; 2014 WREA; 2014 PA; 79 FR 75234, December 17, 2014; 80 FR 65292, October 26, 2015).
5-25
-------
species-specific E-R functions to develop estimates of O3 impacts related to crop yield, including
annual yield losses, for 10 commodity crops grown in the U.S. and estimates of how these losses
might be expected to affect producer and consumer economic surpluses (2014 WREA, sections
6.2 and 6.5). The WREA derived estimates of crop RYL nationally and in a county-specific
analysis, relying on information regarding crop distribution (2014 WREA, section 6.5). As with
the tree analyses described above, the county analysis included estimates based on the median O3
response across the studied crop species (2014 WREA, section 6.5.1, Appendix 6B).
Overall effects on agricultural yields and producer and consumer surplus depend on the
ability of producers/farmers to substitute other crops that are less O3 sensitive, and the
responsiveness, or elasticity, of supply and demand (2014 WREA, section 6.5). The WREA
discusses multiple areas of uncertainty associated with the crop RYL estimates, including those
associated with the model-based adjustment methodology as well as those associated with the
projection of yield loss using the Forest and Agriculture Sector Optimization Model (with
greenhouse gases) at the estimated O3 concentrations (2014 WREA, Table 6-27, section 8.5) and
the lack of a role in the assessment for agricultural crop management practices which have
substantial influence on crop yield. Because the W126 index estimates generated in the air
quality scenarios are inputs to the vegetation risk analyses for crop yield loss, any uncertainties
in the air quality scenario estimation of W126 index values are propagated into those analyses
(2014 WREA, Table 6-27, section 8.5). Therefore, the air quality scenarios in the crop yield
analyses have the same uncertainties and limitations as in the biomass loss analyses (summarized
above), including those associated with the model-based adjustment approach (2014 WREA,
section 8.5).
5.2.1.2 Foliar Injury Assessments
The foliar injury assessments involved analysis of W126 cumulative exposure estimates
and foliar injury scores at USFS biosites for five years (2006-2010), and consideration of the
implications of the analysis with regard to risk of 03-related foliar injury in nationally protected
areas such as national parks (2014 WREA, Chapter 7; Smith and Murphy, 2015; 80 FR 65376,
65395-65396, October 26, 2015). In the biosite data analysis, the WREA used the biomonitoring
site data from the USFS FHM/FIA Network (USFS, 2011), associated soil moisture data during
the sample years, and national surfaces of ambient air O3 concentrations based on spatial
interpolation of monitoring data from 2006 to 2010117 in a cumulative analysis of the proportion
of biosite records with any visible foliar injury, as indicated by a nonzero biosite index score
117 Estimates of W126 were drawn from national-scale spatial surfaces of single-year, unadjusted W126 index values
created for each year from 2006 through 2010 using the VNA interpolation technique applied to the monitor
location index values for these years (2014 WREA, section 4.3.2, Appendix 4A).
5-26
-------
(2014 WREA, section 7.2). This analysis was done for all records together, and also for subsets
based on soil moisture conditions (normal, wet or dry).
In each cumulative analysis, the biosite records were ordered by W126 index and then,
moving from low to high W126 index, the records were cumulated into a progressively larger
dataset. With the addition of each new data point (composed of biosite index score and W126
index value for a biosite and year combination) to the cumulative dataset, the percentage of sites
with a nonzero biosite index score was derived and plotted versus the W126 index estimate for
the just added data point. This analysis was found to be appreciably affected by the larger
representation within the subset of the lower W126 conditions which are associated with a lower
occurrence or extent of foliar injury.118 Nearly two thirds of the dataset included records for
which the W126 index estimates are at or below 11 ppm-hrs (Smith and Murphy, 2015, Table 1).
In a technical memorandum prepared subsequent to the WREA, the same dataset was re-
presented in a different format to more directly consider what the data indicate with regard to a
relationship between O3 exposure in terms of W126 and foliar injury. This presentation indicated
the reduction in the occurrence (and severity) of visible foliar injury with decreasing exposures
across a range that extended from above 19 ppm-hrs to below 7 ppm-hrs (Smith and Murphy,
2015, Table 2).119
5.2.1.3 Additional Air Quality/Exposure and E-R Analyses
Additional analyses developed in the last review included two air quality and exposure
analyses, summarized below, and a separate tabular presentation involving tree and crop E-R
functions. The tabular presentation was based on the robust established E-R functions for growth
effects on tree seedlings and crops was developed for the 2014 PA (2014 PA, Appendix 5C).
118 The cumulative analysis for all sites indicated that (1) as the cumulative set of sites grows with addition of sites
with progressively higher W126 index values, the proportion of the dataset for which no foliar injury was
recorded changes (increases) noticeably prior to about 10 ppm-hrs, and (2) as the cumulative dataset grows still
larger with the addition of records for higher W126 index estimates, the proportion of the cumulative dataset with
no foliar injury remains relatively constant (2014 WREA, Figure 7-10). This "leveling off' (e.g., observed above
~10 ppm-hrs in the "all sites" analysis) likely reflects the counterbalancing of visible foliar injury occurrence at
the relatively fewer higher O3 sites by the larger representation within the subset of the lower W126 conditions
associated with which there is lower occurrence or extent of foliar injury (Smith and Murphy, 2015).
119 Criteria derived from the WREA cumulative analyses were used in two additional WREA analyses. The national-
scale screening-level assessment compared W126 index values estimated within 214 national parks using the
VNA technique described above for the individual years from 2006 to 2010 with benchmark criteria developed
from the biosite data analysis (2014 WREA, Appendix 7A and section 7.3). Separate case study analyses
described visits, as well as visitor uses and expenditures for three national parks, and the 3-year W126 index
estimates in those parks for the four air quality scenarios (2014 WREA, section 7.4). Uncertainties associated
with these analyses, included those associated with the W126 index estimates, are discussed in the WREA,
sections 7.5 and 8.5.3, and in WREA Table 7-24, and also summarized in the PA (2014 PA, section 6.3).
5-27
-------
This analysis presented the estimates of RBL120 (and RYL) at a range of W126-based exposure
levels for 11 tree species and 10 crop species, respectively (2014 PA, Tables 5C-1 and 5C-2).
Additionally, the median tree species RBL (or crop RYL) was presented for each W126 level
(2014 PA, Table 5C-3; 80 FR 65391 [Table 4], October 26, 2015). As summarized in section
3.2.2 above, the 2015 decision on the secondary standard included a focus on Cb-related RBL in
tree seedlings as a surrogate or proxy for the broader array of vegetation-related effects of
potential public welfare significance, which include effects on growth of individual sensitive
species and extend to ecosystem-level effects, such as community composition in natural forests,
particularly in protected public lands, as well as forest productivity (80 FR 65406, October 26,
2015).
The first of the two sets of air quality/exposure analyses included the development of
W126-based cumulative exposure estimates in Class I areas during 3-year periods that met the
then-current standard (75 ppb, in terms of the 3-year average of consecutive year fourth highest
daily maximum 8-hour averages). The second set of air quality/exposure analyses investigated
the W126-based cumulative exposure estimates for locations and time periods that met the then-
current and several potential alternative standards, in terms of 3-year averages of the fourth
highest daily maximum 8-hour average concentration. The former analysis was particularly
informative to the decision regarding the need to revise the then-current standard of 75 ppb (80
FR 65389-65390, October 26, 2015), while the second set of analyses informed the
Administrator's decision on the appropriate revision (80 FR 65403-65410, October 26, 2015).
The first set of air quality/exposure analyses, as presented and relied upon in the final
decision, was an update of an analysis initially presented in the 2014 PA (2014 PA, pp. 5-27 to
5-29). Based on air quality data for the period from 1998 to 2013, the analysis focused
consideration on 17 Class I areas,121 in which during one or more three-year periods the air
quality met the current standard and the three-year average W126 index value was at or above 15
ppm-hrs. The analysis that informed the 2015 decision was restricted to data for monitors sited in
or within 15 kilometers of a Class I area.122
120 These functions for RBL estimate the reduction in a year's growth as a percentage of that expected in the absence
of O3 (2013 ISA, section 9.6.2; 2014 WREA, section 6.2). In specifically evaluating exposure levels, in terms of
the W126 index the 2014 PA focused particularly on RBL estimates for the median across the 11 tree seedling
species for which robust E-R functions are available (80 FR 65391-65392 [Table 4], October 26, 2015; 2014
WREA, Appendix 5C, Table 5C-3).
121 For the four modeled air quality scenarios in the WREA, the WREA also derived detailed estimates of 3-year
W126-based exposures in a screening-level national park assessment and in three individual national parks. (2014
WREA, section 4.3.2, Appendix 4A). Limitations and uncertainties associated with the WREA air quality
adjustment approach limited their usefulness in the EPA's final decision-making.
122 The 15 km distance was selected as a natural breakpoint in distance of O3 monitoring sites from Class I areas and
as still providing similar surroundings to those occurring in the Class I area. We note that given the strict
5-28
-------
This analysis considered cumulative exposure estimates in Class I areas during times that
met the then-current standard in the context of such estimates associated with varying RBL
values for the median tree species derived using the robust E-R functions for RBL in seedlings of
11 tree species. The analysis gave particular weight to the W126 index values at or above 19
ppm-hrs, which were associated with a 6% median RBL, described as "unacceptably high" by
the CASAC (80 FR 65391-92, October 26, 2015; Frey, 2014c). In the analysis, the numbers of
areas, states and NOAA climatic regions, for which the 3-year W126 exposure index values
ranged at or above 19 ppm-hrs were tallied and characterized as to magnitude and variation
across the three years.
The second set of air quality/exposure analyses were focused on air quality monitoring
for O3 monitoring sites with complete data for the most recent 3-year period and also for periods
extending back to 2001.123 This set was comprised of several analyses of air quality that
considered relationships between 3-year W126 index based exposure estimates and the design
value for the then current standard (referred to as the "fourth-high" metric) (2014 PA, Chapter 2,
Appendix 2B and section 6.4; Wells, 2015). These analyses indicated that, depending on the
level, a standard of the then-current averaging time and form could be expected to control
cumulative seasonal O3 exposures to such that they may meet specific 3-year average W126
index values. The fourth-high and W126 metrics, and changes in the two metrics over the past
decade, were found to be highly correlated (2014 PA, section 6.4 and Appendix 2B; Wells,
2015).
These analyses were performed for two recent periods (2009-2011 and 2011-2013), as
well as extending back to 2001 (2014 PA, section 6.4; Wells, 2015). All NOAA climatic regions
in the contiguous U.S. were represented. These analyses illustrated the extent and magnitude of
W126-based exposures at monitoring sites meeting the then existing standard and alternate
standards, including the now-current standard of 70 ppb (2014 PA, section 6.4 and Appendix 2B;
Wells, 2015).
5.2.2 Assessments for this Review
In the preceding section we have briefly summarized air quality, exposure and risk
analyses developed in the last review, noting key uncertainties or limitations associated with the
restrictions on structures and access within some of these areas, it is common for monitors intended to collect data
pertaining to air quality in these types of areas to be sited outside their boundaries.
123 These analyses are summarized and discussed in sections IV.C.l.c, IV.C.2.d and IV.C.3 of the 2015 decision
notice and presented in detail in a technical memorandum to the rulemaking docket (80 FR 65292, 65400-65401,
65408-65409, October 26, 2015; Wells, 2015).
5-29
-------
various assessments. The two sections below briefly summary key considerations in our planning
for assessments in the current review and our initial plans for such analyses.
5.2.2.1 Key Considerations
In identifying the types of assessments to be developed or updated in this review, we give
particular attention to those types of analyses that formed the main foundation for conclusions in
the last review due to their relatively lesser uncertainty and fewer limitations. In so doing, we
consider the availability at this time of any new information that may address limitations or
uncertainties in any of the analyses from the last review. In this regard, we consider both the
analyses based on regional air quality modeling scenarios (e.g., as summarized in section 5.2.1.1)
and environmental exposure analyses based on air quality monitoring data (summarized in
sections 5.2.1.3).
As in any review, key considerations in planning risk and exposure analyses that may be
appropriate in this review include:
• Availability of new information (including more recent air quality patterns), models
and tools since completion of the prior assessment that have potential to address key
areas of uncertainty;
• Identification of model/assessment aspects for which updates are available and
feasible within the constraints of the timeline for the review that may reduce
uncertainty or address limitations, thus improving appropriateness of model outputs
for their intended purposes.
The analyses developed in the last review, along with key limitations and uncertainties, and also
the availability of relevant more recent information or updates, are briefly summarized in
Appendix 5B.
It is the analyses with relatively lesser uncertainty or fewer limitations regarding their
interpretation, which include those most informative in the last review, that we plan to emphasize
in considering analyses that may be appropriate to conduct for the current review. In so doing,
our objective is to focus on analyses for which there are updated models, tools, or data that
would have the potential to substantially improve the utility of risk estimates in informing the
current review. The matrix in Appendix 5B has informed these considerations. Based on this
approach, we expect to focus any new quantitative analyses in this review on the types of
analyses that can benefit from updated information or methods, with the goal of ensuring that the
exposure and risk estimates for this review reflect consideration of newly available information
or methods. Accordingly, we expect that in this new review we will develop updated analyses for
types of assessments for which new/updated information, methods or tools provide a basis for
producing appreciably improved or more targeted exposure and risk information. Thus, we do
5-30
-------
not expect to develop updated analyses for types of assessments for which associated
uncertainties limited their usefulness in the 2015 decision and are unlikely to be addressed by
information available in this review.
5.2.2.2 Initial Plans for the Current Review
Based on the considerations identified above, including consideration of the array of
complex and extensive exposure and risk analyses performed in the last review, and given the
expedited nature of this review, we are preliminarily planning that any new analyses in this
review include the two exposure-based analyses based on air quality monitoring data,
summarized in section 5.2.1.3 above. These analyses both include updates to the derivation of
cumulative exposure estimates at monitoring sites nationwide, providing for the assessment of
such exposures under air quality conditions that meet the current standard or any potential
alternatives for consideration. These two sets of analyses are (1) the analysis of O3
concentrations and derivation of W126 index values for Class I areas and (2) the similar analysis
for monitoring sites nationally. A decision to update these analyses would reflect the relatively
lesser uncertainty associated with these types of analyses as compared to the analyses based on
the regional air quality modeling approach; that lesser uncertainty contributed to the air quality
monitoring-based analyses being more informative in the last review.
Updates to these analyses can reflect the more recent, now available, air quality
monitoring data. These analyses are expected to inform our understanding of current patterns of
air quality and their impact on vegetation exposures under conditions just meeting the now-
current standard. Given the array of monitoring sites for which recent design values indicate
conditions just meeting the current standard, such a focus on monitoring data is expected to again
be accompanied by reduced uncertainty compared to the regional modeling approach described
in the 2014 WREA. Preliminary consideration of such analyses based on model-adjusted air
quality scenarios does not indicate a potential for appreciably addressing key uncertainties, such
that we expect that those analyses would not be updated, but the results from the last review may
be considered as relevant in the current review (e.g., with regard to exposure/risk considerations
in the PA).124
Interpretation of the cumulative exposure estimates in the two types of air quality and
exposure analyses will be informed by the consideration of the currently available evidence on
relationships of cumulative O3 exposure with tree seedling growth and visible foliar injury. Such
information is expected to include up-to-date tree seedling E-R functions for RBL based on the
124 Note that the approach for the WREA differed from that used in the HREA, with the latter focused on urban areas
(as summarized in section 5.1 above) as compared to the large regions that were the focus of the adjustment
approach in the WREA (2014 WREA, section 4.3; 2014 HREA, section 2.2).
5-31
-------
currently available evidence, as well as currently available information on relationships between
cumulative O3 exposures and visible foliar injury, building on the information available in the
last review (e.g., Smith and Murphy, 2015). As indicated in Chapter 4, the ISA for this review
will consider and assess the currently available evidence on the role of O3 in effects on
vegetation growth, and in visible foliar injury. Quantitative exposure-based analyses performed
for the PA will be interpreted considering this current evidence as presented in the ISA.
All of the analyses developed in this review will be described in the PA, with details
documented in appendices or accompanying volumes, as appropriate. We expect to also consider
in the PA any other types of analyses from the last review that we do not update in this review
but that are still informative to this review when viewed in the context of the currently available
evidence as characterized in the ISA and of updated air quality and other analyses performed for
this review. Accordingly, the PA will include description and discussion of all risk and exposure
analyses being considered in this review, both those newly performed in this review as well as
analyses performed for the last review for which an updated assessment was not performed but
that are still informative for this review. The draft PA will be released for public comment and
provided to the CASAC for its review. Advice and comments received will be considered in
completing the final version of the risk and exposure analyses and drawing on all of the analyses
considered in the policy evaluations presented in the final PA.
5-32
-------
6 POLICY ASSESSMENT
As described in section 1.2 above, the PA is a document that provides an evaluation of
the currently available information with regard to the adequacy of the current standards and
potential alternatives, if any are appropriate to consider in the current review. The PA integrates
and interprets the information from the ISA and available information from quantitative
exposure/risk analyses to frame policy options for consideration by the Administrator. This
evaluation of policy implications is intended to "bridge the gap" between the Agency's scientific
assessments and the judgments required of the EPA Administrator in determining whether it is
appropriate to retain or revise the NAAQS.
The discussion in the O3 PA in this review will be framed by consideration of a series of
the policy-relevant questions drawn from those outlined in chapter 3, including the fundamental
questions associated with the adequacy of the current standards and, as appropriate,
consideration of alternative standards that involve revision to any of the specific elements of the
standards: indicator, averaging time, level, and form. The PA conclusions will be based on the
assessment of the scientific information contained in the ISA, any updated exposure/risk
assessments or other additional evaluations and assessments discussed in the PA. Thus, the PA
will address the implications of the science and quantitative assessments for the adequacy of the
current standards, and, as appropriate, for any potential alternative standards. To the extent it is
concluded to be appropriate to consider potential alternative standards, the PA will also describe
a range of policy options for such revisions that is supported by the available information. In so
doing, the PA will describe the underlying interpretations of the scientific evidence,
risk/exposure information and any other quantitative analyses that might support such alternative
policy options and that could be considered by the Administrator in making decisions for the O3
standards. Additionally, the PA will identify key uncertainties in this policy evaluation and areas
for future research and data collection.
With regard to the primary standard, it is recognized that the final decision will be largely
a public health policy judgment by the Administrator. A final decision must draw upon scientific
information and analyses about health effects and risks, as well as judgments about how to deal
with the range of uncertainties that are inherent in the scientific evidence and analyses.
Consistent with the Agency's approach across all NAAQS reviews, the approach of the PA to
informing these judgments is based on a recognition that the available health effects evidence
generally reflects continuums that include ambient air exposures for which scientists generally
agree that health effects are likely to occur through lower levels at which the likelihood and
magnitude of response become increasingly uncertain. This approach is consistent with the
6-1
-------
requirements of the NAAQS provisions of the Act and with how the EPA and the courts have
historically interpreted the Act. These provisions require the Administrator to establish primary
standards that are requisite to protect public health with an adequate margin of safety. In so
doing, the Administrator seeks to establish standards that are neither more nor less stringent than
necessary for this purpose. The provisions do not require that standards be set at a zero-risk level,
but rather at a level that avoids unacceptable risks to public health, including the health of
• • 195
sensitive groups.
With regard to the secondary standard, it is recognized that the final decision will be
largely a public policy judgment by the Administrator. A final decision must draw upon
scientific evidence and analyses about effects on public welfare, as well as judgments about how
to deal with the range of uncertainties that are inherent in the relevant information. This approach
is consistent with the requirements of the NAAQS provisions of the Act and with how the EPA
and the courts have historically interpreted the Act. These provisions require the Administrator to
establish secondary standards that are requisite to protect public welfare from any known or
anticipated adverse effects associated with the presence of the pollutant in the ambient air. In so
doing, the Administrator seeks to establish standards that are neither more nor less stringent than
necessary for this purpose. The provisions do not require that secondary standards be set to
eliminate all welfare effects, but rather to protect public welfare from those effects that are
judged to be adverse.
The O3 PA will include pertinent background information, such as information on current
air quality as well as the decisions in the last NAAQS review, as well as discussion of the
currently available health and welfare effects evidence and exposure/risk information. These
discussions will be focused on policy-relevant aspects important for the Agency to consider in
reviewing the current standards. With regard to the exposure and risk information, the details of
any new analyses will be documented with the PA (e.g., in appendices or associated volumes)
and the findings presented and discussed within the main body of the PA.
The draft PA, with associated appendices that fully describe and document updated risk,
exposure and other quantitative analyses, will be distributed to the CASAC for its consideration
and released to the public for review and comment. Review of the draft PA by the CASAC also
facilitates CASAC's advice to the Agency and recommendations to the Administrator on the
adequacy of the existing standards or revisions that may be appropriate to consider, as provided
125 More than one population group may be identified as sensitive or at-risk in a NAAQS review. The decision in the
review will reflect consideration of the degree to which protection is provided for these sensitive population
groups. To the extent that any particular population group is not among the identified sensitive groups, a decision
that provides protection for the sensitive groups would be expected to also provide protection for other population
groups.
6-2
-------
for in the Clean Air Act. The CASAC will discuss its review of the draft PA at public meetings
that will be announced in the Federal Register. Based on past practice by the CASAC, the EPA
expects that key advice and recommendations for revision of the document would be
summarized by the CASAC in a letter to the EPA Administrator. In revising the draft PA
document, any such advice and recommendations will be taken into account, and comments
received from the public will also be considered. The final document will be made available on
an EPA website, with its public availability announced in the Federal Register.
6-3
-------
7 PROPOSED AND FINAL DECISIONS
Following issuance of the final PA and consideration of analyses and conclusions
presented therein, and taking into consideration CASAC advice and recommendations, the
Agency will develop a notice of proposed decisions. This notice will convey the Administrator's
proposed conclusions, reached in consideration of the analyses and conclusions in the documents
developed in the review (e.g., as described in the preceding chapters) and advice and
recommendations from the CASAC, regarding the adequacy of the current standards and any
revision(s) that may be appropriate. Development of the notice of the proposed (and final)
decisions will take into account issues related to the NAAQS process (e.g., Pruitt, 2018), as
appropriate in this review. As appropriate, a draft notice of proposed decision will be submitted
to the Office of Management and Budget (OMB) for its review and comment. In this interagency
review step, the OMB also provides to other federal agencies the opportunity for review and
comment. After the completion of interagency review, the notice of proposed action is published
in the Federal Register.
At the time of publication of the notice of the proposed action, all materials on which the
proposal is based are made available in the public docket for the review.126 Publication of the
proposal notice is followed by a public comment period, generally lasting 60 to 90 days, during
which the public is invited to submit comments on the proposal to the docket and one or more
public hearings may be held. Taking into account comments received on the proposed action, the
Agency will then develop a notice of final action, which communicates the Administrator's
decisions regarding this review and which may again undergo OMB-coordinated interagency
review prior to issuance by the EPA. At the time of the final action, the Agency responds to all
significant comments on the proposal.127 Publication of the notice of the final action in the
Federal Register will complete the review process.
126 The docket for the current O3 NAAQS review is identified as EPA-HQ-OAR-2018-0279. This docket has
incorporated the ISA docket (EPA-HQ-ORD-2018-0274) by reference. Both dockets are publicly accessible at
www.regulations.gov.
127 For example, Agency responses to all significant comments on the 2014 notice of proposed rulemaking in the last
review were provided in the preamble to the final rule and in a document titled "Response to Significant
Comments on the 2014 Proposed Rule on the National Ambient Air Quality Standards for Ozone (December 17,
2014; 79 FR 75234)", which is available at: https://www.epa.gov/naaas/responses-significant-comments-2014-
proposed-rule-national-ambient-air-qualitv-standards-ozone.
7-1
-------
8 REFERENCES
Adams WC. (2006). Comparison of chamber 6.6 hour exposures to 0.04-0.08 ppm ozone via
square-wave and triangular profiles on pulmonary responses. Inhal Toxicol. 18:127-136.
http://dx.doi.org/10.1080/08958370500306107.
Akinbami LJ, Simon AE, and LM Rossen. (2016). Changing trends in asthma prevalence among
children. Pediatrics. 137(l):e20152354.
American Thoracic Society. (2000). What Constitutes an Adverse Health Effect of Air Pollution?
Am J Respir Crit Care Med. 161:665-673.
American Thoracic Society. (1985). Guidelines as to What Constitutes an Adverse Respiratory
Health Effect, with Special Reference to Epidemiologic Studies of Air Pollution. Am Rev
Respir Dis. 131:666-668.
Appel W, Napelenok S, Hogrefe C, Pouliot G, Foley K, Roselle S, Pleim J, Bash J, Pye H, Heath
N, Murphy B, and R Mathur. (2017). Overview and Evaluation of the Community
Multiscale Air Quality (CMAQ) Modeling System Version 5.2. Chapter 11, Air Pollution
Modeling and its Application XXV. Springer International Publishing AG, Cham (ZG),
Switzerland, pp. 69-73. https://doi.org/10.1007/978-3-319-57645-9 11
Brown JS, Bateson TF, and WF McDonnell. (2008). Effects of exposure to 0.06 ppm ozone on
FEV1 in humans: A secondary analysis of existing data. Environ Health Perspect. 116:
1023-1026. http://dx.doi.org/10.1289/ehp.11396
Bais AF, Lucas RM, Bornman JF, Williamson CE, Sulzberger B, Austin AT, Wilson SR,
Andrady AL, Bernhard G, McKenzie RL, Aucamp PJ, Madronich S, Neale RE, Yazar S,
Young AR, de Gruijl FR, Norval M, Takizawa Y, Barnes PW, Robson TM, Robinson
SA, Ballare CL, Flint SD, Neale PJ, Hylander S, Rose KC, Wangberg SA, Hader DP,
Worrest RC, Zepp RG, Paul ND, Cory RM, Solomon KR, Longstreth J, Pandey KK,
Redhwi HH, Torikai A, and AM Heikkila. (2018). Environmental effects of ozone
depletion, UV radiation and interactions with climate change: UNEP Environmental
Effects Assessment Panel, update 2017. Photochem Photobiol Sci. 17:127-179.
Bennett M, Schofield K, Lee S, and S Norton.(2017). Response of chlorophyll a to total nitrogen
and total phosphorus concentrations in lotic ecosystems: a systematic review protocol.
Environmental Evidence. 6:18
Center for Disease Control and Prevention. (2016). Current Asthma Prevalence by Weight Status
Among Adults: United States, 2001-2014. NCHS Data Brief No. 239, March 2016.
Available at: https://www.cdc.gov/nchs/products/databriefs/db239.htm
Cohen AM, Hersh WR, Peterson K, and PY Yen. (2006). Reducing workload in systematic
review preparation using automated citation classification Journal of the American
Medical Informatics Association. 13:206-219. http://dx.doi.Org/10.l 197/iamia.ml929
8-1
-------
Collet S, Kidokoro T, Karamchandani P, Shah T, and J Jung. (2017). Future-year ozone
prediction for the United States using updated models and inputs. J. Air Waste Manage.
Assoc. 67 (8):938-948. https://doi: 10.1080/10962247.2017.1310149
Cox LA, Jr. (2019) Letter from Dr. Louis Anthony Cox, Jr, Chair, Clean Air Scientific Advisory
Committee, to Acting Administrator Andrew R. Wheeler. Re: Consultation on the EPA's
Integrated Review Plan for the Review of the Ozone National Ambient Air Quality
Standards (External Review Draft - October 2018). December 10, 2018. EPA-CASAC-
19-001.
Darrow LA, Klein M, Sarnat JA, Mulholland JA, Strickland MJ, Sarnat SE, et al. (2011). The
use of alternative pollutant metrics in time-series studies of ambient air pollution and
respiratory emergency department visits. JESEE. 21:10-19.
Folinsbee LJ, McDonnell WF, and DH Horstman. (1988). Pulmonary function and symptom
responses after 6.6-hour exposure to 0.12 ppm ozone with moderate exercise. JAPCA
38:28-35.
Frey HC. (2014a). Letter from Dr. H. Christopher Frey, Chair, Clean Air Scientific Advisory
Committee, to Administrator Gina McCarthy. Re: CASAC Review of the EPA's Health
Risk and Exposure Assessment for Ozone (Second External Review Draft - February
2014). EPA-CASAC-14-005. July 1, 2014.
Frey HC. (2014b). Letter from Dr. H. Christopher Frey, Chair, Clean Air Scientific Advisory
Committee, to Administrator Gina McCarthy. CASAC Review of the EPA's Welfare
Risk and Exposure Assessment for Ozone (Second External Review Draft). EPA-
CASAC-14-003. June 18, 2104.
Frey HC. (2014c). Letter from Dr. H. Christopher Frey, Chair, Clean Air Scientific Advisory
Committee, to Administrator Gina McCarthy. CASAC Review of the EPA's Second
Draft Policy Assessment for the Review of the Ozone National Ambient Air Quality
Standards. EPA-CASAC-14-004. June 26, 2014.
Frey HC, and JM Samet. (2012). Letter from Dr. H. Christopher Frey, Chair and Dr. Jonathan M.
Samet, Immediate Past Chair, Clean Air Scientific Advisory Committee, to Administrator
Lisa P. Jackson. Re: CASAC Review of the EPA's Policy Assessment for the Review of
the Ozone National Ambient Air Quality Standards (First External Review Draft -
August 2012). EPA-CASAC-13-003. November 26, 2012.
Hazucha MJ, Folinsbee LJ, and PA Bromberg. (2003). Distribution and reproducibility of
spirometric response to ozone by gender and age. J Appl Physiol. 95(5): 1917-1925.
He H, Liang, X, Lei, H and DJ Wuebbles. (2016). Future U.S. ozone projections dependence on
regional emissions, climate change, long-range transport and differences in modeling
design. Atmos. Environ. 128, 124-133. https://doi.Org/10.1016/i.atmosenv.2015.12.064
Heck WW, and EB Cowling. (1997). The need for a long term cumulative secondary ozone
standard - An ecological perspective. EM January: 23-33.
8-2
-------
Henderson R. (2006). Letter from Dr. Rogene Henderson, CAS AC Chair to EPA Administrator
Stephen Johnson. Re: Clean Air Scientific Advisory Committee's (CASAC) Peer Review
of the Agency's 2nd Draft Ozone Staff Paper. October 24, 2006, EPA-CASAC-07-001.
Henderson, BH., Akhtar, F, Pye, HOT, Napelenok, SL, and WT Hutzell. (2014). A database and
tool for boundary conditions for regional air quality modeling: description and
evaluation. Geosci Model Dev, 7: 339-360.
Howard BE, Phillips J, Miller K, Tandon A, Mav D, Shah MR, Holmgren S, Pelch KE, Walker
V, Rooney AA, MaCleod M, Shah RR, and K Thayer. (2016). SWIFT-Review: a text-
mining workbench for systematic review Systematic Reviews. 5:87.
IPCC. (2014). Climate Change 2014: Synthesis Report. Contribution of Working Groups I, II,
and III to the Fifth Assessment Report of the Intergovernmental Panel on Climate
Change. [Core Writing Team, R.K. Pachauri and L.A. Meyer (eds.)]. IPCC, Geneva,
Switzerland.
Ito K, Thurston GD, and RA Silverman. (2007). Characterization of PM2.5, gaseous pollutants,
and meteorological interactions in the context of time-series health effects models.
JESEE. 17(S2):S45-60.
Jackson L. (2009). Memo from Lisa Jackson, Administrator, U.S. EPA to Elizabeth Craig,
Acting Assistant Administrator of Office of Air and Radiation, and Lek Kadeli, Acting
Assistant Administrator of Office of Research and Development. RE: Process for
Reviewing National Ambient Air Quality Standards. May 21, 2009. Available:
https://www3.epa.gov/ttn/naaqs/pdfs/NAAQSReviewProcessMemo52109.pdf
Jerrett M, Burnett RT, Pope III CA, Ito K, Thurston G, Krewski D, Shi Y, Calle E, and M Thun.
(2009). Long-term ozone exposure and mortality. New England Journal of Medicine.
360:1085-1095.
Katsouyanni K, Samet JM, Anderson HR, Atkinson R, Le Tertre A, Medina S, Samoli E,
Touloumi G, Burnett RT, Krewski D, Ramsay T, Dominici F, Peng RD, Schwartz J, and
A Zanobetti. (2009). Air pollution and health: A European and North American approach
(APHENA). (Research Report 142). Boston, MA: Health Effects Institute.
http://pubs.healtheffects.org/view.php?id=327
Langford AO, Alvarez RJ, Brioude J, Fine R, Gustin MS, Lin MY, Marchbanks RD, Pierce RB,
Sandberg SP, Senff CJ, Weickmann AM, and EJ Williams. (2017). Entrainment of
stratospheric air and Asian pollution by the convective boundary layer in the
southwestern U.S. J Geophys Res Atmos. 122:1312-1337.
Linn, W.S.; Y. Szlachcic; H. Gong, Jr.; P.L. Kinney and K.T. Berhane. 2000. Air pollution and
daily hospital admissions in metropolitan Los Angeles. Environ Health Perspect.
108(5):427-434.
Loomis D, Groose Y, Lauby-Secretan B, El Ghissassi F, et al. (2013). The carcinogenicity of
outdoor air pollution. Lancet Oncol. 14(13): 1262-1263.
8-3
-------
Mar TF, and JQ Koenig. (2009). Relationship between visits to emergency departments for
asthma and ozone exposure in greater Seattle, Washington. Ann Allergy Asthma
Immunol. 103:474-479.
McCurdy T, Glen G, Smith L, and Y Lakkadi. (2000). The National Exposure Research
Laboratory's Consolidated Human Activity Database. J Expo Anal Environ Epidemiol.
10:566-578.
McCurdy T and SE Graham. (2003). Using human activity data in exposure models: Analysis of
discriminating factor. J Expo Anal Environ Epidemiol. 3:294-317.
McDonnell WF, Chapman RS, Leigh MW, Strope GL, and AM Collier. (1985). Respiratory
responses of vigorously exercising children to 0.12 ppm ozone exposure. American
Review of Respiratory Disease. 132:875-879.McDonnell WF, Stewart PW, and MV
Smith. (2007). The temporal dynamics of ozone induced FEVi changes in humans: an
exposure-response model. Inhal Toxicol. 19:483-494.
McDonnell WF, Stewart PW, and MV Smith. (2010). Prediction of ozone-induced lung function
responses in humans. Inhal Toxicol. 22(2): 160-168.
McDonnell WF, Stewart PW, Smith MV, Kim CS and ES Schelegle. (2012). Prediction of lung
function response for populations exposed to a wide range of ozone conditions. Inhal
Toxicol. 24:619-633.
McDonnell WF, Stewart PW, and MV Smith. (2013). Ozone exposure-response model for lung
function changes: an alternate variability structure. Inhal Toxicol. 25:348-353.
Medina-Ramon, M.; A. Zanobetti and J. Schwartz. 2006. The effect of ozone and PM10 on
hospital admissions for pneumonia and chronic obstructive pulmonary disease: a national
multicity study. American Journal of Epidemiology. 163(6):579-588.Melillo JM,
Richmond TC, and GW Yohe, Eds. (2014). Climate Change Impacts in the United States:
The Third National Climate Assessment. U.S. Global Change Research Program.
Available:
http:// s3. amazonaws.com/nca2014/low/NC A3_Climate_Change_Impacts_in_the_United
%20States_LowRes.pdf?download=l.
National Research Council. (2009). Science and decisions: Advancing risk assessment.
Washington, D.C.: National Academies Press.
https://www.nap.edu/catalog/12209/science-and-decisions-advancing-risk-assessment
Nolte CG, Spero TL, Bowden JH, Mallard MS, and PD Dolwick. (2018). The potential effects of
climate change on air quality across the conterminous U.S. at 2030 under three
Representative Concentration Pathways (RCPs). Atmos Chem Phys Discuss.
https://doi.org/10.5194/acp-2018-510, 32pp.
Norval M, Lucas RM, Cullen AP, de Gruijl FR, Longstreth J, Takizawa Y and JC van der Leun.
(2011). The human health effects of ozone depletion and interactions with climate
change. Photochem Photobiol Sci. 10:199-225.
8-4
-------
Passannante AN, Hazucha MJ, Bromberg PA, Seal E; Folinsbee L, and G Koch. (1998).
Nociceptive mechanisms modulate ozone-induced human lung function decrements. J
Appl Physiol. 85:1863-1870.
Peacock M. (2006). Memo from Marcus Peacock, Deputy Administrator, U.S. EPA to George
Gray, Assistant Administrator of Office of Research and Development, and Bill Wehrum,
Assistant Administrator of Office of Air and Radiation. RE: Process for Reviewing
National Ambient Air Quality Standards. December 7, 2006. Available:
https://www3.epa.gov/ttn/naaqs/pdfs/memo_process_for_reviewing_naaqs.pdf
Peacock M (2008). Letter from Marcus Peacock, Deputy Administrator, U.S. EPA to Rogene
Henderson, Chair, Clean Air Scientific Advisory Committee, September 8, 2008.
Available: http://www3.epa.gov/ttn/naaqs/pdfs/PeacocklettertoCASAC090808.pdf.
Pruitt ES (2018). Memo from E. Scott Pruitt, Administrator, U.S. EPA to Assistant
Administrators. SUBJECT: Back-to-Basics Process for Reviewing National Ambient Air
Quality Standards. May 9, 2018. Available: https://www.epa.gov/criteria-air-
pollutants/back-basics-process-reviewing-national-ambient-air-quality-standards.
Seal E, McDonnell WF, and DE House. (1996). Effects of age, socioeconomic status, and
menstrual cycle on pulmonary response to ozone. Archives of Environmental and
Occupational Health. 51:132-137.
Samet J. (2009). Letter from Dr. Jonathan M. Samet, Chair, Clean Air Scientific Advisory
Committee to the Honorable Lisa P. Jackson, Administrator, U.S. EPA. CASAC Review
of EPA's Integrated Science Assessment for Particulate Matter - First External Review
Draft, December 2008. (May 21, 2009). Available:
http://ofmpub. epa.gov/eims/eimscomm. getfile?p_download_id=508493.
Samet JM. (2011). Letter from Dr. Jonathan M. Samet, Chair, Clean Air Scientific Advisory
Committee to Administrator Lisa P. Jackson. Re: Clean Air Scientific Advisory
Committee (CASAC) Response to Charge Questions on the Reconsideration of the 2008
Ozone National Ambient Air Quality Standards. EPA-CASAC-11-004. March 30, 2011.
http://yosemite.epa.gOv/sab/sabproduct.nsf/0/F08BEB48C1139E2A8525785E006909AC/
$File/EPA-CASAC-ll-004-unsigned+.pdf.
Samet JM. (2010). Letter from Dr. Jonathan M. Samet, Chair, Clean Air Scientific Advisory
Committee to Administrator Lisa P. Jackson. Re: Review of EPA's Proposed Ozone
National Ambient Air Quality Standard (Federal Register, Vol. 75, Nov. 11, January 19,
2010). EPA-CASAC-10-007. February 19, 2010.
http://yosemite.epa.gov/sab/sabproduct.nsf/264cbl227d55e02c85257402007446a4/610B
B57CFAC8A41C852576CF007076BD/$File/EP A-CASAC-10-007-unsigned.pdf.
Schelegle ES, Morales CA, Walby WF, Marion S, and RP Allen. (2009). 6.6-hour inhalation of
ozone concentrations from 60 to 87 parts per billion in healthy humans. Am J Respir Crit
Care Med. 180:265-272. http://dx.doi.org/10.1164/rccm.200809-14840C.
8-5
-------
Smith JT, and DL Murphy. (2015). Additional Observations From WREA Datasets for Visible
Foliar Injury. Memorandum to the Ozone NAAQS Review Docket, EPA-HQ-OAR-
2008-0699.
Smith RL, Xu B, and P Switzer. (2009). Reassessing the relationship between ozone and short-
term mortality in U.S. urban communities. Inhal Toxicol. 21:37-61.
Spektor DM, Lippmann M, Lioy PJ, Thurston GD, Citak K, James DJ, Bock N, Speizer FE, and
C Hayes. (1988). Effects of ambient ozone on respiratory function in active, normal
children. American Review of Respiratory Disease. 137(2):313-320.
Spektor DM, and M Lippmann. (1991). Health effects of ambient ozone on healthy children at a
summer camp, in Tropospheric Ozone and the Environment: Papers from an
International Conference; March 1990; Los Angeles, CA. Eds., R. L. Berglund; D. R.
Lawson; D. J. McKee. Air & Waste Management Association, Pittsburgh PA.
Strickland MJ, Darrow LA, Klein M, Flanders WD, Sarnat JA, Waller LA, et al. (2010). Short-
term associations between ambient air pollutants and pediatric asthma emergency
department visits. American Journal of Respiratory Critical Care Medicine. 182:307-316.
Tolbert PE, Klein M, Peel JL, Sarnat SE, and JA Sarnat. (2007). Multipollutant modeling issues
in a study of ambient air quality and emergency department visits in Atlanta. JESEE.
17(S2):29-35.
U.S. Department of Health, Education, and Welfare. (1970). Air Quality Criteria for
Photochemical Oxidants. Washington, D.C.: National Air Pollution Control
Administration; publication no. AP-63. Available from NTIS, Springfield, VA; PB-
190262/BA.
U.S. EPA. (1978). Air quality criteria for ozone and other photochemical oxidants. Washington,
DC. EPA/600/8-78/004.
U.S. EPA (1986). Air quality criteria for ozone and other photochemical oxidants. Research
Triangle Park, NC. EPA-600/8-84-020aF - EPA-600/8-84-020eF.
http://www.ntis.gov/search/product.aspx?ABBR=PB87142949.
U.S. EPA. (1989). Review of the National Ambient Air Quality Standards for Ozone,
Assessment of Scientific and Technical Information: OAQPS Staff Paper. Office of Air
Quality Planning and Standards, Research Triangle Park, NC. EPA Report No. 450/2-02-
001.
U.S. EPA. (1992). Summary of Selected New Information on Effects of Ozone on Health and
Vegetation, Supplement to Air Quality Criteria for Ozone and Other Photochemical
Oxidants. U.S. Environmental Protection Agency, Research Triangle Park, NC.
EPA/600/8-88/105F.
8-6
-------
U.S. EPA. (1996a). Air Quality Criteria for Ozone and Related Photochemical Oxidants.
Volumes 1 to 3. U.S. Environmental Protection Agency, Research Triangle Park, NC.
EPA/600/P-93/004aF, EPA/600/P-93/004bF, and EPA/600/P-93/004cF.
U.S. EPA. (1996b). Review of the National Ambient Air Quality Standards for Ozone,
Assessment of Scientific and Technical Information: OAQPS Staff Paper. Office of Air
Quality Planning and Standards, Research Triangle Park, NC. EPA Report No. 452/R-96-
007.
U.S. EPA. (2006a). Air Quality Criteria for Ozone and Related Photochemical Oxidants (2006
Final). U.S. Environmental Protection Agency, Washington, DC. EPA/600/R-05/004aF-
cF. March 2006. http://www.epa.gov/ttn/naaqs/standards/ozone/s_o3_cr_cd.html.
U.S. EPA. (2007). Ozone Population Exposure Analysis for Selected Urban Areas. Research
Triangle Park, NC: EPA Office of Air Quality Planning and Standards. Available at:
https://www3.epa.gov/ttn/naaqs/standards/ozone/data/2007_07_o3_exposure_tsd.pdf
U.S. EPA. (2008). Risk and Exposure Assessment to Support the Review of the NO2 Primary
National Ambient Air Quality Standard. Office of Air Quality Planning and Standards,
Research Triangle Park, NC, EPA-452/R-08-008a, November 2008. Available at:
https ://www3. epa.gov/ttn/naaqs/ standards/nox/ s_nox_cr_rea.html
U.S. EPA. (2009). Risk and Exposure Assessment to Support the Review of the SO2 Primary
National Ambient Air Quality Standard. Office of Air Quality Planning and Standards,
Research Triangle Park, NC, EPA-452/R-09-007, July 2009. Available at:
https://www3.epa.gov/ttn/naaqs/standards/so2/data/200908S02REAFinalReport.pdf
U.S. EPA. (2010). Quantitative Risk and Exposure Assessment for Carbon Monoxide -
Amended. Office of Air Quality Planning and Standards, Research Triangle Park, NC,
EPA-452/R-10-006, July 2010. Available at: https://www.epa.gov/naaqs/carbon-
monoxide-co-standards-risk-and-exposure-assessments-current-review
U.S. EPA. (2012a). Total Risk Integrated Methodology (TRIM) - Air Pollutants Exposure Model
Documentation (TRIM.Expo / APEX, Version 4.4) Volume I: User's Guide. Research
Triangle Park, NC: EPA Office of Air Quality Planning and Standards. (EPA document
number EPA-452/B-12-001a). Available at:
http://www.epa.gov/ttn/fera/human apex.html
U.S. EPA. (2012b). Total Risk Integrated Methodology (TRIM) - Air Pollutants Exposure Model
Documentation (TRIM.Expo / APEX, Version 4.4) Volume II: Technical Support
Document. Research Triangle Park, NC: EPA Office of Air Quality Planning and
Standards. (EPA document number EPA-452/B-12-001b). Available at:
http://www.epa.gov/ttn/fera/human apex.html
U.S. EPA. (2013a). Integrated Science Assessment for Ozone and Related Photochemical
Oxidants, U.S. Environmental Protection Agency, Research Triangle Park, NC, EPA
600/R-10/076F. Available: https://cfpub.epa.gov/ncea/isa/recordisplay.cfm?deid=247492
8-7
-------
U.S. EPA. (2013b). Environmental Benefits Mapping Analysis Program (BenMAP v4.0). Posted
January 2013. Available at: https://www.epa.gov/benmap/benmap-downloads
U.S. EPA. (2014a). Health Risk and Exposure Assessment for Ozone. Office of Air Quality
Planning and Standards, Research Triangle Park, NC. EPA-452/P-14-004a.
http://www.epa.gov/ttn/naaqs/standards/ozone/s_o3_index.html.
U.S. EPA. (2014b). Welfare Risk and Exposure Assessment for Ozone. Office of Air Quality
Planning and Standards, Research Triangle Park, NC. EPA-452/P-14-005a.
http://www.epa.gov/ttn/naaqs/standards/ozone/s_o3_index.html.
U.S. EPA. (2014c). Policy Assessment for Ozone. Office of Air Quality Planning and Standards,
Research Triangle Park, NC. EPA-452/R-14-006.
http://www.epa.gov/ttn/naaqs/standards/ozone/s_o3_index.html.
U.S. EPA. (2015a). Responses to Significant Comments on the 2014 Proposed Rule on the
National Ambient Air Quality Standards for Ozone (December 17, 2014; 79 FR 75234).
Docket Document Number EPA-HQ-OAR-2008-0699-4309.
U.S. EPA. (2015b). Regulatory Impact Analysis of the Final Revisions to the National Ambient
Air Quality Standards for Ground-Level Ozone, U.S. Environmental Protection Agency,
Research Triangle Park, NC, EPA-452/R-15-007.
U.S. EPA. (2015c). Preamble to the Integrated Science Assessments. U.S. Environmental
Protection Agency, Washington, DC, EPA/600/R-15/067. Available at:
https://cfpub.epa.gOv/ncea/i sa/recordisplay.cfm?deid=310244.
U.S. EPA. (2015d). Peer Review Handbook. U.S. Environmental Protection Agency, Science
Technology and Policy Committee, Washington, DC, EPA/100/B-15/001.
U.S. EPA. (2016). Integrated Science Assessment (ISA) for Oxides of Nitrogen - Health Criteria
(Final Report). U.S. Environmental Protection Agency, Washington, DC, EPA/600/R-
15/068, 2016.
U.S. EPA. (2017). The Consolidated Human Activity Database - Master Version (CHAD-
Master). Technical Memorandum. Office of Research and Development, National
Exposure Research Laboratory, Research Triangle Park, NC, 27711. In preparation.
Previous version (09/15/2014). Available at:
https://www.epa.gov/healthresearch/consolidated-human-activity-database-chad-use-
human-exposure-and-health-studies-and
U.S. EPA. (2018a). Technical Note - Guidance for Developing Enhanced Monitoring Plans.
U.S. Environmental Protection Agency, Ambient Air Monitoring Group, December 2018.
Available at: https://www3.epa.gov/ttn/amtic/pamsguidance.html
U.S. EPA. (2018b). 2014 National Emissions Inventory, version 2 Technical Support Document,
U.S. Environmental Protection Agency, Emissions Inventory and Analysis Group,
Research Triangle Park, NC.
-------
U.S. EPA. (2018c). Ozone Trends, U.S. Environmental Protection Agency. Available at:
https://www.epa.gov/air-trends/ozone-trends
U.S. EPA. (2018d). Integrated Science Assessment for Oxides of Nitrogen, Oxides of Sulfur, and
Particulate Matter-Ecological Criteria (2nd External Review Draft). EPA/60/R-18/097.
Available at: https://cfpub.epa.gov/ncea/isa/recordisplay.cfm?deid=340671
U.S. EPA. (2018e). Risk and Exposure Assessment for the Review of the Primary National
Ambient Air Quality Standard for Sulfur Oxides. Office of Air Quality Planning and
Standards, Research Triangle Park, NC, EPA-452/R-18-003, May 2018. Available at:
https://www.epa.gOv/sites/production/files/2018-05/documents/primary_so2_naaqs_-
_final_rea_-_may_2018. pdf
USGCRP. (2017). Climate Science Special Report: Fourth National Climate Assessment,
Volume I. Wuebbles DJ, Fahey DW, Hibbard KA, Dokken DJ, Stewart BC, and TK
Maycock [eds.]. U.S. Global Change Research Program, Washington, DC, USA, 470 pp,
doi: 10.7930/J0J964J6
Wells B, Wesson K, and S Jenkins. (2012). Analysis of Recent U.S. Ozone Air Quality Data to
Support the 03 NAAQS Review and Quadratic Rollback Simulations to Support the First
Draft of the Risk and Exposure Assessment. Memorandum to the EPA Docket # EPA-
HQ-EPA-2008-0699. Document Identifier EPA-HQ-EPA-2008-0699-4253.
Wells B. (2015). Expanded Comparison of Ozone Metrics Considered in Current NAAQS
Review. Memorandum to the Ozone NAAQS Review Docket, EPA-HQ-OAR-2008-
0699. Document Identifier EPA-HQ-OAR-2008-0699-4325.
Whitfield R, Biller W, Jusko M, and J Keisler. (1996). A Probabilistic Assessment of Health
Risks Associated with Short- and Long-Term Exposure to Tropospheric Ozone. Argonne,
IL: Argonne National Laboratory. Available at:
https://www3.epa.gov/ttn/naaqs/standards/ozone/data/riskrep.pdf
Zanobetti A and J Schwartz. (2008). Mortality displacement in the association of ozone with
mortality: an analysis of 48 cities in the United States. American Journal of Respiratory
and Critical Care Medicine. 177:184-189.
8-9
-------
Appendix 5A
-------
Appendix 5A. Limitations and uncertainties of exposure and risk analyses developed in the last review of the primary
standard and consideration of related newly available information and tools. Drawn from the 2014 HREA, Tables 4-7, 5-10, 6-
20, 7-4, notice of final decision and response to comments document for the review.
Analysis Element
Limitations/Uncertainty identified in 2014 HREA
2014 Uncertainty Characterization and Newly Available
Information for Current Review
Ambient Air Concentrations
Ambient air monitoring data
The monitoring datasets used for the 2014 HREA were for the period from 2006
through 2010.
Overall, O3 measurement data are of high quality and have
low uncertainty. Newly available are data for more recent
3-year period (2015-2017).
Approach used to derive
factors to adjust air quality
to just meet then-existing
and potential alternate
standards
Modelina Platforms and Approaches: Model predictions from the Community
Multiscale Air Quality (CMAQ) model, like all deterministic photochemical models,
have both parametric and structural uncertainty associated with them. Higher
Order Decoupled Direct Method (HDDM) allows for the efficient approximation of
O3 concentrations under alternate emissions scenarios. This approximation is less
accurate for larger emissions perturbations, especially under nonlinear chemistry
conditions.
Low to moderate magnitude of impact on exposure and
FEV1 risk estimates potentially resulting in both under- and
over-estimation of ambient concentrations.
Updated modeling platforms are available since completion
of the 2014 HREA. We could apply HDDM in the
CAMxv6.5 photochemical model (somewhat faster
approach than HDDM with CMAQ) which includes updated
chemical mechanisms reflecting understanding of
important chemical pathways for ozone formation and
destruction that have been extended since the chemistry
available during the last review. We would use modeling
inputs that reflect emissions, meteorology and international
transport representing a more recent year (2016).
Based on results from modeling performed in the 2014
HREA and time constraints for this review, we would focus
primarily on NOx reductions alone.
To reduce uncertainty in analyses for this review, we may
select a subset of study areas based on consideration of
CMAQ/HDDM model performance in different urban areas
as well as occurrence of any atypical O3 episodes during
the modeled period.
Application of HDDM sensitivities to ambient data: there is uncertainty in the
statistical regressions used to relate O3 response to emissions perturbations with
ambient O3 concentrations for every season, hour-of-the-day, and monitor
location. Further, functional relationships between O3 response and hourly O3
concentration were developed based on 8 months of modeling: January and April-
October 2007 and applied to ambient data from 2006-2010. Some locations
monitor for months not included in this modeling (February, March, November,
and December) while others do not.
Emissions Reduction Assumptions: In cases where VOC reductions were
modeled, equal percentage NOx and VOC reductions were applied in the
adjustment methodology. Assumption of across-the-board emissions reductions:
Ozone response is modeled for across-the-board reductions in U.S. anthropogenic
NOx (and VOC). These across-the-board cuts do not reflect actual emissions
control strategies.
Approach used to spatially
interpolate ambient air
monitor concentrations to
census tracts
Voronoi Neighbor Averaging (VNA) is a spatial interpolation technique used to
estimate O3 concentrations in unmonitored areas, which has inherent uncertainty.
The relative influence on exposure and risk estimates range from low to moderate,
with greatest uncertainties when interpolating large distances between monitors.
The uncertainty in this approach could lead to both under-
and over-estimation of ambient concentrations. However,
the magnitude of impact to exposure and FEV1 risk
estimates was estimated to range between low to
5A-1
-------
Analysis Element
Limitations/Uncertainty identified in 2014 HREA
2014 Uncertainty Characterization and Newly Available
Information for Current Review
moderate. Several other methods are available, with
associated limitations. For this review, preferred study
areas could include spatial coverage of ambient air
monitors relative to study area dimensions as a study
selection criterion.
Exposure Modeling
APEX general input
databases
There are several general databases used including year 2000 population
demographics and commuting, CHAD activity diaries, area-specific meteorological
data, and 2006-2010 asthma prevalence.
2014 HREA characterization indicated most databases
were of high quality and had low impact to estimated
exposures. Meteorological and asthma prevalence data
could be updated to appropriately correspond with the
selected study areas and exposure periods. There are no
new activity pattern data however the CHAD activities have
been expanded and the associated METs distributions
were revised. The demographic data have been updated to
reflect the 2010 census. However, a limited sensitivity
analysis in the 2014 HREA using the 2010 census
indicated a small effect, though consistently yielding lower
FEVi risk estimates (Table 6-18, 2014 HREA).
APEX anthropometric
attributes and physiological
processes
There are several databases and algorithms used to estimate body weight (BW),
resting metabolic rate (RMR), normalized oxygen consumption rate (nV02),
metabolic equivalents of work (METS), and ventilation rates (VE) that may
contribute to uncertainty in the estimated exposures.
The 2014 HREA characterized these as having between a
low to moderate impact on estimated exposures, with two
(VE and METS) potentially contributing to overestimates.
We have since updated each of these to some extent
using either recent data or new algorithms except for the
nV02.
APEX microenvironmental
concentrations
There was uncertainty associated with approaches and factors used to estimate
concentrations within indoor, outdoor, and inside vehicle microenvironments
including air exchange rates, air conditioning (A/C) prevalence, indoor removal
rates, proximity factors to adjust for near road concentrations, and penetration
factors.
Because the highest O3 exposures occur in outdoor
environments, these factors were characterized as having
low impact to estimated exposures. While some data
would be updated (e.g., A/C prevalence), most factors
used in 2014 would be reapplied.
Representation of time
outdoors considering air
quality advisories
Limited availability of data on averting behavior in response to air quality alerts
indicates that a small percentage of the population may engage in averting
behavior. The lack of representation of this in the exposure modeling may
contribute to overestimates of actual exposures in such circumstances (2014
HREA, pp. 5-53 to 5-54; p. 9-11). A sensitivity analysis performed for the 2014
HREA estimated 1-2 percentage point reductions in the percent of simulated
children at or above benchmark levels when accounting for averting by a portion of
the population and for a particular duration. These results indicate that, depending
While not specifically characterized in the 2014 HREA,
simulating the averting of high concentration air pollution
events had a moderate impact on the estimated
exposures, suggesting the number of people exposed at or
above benchmark levels may be overestimated. There
may be recent published literature to support the
parameters used to develop the averting scenario or to
develop a new scenario to better reflect current averting
behavior and better characterize the impact to exposures.
5A-2
-------
Analysis Element
Limitations/Uncertainty identified in 2014 HREA
2014 Uncertainty Characterization and Newly Available
Information for Current Review
on benchmark levels, averting could lead to 20% or greater reductions in the
number of people experiencing exposures of interest.
Estimating repeated
exposures for select at-risk
populations
The limited availability of longitudinal activity diary data and the general population
modeling approach used may underestimate the correlation in activity patterns for
certain potentially at-risk populations (e.g., outdoor workers or the subset of
children with systematically high outdoor activity levels). Accordingly, the results
may underestimate how often there are repeated exposures to exposures above
benchmarks and we are limited in our ability to identify the percent of the
population with unusually high numbers of multiple exposures (2014 HREA,
section 9.5.2). The simulated scenarios were highly dependent on existing activity
pattern data and several assumptions made to characterize a particular at-risk
population.
While not specifically characterized in the 2014 HREA,
simulating potentially at-risk populations having repeated
exposure to high air pollution events had a moderate
impact on the estimated exposures, suggesting the
number of people exposed at or above benchmark levels
may have been underestimated. Unclear as to whether
new data are available to enhance the approach used.
Comparison of Simulated Exposures to Benchmarks
Cut point for moderate or
greater ventilation
An equivalent ventilation rate (EVR in L/min-m2) served as a cut point for selecting
simulated individuals performing moderate or greater exertion activities. The EVR
was used to allow for extrapolation of information obtained from adults to children.
The value used (13 L/min-m2) was a lower bound based on approximating the 5th
percentile of the distribution of targeted ventilation rates maintained by the study
subjects (Whitfield et al., 1996). There is uncertainty in the extrapolation of adult
data to simulated children and the use of a lower bound value.
The 2014 HREA recognized that the simulated number of
people achieving this level of exertion could be
moderatelyoverestimated, affecting the results for
comparison to benchmarks and the population-based E-R
approach used to estimate lung function risk. A new
approach to identifying when individuals may be at
moderate or greater exertion could be explored using
available exposure study data. Consideration will also be
given to the total time-averaged ventilation rate in
calculating the EVR of study subjects (see exposure
duration below), rather than using the exercise ventilation
rate alone to calculate EVR (as was done in prior reviews).
Exposure duration
The exposure duration for the studies from which the benchmark concentrations
are drawn is 6.6 hours (6 x 50 min exercise periods separated by 10-minute rest
periods, and with a 35-minute lunch after 3rd hour). Simulated exposures relied on
a daily maximum 8-hour averaging time. Therefore, health responses observed at
a 6.6-hour concentration would directly relate to a lower 8-hour average
concentration. Further, there is some indication that the pattern of the exposure
may be important to generating the adverse health response (2013 ISA, section
6.2.1.1, pp. 6-10 to 6-11). The approach used to define the exposure benchmark
considered average concentration over the exposure period without consideration
of exposure pattern or peak concentrations within the exposure averaging time.
The simulated number of people with exposures at or
above benchmarks and those expected to experience lung
function decrements via the population-based E-R
approach could have been 1) underestimated when
considering the different averaging periods, and 2)
underestimated or overestimated when ignoring the pattern
of exposure within the averaging period. New benchmarks
that better reflect the averaging time used in the controlled
human exposure study data could be used (e.g., 6 or 7
hours)
Benchmark concentrations
An important uncertainty is that there is only very limited evidence from controlled
human exposure studies of population groups potentially at greater risk.
Compared to the healthy young adults included in the controlled human exposure
Although not directly characterized in the 2014 HREA, the
benchmark levels derived from the controlled human
exposure studies may not be entirely representative of
5A-3
-------
Analysis Element
Limitations/Uncertainty identified in 2014 HREA
2014 Uncertainty Characterization and Newly Available
Information for Current Review
studies, members of some populations (e.g., children with asthma) are considered
more likely to experience adverse effects following exposures to lower O3
concentrations (80 FR 65322, 65346, October 26, 2015; Frey 2014a, p. 7).
effects likely to be exhibited by the simulated population
and could underestimate the size of the population at risk
and/or the magnitude of adverse effects.
Exposure to O3 alone vs.
O3 plus all related oxidants
in ambient air
The controlled human exposure studies evaluated the adverse health effects
resulting from O3 exposure only. Exposure and risk estimated for simulated
individuals also considers O3 exposures alone (as is appropriate given the E-R
functions that are derived from the 03-only exposure studies). However, O3 serves
as a surrogate for all oxidants that exist in ambient air and, as such, it is possible
that individuals could be exposed to these pollutants in addition to O3. Adverse
health effects (if any) resulting from oxidants other than O3 are not accounted for
when using the current 03-only exposure approach.
This element was not characterized in the 2014 HREA. We
are unaware of any controlled human exposure studies
that evaluated health effects resulting from exposure to a
mixture of O3 and other oxidants. Therefore, it is largely
unknown how health effects might be altered following
exposure to oxidants other than O3 and as such, it is
uncertain as to how this may relate to the estimated risk in
the assessment.
MSS FEV1 Lung Function Risk Assessment
The McDonnell-Stewart-
Smith (MSS) FEV1
model for ages 18 to 35
While there is a good conceptual foundation for the structure of the MSS model,
the variability in measurements of FEV1 and estimated parameters of the model
introduce uncertainty into estimates of FEV1 reductions. For instance, some of the
estimated parameters have wide confidence intervals (2014 HREA, Table 6-14).
Sensitivity analyses in the 2014 HREA additionally addressed how the general
pattern of exercise/ventilation of study subjects affects estimated risks, however
there were no evaluations of how exposure patterns of study subjects or changes
in other influential attributes may affect risk estimates.
A new MSS model (McDonnell et al., 2013) is available for
use in this review.
Representation of inter-
individual variability
There is uncertainty in the degree to which the MSS model represents inter-
personal variability in FEV1 reductions (i.e., via the MSS model variable Var(U)).
This is the result of having very few exposure studies with repeated clinical trials
using the same individuals, likely yielding an underestimate in the Var(U)
parameters. In addition, the method used for adjusting for filtered air (FA)
exposures in the data used to fit the MSS model does not use the subject-specific
adjustments, rather the mean FA response across a study is used to adjust the O3
responses of each subject in the study. Furthermore, there are few clinical data for
population with diseased lungs (i.e., asthma), thus the MSS model may not
account for the increase in inter-individual variability that would result from
inclusion of exposure-response (E-R) data from such individuals. A higher Var(U)
indicates greater between-individual variability and less within-individual variability,
therefore more responsive individuals are more likely to see repeated occurrences
of high AFEV1 (and thus less responsive individuals are more likely to see no
occurrences of high AFEV1).
The 2014 HREA concluded that the number of people
experiencing FEV1 decrements could be moderately
overestimated given underestimates in the MSS FEV1
model Var(U) parameter (absent the influence by other
sources of uncertainty).
Representation of intra-
individual variability
There is uncertainty in the degree to which the MSS model represents intra-
personal variability in FEV1 reductions (i.e., via the MSS model variable Var(s)).
The Var(s) term is assumed to have a Gaussian distribution {mean=0, standard
Sensitivity analyses conducted in the 2014 HREA indicated
that how the MSS FEV1 model Var(s) parameters are
bounded has a moderate or greater influence in predicting
5A-4
-------
Analysis Element
Limitations/Uncertainty identified in 2014 HREA
2014 Uncertainty Characterization and Newly Available
Information for Current Review
deviation=4.14} and for our purposes in estimating risk was bounded at ±2
standard deviations (i.e., ±8.3). Extending or restricting these bounds will result in
either greater or fewer simulated individuals experiencing lung function
decrements, respectively. The assumption that the distribution of this term is
Gaussian is convenient for fitting the data but may not be accurate.
the proportions of the population with FEV1 decrements >
10 and 15%. It is not clear how potential misspecification of
the Var(s) distribution shape affects its parameters and that
of other variables in the MSS model, and how these
changes may affect risk estimates.
Extrapolation of MSS
variable parameters
estimated for adults (18-
35) to children (ages 5 to
18)
There are virtually no controlled human exposure data for children (i.e., the
youngest age for which controlled human exposure data are generally available is
18 years old). Thus, the 2014 HREA essentially applied the same lung function
response following O3 exposures to children as was applied for adults (2014
HREA, section 6.5.3). This assumption is justified in part by the findings of
McDonnell et al. (1985), who reported that children (8-11 years old) experienced
FEV1 responses similar to those observed in adults (18-35 years old) (2014
HREA, p. 3-10) and from summer camp studies of school-aged children reported
03-induced lung function decrements similar in magnitude to those observed in
controlled human exposure studies using adults (2013 ISA, section 6.2.1). To
estimate health risk for children, a constant value was used for the MSS model
age variable (and derived from 18-year olds, and as a maximum value). There is
uncertainty in this approach, depending on how this age term influences overall
risk estimates for children compared to adults in controlled human exposure
studies (2014 HREA, section 6.5.3).
The 2014 HREA concluded that the extrapolation approach
could result in moderate over- or underestimates of 03-
induced lung function decrements in simulated children.
Extrapolation of
exposure-response data
from healthy subjects to
simulated people with
asthma
There is uncertainty associated with using E-R functions derived from healthy
subjects in the controlled exposure studies to estimate 03-induced lung function
risk in simulated individuals with asthma (2014 HREA, section 6.5.4). Although the
evidence is mixed (2013 ISA, section 6.2.1.1), several studies have reported
statistically larger, or a tendency toward larger, 03-induced lung function
decrements in asthmatics than in non-asthmatics (Kreit et al., 1989; Horstman et
al., 1995; Jorres et al., 1996; Alexis et al., 2000). On this issue, CASAC noted that
"[ajsthmatic subjects appear to be at least as sensitive, if not more sensitive, than
non-asthmatic subjects in manifesting 03-induced pulmonary function decrements"
(Frey, 2014c, p. 4). Furthermore, the response could depend on a variety of
factors that have not been well-evaluated, including the severity of asthma and the
prevalence of medication use. For instance, responses to O3 increase with
severity of asthma (Horstman et al., 1995) and corticosteroid usage does not
prevent 03-induced lung function decrements or respiratory symptoms in people
with asthma (Vagaggini et al., 2001, 2007).
The 2014 HREA indicated that if asthmatics experience
larger 03-induced lung function decrements than the
healthy adults used to develop E-R functions, the impacts
of O3 exposures on lung function in asthmatics, including
asthmatic children, could be underestimated, albeit to an
unknown extent.
Population-based Exposure-Response model
5A-5
-------
Analysis Element
Limitations/Uncertainty identified in 2014 HREA
2014 Uncertainty Characterization and Newly Available
Information for Current Review
Cut point for moderate or
greater ventilation
See entry for this element under Comparison to Benchmarks section. The
approach used could overestimate the number of individuals at moderate or
greater exertion.
While not directly characterized in the 2014 HREA, the
reported number and percent of individuals estimated to
experience a lung function decrement would likely be
greater than that estimated using a higher, alternative EVR
value to estimate elevated exertion.
Exposure duration
See entry for this element under Comparison to Benchmarks section. The duration
used results in fewer simulated individuals identified as having the exposure of
interest than expected for the E-R function.
While not directly characterized in the 2014 HREA, the
reported number and percent of individuals estimated to
experience a lung function decrement would be
underestimated given the difference in exposure durations.
E-R function shape
In both the 2007 O3 Staff Paper and 2014 HREA, an E-R function was derived
using a combination of two functions (90% logistic fit and 10% linear-threshold).
The selection of this parameterization was based largely on 1) linearity of E-R
function for exposures between 0.08 - 0.12 ppm (and used in the 1997 O3 risk
assessment), a "very good" logistic model fit (2007 Staff Paper), and CASAC
advice noting a linear model cannot entirely be ruled out given the are limited data
at the two lowest exposure levels (Henderson, 2006). Sensitivity analyses of three
different logistic/linear-threshold forms (90/10, 80/20, 50/50) indicated differences
in the estimated risks, most notably lower risks estimated with increasing
proportion of the linear threshold form and when considering the air quality
adjusted to the lowest standard level of 64 ppb (2007 Staff Paper). A key issue of
concern regarding each of these model fits is how responses are estimated at
concentrations below which we have controlled human exposure study data (i.e.,
<40 ppb).
While not directly characterized in the 2014 HREA, the
reported number and percent of individuals estimated to
experience a lung function decrement may be greater
when using primarily a logit fit than when using a probit fit.
Based on the 2009 and 2018 SOx REAs, the use of a
probit form of a logistic model is more appropriate than
using a logit form. This is based on assumptions regarding
the distribution of individual thresholds for response
supporting the use of a probit function, which is based on
the inverse of the cumulative standard normal distribution
function, rather than a logistic function which assumes a
logistic distribution, for estimating risk (U.S. EPA, 2009,
2018e). It is possible the combined 90% logistic/10% linear
may be more similar to a probit form (i.e., have lower
response at lowest concentrations), the impact to risk
estimates remains uncertain.
Ambient AQ (epidemiologic study)-based risk
Ambient air concentrations
Relationship between population exposures and ambient air monitor
concentrations: One of the assumptions in the use of ambient air concentration-
response functions drawn from epidemiological studies to estimate risk associated
with a pollutant for a modeled air quality scenario and population is that the
relationship between ambient air monitor concentrations (usually represented in
the studies by an area-wide average) and the exposure of the population is the
same in the modeled air quality scenario and population as what existed in the
epidemiologic study situation. Listed below are several aspects of that
relationship.
Use ofareawide averaae concentrations: The use of areawide averaaes can miss
important patterns of exposure within urban study areas introducing uncertainty
It is difficult to quantitatively characterize the direction and
magnitude the uncertainty in monitor averaging might have
on risk estimates. The issue could be a greater concern in
large urban areas which may exhibit greater variation in O3
levels compared to small urban areas due to diverse
sources, topography, and patterns of commuting. In
addition, populations living near heavily-trafficked
roadways may experience different patterns of exposure
relative to more generalized urban populations (both for O3
and co-pollutants such as PM2.5). Further, while there is
increased uncertainty in the response at lower
concentration levels, it remains difficult to characterize
5A-6
-------
Analysis Element
Limitations/Uncertainty identified in 2014 HREA
2014 Uncertainty Characterization and Newly Available
Information for Current Review
into the epi study effect estimates and accordingly into the C-R functions applied
in the HREA.
Monitor locations used for area wide averaaes: For some of the HREA analyses,
the locations of the ambient air monitors used to characterize air quality in the
HREA urban study areas do not necessarily match directly with the locations of
monitors used in the original epidemiological study. This may be due to
differences in the monitors operating during and used in the study and those for
which data are available in the years included in the HREA. This may additionally
occur due to the use of CBSAs for the HREA study area, given that CBSAs are
generally larger areas than the epi study areas.
PoDulation Residence and Activity: Differences in the residences and activity
patterns of the simulated population and the epi study population can contribute
uncertainty to risk estimates given the relationships between individual activity and
exposure to pollutants in ambient air are not accounted for in an epi study. For
instance, in one HREA study area, the O3 C-R functions were based on an
epidemiological study in a region (northern Connecticut and Springfield) that did
not encompass the actual urban study area assessed for risk (Boston).
Another area of uncertainty relates to the location of exposure events vs location
of the ambient air monitors and the relationship of the associated ME
concentrations vs ambient air monitors.
All of these can contribute to differences between the HREA and the
epidemiologic studies in the relationship between ambient air monitor
concentration and population exposure, which can contribute uncertainty to the
risk estimates.
whether there are known and quantifiable biases in these
low concentrations.
Regardless, we expect there to be similar uncertainties in
appropriately and accurately representing hypothetical
ambient air conditions used in concert with C-R functions
previously used and any functions derived from newly
available epidemiologic studies identified in the current
review.
Differences in population representation in the risk
assessment compared with the population in O3
epidemiologic studies could have low to moderate
magnitude of impact on the estimated risks and potentially
lead to instances of over and underestimations (HREA,
Table 7-4). We expect there to be similar uncertainties in
population representation when using any newly available
information for the current review.
Population baseline
incidence of health
outcome being assessed
At-risk populations: To some extent, differences in risk factors for the outcomes
being quantified are accounted for by using baseline incidence rates. Uncertainty
can be introduced into the characterization of baseline incidence in varying ways
(e.g., error in reporting incidence for specific endpoints, mismatch between the
spatial scale in which the baseline data were captured and the level of the risk
assessment).
We would anticipate that sources of uncertainty related to
baseline incidence (e.g., potential mismatch between the
spatial scale of reporting in epidemiology studies versus
risk modeling) would still apply if an updated analysis were
completed.
Concentration- Response
(C-R) functions
Use of effect estimates obtained from epidemioloav studies as the basis for C-R
functions: Exposure measurement error combined with other factors (e.a.
magnitude of the effect, sample size, controls for confounding variables,
The HREA recognized that the uncertainty in these
features associated with the O3 C-R functions could have a
moderate impact on risk estimates and, in some instances,
5A-7
-------
Analysis Element
Limitations/Uncertainty identified in 2014 HREA
2014 Uncertainty Characterization and Newly Available
Information for Current Review
consideration for effect modification) can affect the statistical models and
associated effect estimates obtained from O3 epidemiological studies. Uncertainty
in effect estimates due to these influential factors contributes to uncertainty in the
O3 C-R functions used to estimate risk. Consequently, this introduces uncertainty
to the epidemiological-based risk estimates. See discussion in 2014 HREA (p. 7-
43) regarding statistical fit of the O3 C-R functions.
Shape of the C-R curve at lower concentrations: The shape of the curve at the
most prevalent ambient air concentrations can have an important influence impact
on the risk estimates. Most of the population will experience relatively low ambient
air concentrations compared with a lesser proportion of the population
experiencing concentrations having a high level of risk. The 2013 ISA indicates
reduced certainty in the shape of O3 C-R functions at lower ambient air
concentrations due to lesser prevalence of these concentrations in the
epidemiological studies (2014 HREA, pp. 7-43 to 7-44; 2013 ISA, section 2.5.4.4).
As a result, the HREA provides estimates of epidemiology-based mortality risks
using the entire distribution of ambient O3, as well as providing estimates of
mortality associated with specific ambient O3 concentrations.
Specifvina laa structure (short-term exposure studies): There is uncertainty
associated with specifying the exact lag structure to use in modeling short-term
03-attributable mortality and respiratory-related morbidity. Most studies examining
different lag models suggest that O3 effects occur within a few days of exposure
(see O3 ISA, section 2.5.4.3). While the nature of an ideal lag model remains
uncertain, we consider this uncertainty to be relatively small in magnitude
compared with other the identified uncertainties.
C-R function for lona-term (seasonal averaae 1-hr dailv max) mortality: There is
also uncertainty about the extent to which mortality estimates based on the long-
term metric in Jerrett et al. (2009) (i.e., seasonal average of 1-hour daily maximum
concentrations) reflects associations with long-term average O3 versus repeated
occurrences of elevated short-term concentrations. For example, the CASAC
concluded that "[i]n light of the potential nonlinearity of the C-R function for long-
term exposure reflecting a threshold of the mortality response, the estimated
number of premature deaths avoidable for long-term exposure reductions for
several levels need to be viewed with caution" (Frey, 2014a, p. 3).
could result in either over- or underestimation of health
risks.
Of greatest importance is the uncertainty in risks estimates
for low ambient O3 concentrations. The PA recognizes a
greater public health concern for the risk of adverse 03-
attributable effects at higher ambient O3 concentrations
(which drive higher exposure concentrations, section 3.2.2
of the 2014 PA), compared to risks associated with lower
concentrations. This suggests that application of the C-R
function at the lowest ambient O3 concentrations,
combined with instances of increased low concentrations
resulting from the air quality adjustment approach (see
above), could potentially contribute to over-estimation of
risks. A broader impact of this uncertainty that is discussed
in the last review is associated with the public health
importance of the increases in relatively low O3
concentrations following air quality adjustment (80 FR
65316-17, October 26, 2015). To the extent adverse 03-
attributable effects are more strongly supported for higher
ambient concentrations, the impacts on risk estimates of
increasing low O3 concentrations (an impact of reductions
in some 63 precursors) reflect an important source of
uncertainty in the AO epidemiologic risk estimates (80 FR
65316-17, October 26, 2015).
While it is possible that different C-R relationship shapes
could be considered in addition to the previously used
approach of apportioning the contribution of particular
levels to the risk estimates, we expect there to be similar
uncertainties in the O3 C-R functions when using any newly
available information, approaches, or tools identified for the
current review.
5A-8
-------
Analysis Element
Limitations/Uncertainty identified in 2014 HREA
2014 Uncertainty Characterization and Newly Available
Information for Current Review
Lack of C-R functions that have addressed potential for influence of co-pollutants:
The inclusion or exclusion of co-pollutants in epidemiologic study models may
confound, or in other ways, impact the O3 effect estimates reported in the epi
studies in those instances where other pollutants are causally associated with the
endpoint of interest. Regarding PM as one copollutant, the O3 ISA notes that
across studies where its role was assessed, the potential impact of co-pollutants
such as PM on 03-mortality risk estimates tended to be much smaller than the
variation in 03-mortality risk estimates across epi study cities.
5A-9
-------
Appendix 5B
-------
Appendix 5B. Limitations and uncertainties of the air quality, exposure and risk analyses developed in the last review of
secondary standard, and consideration of related newly available information and tools. Drawn from the 2014
WREA, 2014 PA; notices of proposed and final decisions; and, response to comments document for the review.
Analysis Element
Limitations/Uncertainty Identified in Last Review
Conclusions from Last Review and Newly
Available Information for Current Review
[Section 5.2.1.3] W126-based Cumulative Exposure Estimates for Class I Areas (based on Air Monitoring Data)
Ambient air monitoring data
for O3
The monitoring dataset used was for the period from 1998 through 2013 (80 FR 65385,
October 26, 2018). The data set included SLAMS monitors as well as CASTNET
monitors, thus providing extended representation in rural areas. The monitoring season
varies across states in length from May to September to year-round, with duration
intended to capture the highest concentration periods, thus including highest 3-month
period needed for derivation of W126 index values.
Overall O3 measurements are of high quality and
have low uncertainty (2014 WREA, Section 4.4).
Ambient air monitoring data are now available for
more recent years, e.g., through the 2017 monitoring
year.
Class I area representation
by monitoring sites
This analysis focused on monitors sited in or within 15 km of a Class I area for which
any of the years in the time period had a W126 index value above 15 ppm-hrs (80 FR
65385, October 26, 2015). The 15 km distance was selected as a natural breakpoint in
distance of O3 monitoring sites from Class I areas and as still providing similar
surroundings to those occurring in the Class I area. We note that given the strict
restrictions on structures and access within some of these areas, it is common for
monitors intended to collect data pertaining to air quality in these types of areas to be
sited outside their boundaries. The analysis focused on those sites for which at least
one 3-year period between 1998 and 2013 included a 3-year W126 value at/above 15
ppm-hrs (80 FR 65385, October 26, 2015).
The 17 locations in this analysis represent nearly 25%
of the approximately 70 Class I areas for which there
are ambient air monitors within 15 km, and
approximately 10% of the approximately 160 Class I
areas in the U.S. (80 FR 65385, October 26, 2015).
There is an O3 monitor within approximately 24 of
Class I areas (somewhat less than 15%), and a
monitor in or within 15 km of approximately 70 of them
(somewhat fewer than half) (80 FR 65385, October
26, 2015). More recent monitoring data may include
additional sites.
[Section 5.2.1.3] W126-based Cumulative Exposure Estimates for O3 Monitoring Sites across the U.S. with Design Values at/below 75,70,65 and 60 ppb
Ambient air monitoring data
for O3
The monitoring dataset used was for the period from 1998 through 2013 (80 FR 65400,
October 26, 2015; Wells, 2015). The data set included SLAMS monitors, which are
largely focused in urban and suburban areas, as well as CASTNET monitors, which are
located in rural areas, thus providing extended representation in rural areas (as
summarized in section 2.2 above). The monitoring network in some areas of the
Western U.S. is much less dense than in the eastern portions of the U.S. and the west
coast states (Wells, 2015, Figures 1 and 2). The monitoring season varies across
states in length from May to September to year-round, with duration intended to
capture the highest concentration periods, thus including highest 3-month period
needed for derivation of W126 index.
Overall O3 measurements are of high quality and
have low uncertainty (2014 WREA, Section 4.4). Data
are now available through the more recent 2017
monitoring year, e.g., four more 3-year periods
extending through the 2015-2017 time period are now
available. Data are also available now for a few
additional sites in Montana and Wyoming (Figure 2-3
above).
Nationwide representation
by monitoring sites
The analysis included 1,430 monitoring sites with sufficient data to derive valid air
quality metrics for at least one 3-year period from 2001 to 2013 (Wells, 2015). During
the then-most recent 3-year period (2011-2013), there were more than 500 monitoring
sites that would meet the now-current standard of 70 ppb. These monitors were
Given the reductions in O3 concentrations that have
occurred since then (see section 2.3 above), it is likely
there are more sites that meet the now-current
5B-1
-------
Analysis Element
Limitations/Uncertainty Identified in Last Review
Conclusions from Last Review and Newly
Available Information for Current Review
distributed across all nine NOAA climatic regions and 46 of the 50 states. Across all 11
3-year periods of data over the complete time period, there were nearly 4,000 site-time
period instances for which the now-current standard of 70 ppb would have been met.
standard in an update of such an analysis for the four
more recent 3-year periods now available.
[Section 5.2.1.1] National and Regional/Urban Estimates of 03-attributable Impacts for Model-adjusted O3 Concentrations in Nine NOAA Regions
Ambient Air Concentrations
Ambient air monitoring data
The monitoring dataset used was for the 3-year period from 2006 through 2008
(WREA, Table 4-5).
Overall O3 measurements are of high quality and
have low uncertainty (2014 WREA, section 4.4). Data
are now available for the period 2015-2017.
Approach used to derive
factors to adjust air quality
to just meet then-existing
standard
Modelina Platforms and Approaches: Model predictions from the Community Multiscale
Air Quality (CMAQ) model, like all deterministic photochemical models, have both
parametric and structural uncertainty associated with them. Higher Order Decoupled
Direct Method (HDDM) allows for the efficient approximation of O3 concentrations
under alternate emissions scenarios. This approximation is less accurate for larger
emissions perturbations, especially under nonlinear chemistry conditions (WREA,
Table 4-5).
Medium magnitude of impact potentially resulting from
both under- and over-estimation of ambient
concentrations. Updated modeling platforms are
available since the 2014 WREA, e.g., the CAMxv6.5
photochemical model includes updated chemical
mechanisms for O3 formation and destruction
pathways. Somewhat more recent emissions,
meteorology and international transport information is
available (e.g., for 2016).
As the adjustment is applied to all monitor locations in
each region, the adjustment results in broad regional
reductions in O3, including at some monitors that were
already meeting or below the target level. Thus, the
adjustments performed to develop a scenario meeting
a target level at the highest monitor in each region
resulted in substantial reduction below the target level
in some areas of the region. This result at the
monitors already well below the target indicates an
uncertainty with regard to air quality expected from
specific control strategies that might be implemented
to meet a particular target level (80 FR 65375,
October 26, 2015). Adjustments made across smaller
areas might reduce this uncertainty.
Application of HDDM sensitivities to ambient data: there is uncertainty in the statistical
regressions used to relate O3 response to emissions perturbations with ambient O3
concentrations for every season, hour-of-the-day, and monitor location (WREA, Table
4-5).
Emissions Reduction Assumptions: Assumption of across-the-board emissions
reductions: Ozone response is modeled for across-the-board reductions in U.S.
anthropogenic NOx. These across-the-board cuts do not reflect actual emissions
control strategies. The form, locations, and timing of emissions reductions that would
be undertaken to meet various levels of the O3 standard are unknown. The across-the-
board emissions reductions bring levels down uniformly across time and space to show
how O3 would respond to changes in ambient levels of precursor species but do not
reflect spatial and temporal heterogeneity that may occur in local and regional
emissions reductions (WREA, Table 4-5).
Concentration Adiustment: Adjustments were applied independently for each of the
nine NOAA climate regions in continental U.S. such that the highest monitor location in
each region just met the then-existing standard (WREA, Table 4-5). In regions where
the air quality adjustment was applied, it was based on emissions reductions
determined necessary for the highest monitor in that region to just equal the existing
standard or the W126 target for the scenario. Concentrations at all other monitor
locations in the region were also adjusted based on the same emissions reductions
assumptions.
5B-2
-------
Analysis Element
Limitations/Uncertainty Identified in Last Review
Conclusions from Last Review and Newly
Available Information for Current Review
Approach used to derive
factors to adjust air quality
to just meet the 3-year
W126 targets (15, 11, and
7 ppm-hr)
Model-based adiustments: Beainnina with concentrations at monitor locations that had
been adjusted to just meet the then-existing standard, further adjustments were made
at all sites in each NOAA region in which at least one site was not already at/below the
target W126 value for that scenario (2014 WREA, section 4.3.4.1). In such regions, the
adjustment made at all sites was that determined necessary for the highest monitor in
that region to just equal the W126 target.
See above.
Approach used to spatially
interpolate ambient monitor
concentrations to grid cells
Spatial interpolation techniaue: Voronoi Neiahbor Averaaina (VNA) was used to
estimate O3 concentrations in unmonitored areas (as summarized in section 5.2.1.1
above). The uncertainty tends to increase with greater distance from the monitoring
sites as the VNA estimates are weighted based on distance from neighboring
monitoring sites. Thus, there is less uncertainty in the VNA estimates near urban areas
with more dense monitoring networks, and more uncertainty in sparsely populated
areas where monitors are further apart, such as in the Western U.S. (2014 WREA,
Table 4-5).
The uncertainty in this approach could lead to both
under- and over-estimation of ambient concentrations.
However, the magnitude of potential impact to
exposure and risk estimates ranges from low to
moderate, with greatest uncertainties when
interpolating large distances between monitors (2014
WREA, Table 4-5). Several other methods are
available, with associated limitations.
Impacts on Tree Growth at Species- and Ecosystem-level
Response estimates for
controlled exposures
Robust and well-established E-R functions for RBL are available for eleven tree
species in the seedling growth stage: black cherry, Douglas fir, loblolly pine, ponderosa
pine, quaking aspen, red alder, red maple, sugar maple, tulip poplar, Virginia pine, and
white pine (2013 ISA; 2014 PA; 80 FR 65371-73, 65383-65384, 75393-65395, October
26, 2015). The data for these species come from extensive controlled studies in open
top chambers (OTCs), with most species studied multiple times under a wide range of
exposure and/or growing conditions
New field-based studies available in the last review
qualitatively strengthened support for and confidence
in the evidence from the OTC studies providing
additional evidence that 03-induced tree seedling
biomass loss effects observed in chambers also
occurs in the field (2014 PA, pg. 1-29 to 1-30).
Species-specific E-R
functions
Robust composite species-specific E-R functions were developed for each of the 11
tree species (above) based on the separate E-R functions for each combination of
species, exposure condition and growing condition scenario (2013 ISA, section 9.6.1).
The species-specific composite E-R functions have been successfully used to predict
the biomass loss response from tree seedling species over a range of cumulative
exposure conditions (2013 ISA, section 9.6.2). A 12th species-specific E-R function was
considered but not given the same emphasis as the other eleven, as it lacked the
robust basis of the others given that its underlying data were from a single gradient
study that did not control for O3 and climatic conditions, as contrasted with the more
well controlled OTC exposure studies (Frey, 2014c, p. 10, 80 FR 65372, October 26,
2015).
Shape of E-R function: Relative biomass loss estimates are hiahlv sensitive to the
parameters in the E-R function. Some species are represented by one study, other
species by many studies (WREA, Table 6-27).
Sensitivity analyses showed high intraspecific and
interspecific variability. Among the species for which
robust E-R functions are available are a few very
sensitive species and several with little or no O3
sensitivity. It is unknown how well this reflects the
larger suite of tree species in the U.S. Potential
influence on risk estimates estimated to have high
magnitude (WREA, Table 6-27).
5B-3
-------
Analysis Element
Limitations/Uncertainty Identified in Last Review
Conclusions from Last Review and Newly
Available Information for Current Review
Absence of functions for manv sensitive species: Robust E-R functions are not
available for the majority of trees in the modeled urban areas and Class 1 areas,
precluding their representation. Study data for other species do not support E-R
development (WREA, Table 6-27).
Use of seedlina functions for adult trees: E-R functions for trees are based on analyses
of tree seedlings, but most biomass impacts are from estimated effects to adult trees
(WREA, Table 6-27).
National distribution of species with E-R functions: While the available robust E-R
functions are for species representing only a small fraction (0.8 percent) of the total
number of native tree species in the contiguous U.S. (1,497), this small subset includes
eastern and western species, deciduous and coniferous species, and species that grow
in a variety of ecosystems and represent a range of tolerance to O3 (2013 ISA, section
9.6.2; 2014 WREA, section 6.2, Figure 6-2, Table 6-1). The range of each species is
based on data from USFS and used to specify presence/absence of each species
nationally and, in ecosystem-level analysis were used to scale biomass loss by
proportional presence of each species (WREA, Table 6-27).
Species distribution in urban case studv areas and availability of E-R functions: E-R
functions are available for only small portion of trees in the urban case study areas.
Eighty to 90 percent of the total trees in the urban case study areas are excluded from
the analysis as they are species for which we do not have E-R functions; we have
some data indicating sensitivity for two of these species.
Additional sensitive species are likely to exist in U.S.
Therefore, total tree biomass impacts are likely
underestimated, with medium to high potential
magnitude of impact (WREA, Table 6-27). It is not
known yet if there would be robust E-R functions
available for additional tree species in this current
review.
Generally, RBL estimates in tree seedlings are
comparable to adult tree estimates, with a few
exceptions such as black cherry. Some E-R functions
overestimate and some underestimate RBL in adult
trees, with low to medium potential magnitude of
impact (2014 WREA, Table 6-27)).
The magnitude of the influence is dependent on the
community composition in each area. Magnitude of
potential influence on national-scale risk estimates
estimated to be low to medium, and medium to high
for urban case studies (WREA, Table 6-27).
It is unclear whether robust E-R functions will be
available in this review for additional species.
Species distributions
Tree basal area estimates used to assess laraer scale ecosystem effects: Estimates of
basal area were modeled by the U.S. Forest Service's Forest Health Technology
Enterprise Team (FHTET) at a scale of 240 m2. These values were aggregated to the
144(12x12) km2 CMAQ grid.
Assumption of constant forest composition: Forest and Aariculture Sector Optimization
Model with greenhouse gases (FASOMGHG) modeling (used for the urban case study
analyses) does not reflect changes in tree species mixes within a forest type made by
natural adaptation and adaptive management by landowners due to O3. Less sensitive
tree species may gain relative advantage over more sensitive species. The magnitude
of potential influence of associated uncertainties on risk estimates is estimated to be
low (WREA, Table 6-27, p. 6-70).
The magnitude of the potential influence of the
associated uncertainty on national scale risk
estimates is expected to be low to medium (WREA,
Table 6-27). While USDA's FHTET has been working
on refining its model, the effect of these refinements
on risk estimates, though variable, would likely be
small (WREA Table 6-27).
While updates to FASOM or FASOMGHG models
may be available, we do not expect there to be
appreciable improvements in scaling up of effects or
in incorporation of changes in forest composition.
5B-4
-------
Analysis Element
Limitations/Uncertainty Identified in Last Review
Conclusions from Last Review and Newly
Available Information for Current Review
Crop Yield Impacts
Response estimates for
controlled exposures
Experimental data: There is strona evidence for established E-R functions for 10 crops
(barley, field corn, cotton, kidney bean, lettuce, peanut, potato, grain sorghum, soybean
and winter wheat). The established E-R functions for relative yield loss (RYL) were
developed from OTC-type experiments from the National Crop Loss Assessment
Network (NCLAN) (2013 ISA, section 9.6.3; 2014 WREA, section 6.2; 2014 PA, Figure
5-4 and section 6.3; 80 FR 65372, October 26, 2015). These crops were originally
selected for study based on their significant role among U.S. commodity crops
nationwide (e.g., representing approximately 85% of the commodity crops grown in the
U.S. in the 1980s). Data newly available in the 2015 review continued to confirm earlier
findings, leading to the ISA conclusion of little new evidence that crops are becoming
more tolerant of 03 (U.S. EPA, 2006a; U.S. EPA 2013).
It is not clear what percentage of the commodity crops
grown today the evaluated species represent. Also, it
is not clear to what degree crop sensitivities may have
changed over time due to genetic modification or
change in varieties planted.
Species-specific E-R
functions
Shape of E-R function: Crop vield loss estimates are hiahlv sensitive to the parameters
in the E-R function. Some functions are based on one study and others on many
studies (WREA, Table 6-27).
Sensitivity analyses for 10 crops (in 54
studies) showed high intraspecific and interspecific
variability It is unknown how well the set of species
with E-R functions reflects the larger suite of crops in
the U.S (WREA, Table 6-27).
Agricultural and Timber Market Impacts
Approach to estimating
impacts on agricultural and
timber markets
Use of median parameters for crop species E-R functions used to assess national
aaricultural impacts (in FASOM): In addition to the robust E-R functions developed for
the 10 commodity crops above, this modeling used the median E-R function for
oranges, rice, and tomatoes, three species for which E-R functions in terms of W126
are not available (2014 WREA, Table 6-27, p. 6-69).
Using alternative E-R functions would result in lower
or higher O3 impacts on crop and tree species
biomass productivity, potentially affecting economic
equilibrium outcomes (2014 WREA, Table 6-27).
Crop proxv and forest tvpe assumptions: Actual impacts mav differ from those of the
crop proxy or the forest type as the crops/tree species modeled are only a subset of
species present in U.S. agriculture and forestry systems. Further, FASOMGHG
modeling used a simple average of tree RYLs for all forest types within a region (2014
WREA, Table 6-27).
The extent to which updates to FASOMGHG address
this uncertainty is yet to be examined.
Omission of aariculture/ forestry on public lands: The model used (FASOMGHG) does
not include public lands (2014 WREA, Table 6-27).
International trade proiections in FASOMGHG: FASOMGHG reflects future
international trade projections by USDA based on recent O3 conditions. Soybeans and
wheat are major crop exports and have relatively large responses to O3, which are not
reflected in the trade projections (2014 WREA, Table 6-27).
Because public lands are not affected within the
model, the estimates of changes in consumer and
producer surplus would likely be higher if public lands
were included (2014 WREA, Table 6-27).
5B-5
-------
Analysis Element
Limitations/Uncertainty Identified in Last Review
Conclusions from Last Review and Newly
Available Information for Current Review
Overall effects on aaricultural yields and producer and consumer surplus depend on
the ability of producers/farmers to substitute other crops that are less O3 sensitive, and
the responsiveness, or elasticity, of demand and supply (U.S. EPA, 2014b, section
6.5). The WREA discusses multiple areas of uncertainty associated with the
crop yield loss estimates, including those associated with the model-based adjustment
methodology as well as those associated with the projection of yield loss using the
FASOMGHG at the estimated O3 concentrations (U.S. EPA, 2014b, Table 6-27,
section 8.5). Because the W126 index estimates generated in the air quality scenarios
are inputs to the vegetation risk analyses for crop yield loss, any uncertainties in the air
quality scenario estimation of W126 index values are propagated into those analyses
(U.S. EPA, 2014b, Table 6-27, section 8.5). Therefore, the air quality scenarios in the
crop yield analyses have the same uncertainties and limitations as in the biomass loss
analyses (summarized above), including those associated with the model-based
adjustment methodology (U.S. EPA, 2014b, section 8.5).
While having sufficient crop yields is of high public
welfare value, important commodity crops are typically
heavily managed to produce optimum yields.
Moreover, based on the economic theory of supply
and demand, increases in crop yields would be
expected to result in lower prices for affected crops
and their associated goods, which would primarily
benefit consumers. These competing impacts on
producers and consumers complicate consideration of
these effects in terms of potential adversity to the
public welfare (U.S. EPA, 2014c, sections 5.3.2 and
5.7). (80 FR 65379, October 26, 2015).
Carbon Sequestration
Species-specific estimates
Functions for estimatina carbon seauestration: The functions applied in the models to
estimate carbon sequestration are uncertain and vary by species. Pollution removal is
calculated based on field, pollution concentration, and meteorological data. The
pollution removal functions in iTree are from Nowak et al. (2006).
This uncertainty was judged to have medium
magnitude of potential influence on risk estimates
(2014 WREA, Table 6-27). It is not clear if updates to
these models have reduced this uncertainty.
National-scale estimates
Carbon sequestration
estimates in small set of
urban areas (using iTree
model)
Representation and distribution of trees within assessed urban areas: The base
inventory of urban trees, including species and distribution, in iTree has uncertainty.
The iTree model estimates are based on tree growth and pollution removal functions
that are specific to the forest structure in each urban area, including the species
composition, number of trees, and diameter distribution of trees. Of the 11 species with
E-R functions, only 2-3 species were in each urban area, comprising at most 18.5% of
total tree population (2014 WREA, section 6.6).
The urban tree inventories included in the iTree
analyses are based on field counts and
measurements of trees in the specific urban areas
analyzed. Although such data are generally
considered less uncertain than modeled tree
inventories, any associated uncertainties are
propagated into the estimates of carbon sequestration
and pollution removal based on those inventories
(2014 WREA, Table 6-27).
Pollutant Removal
Pollutant removal
Estimates in small set
urban areas (using iTree
model)
Estimation of pollutant removal: The functions applied in iTree to estimate arowina
trees' removal of some common air pollutants are uncertain and vary by species.
Assumption of zero pollutant emissions: Manv tree species are bioaenic sources of
volatile organic compounds (VOC) that contribute to formation of O3. Additional VOC
emissions associated with biomass gains are not addressed.
Magnitude of potential influence of uncertainty on risk
estimates estimated to be medium (WREA, Table 6-
27). The availability of updated removal functions or
functions addressing potential O3 formation is not yet
known.
[Section 5.2.1.2] Foliar Injury Analyses
5B-6
-------
Analysis Element
Limitations/Uncertainty Identified in Last Review
Conclusions from Last Review and Newly
Available Information for Current Review
Associating foliar injury
data with CMAQ-generated
O3 exposures by grid cell
assignments
Spatial assianment of foliar iniurv biosite data to 12x12 km arids. Because of privacy
laws that require the exact location information of sampling sites to not be made public,
the data were assigned to the CMAQ grid by the USFS, except for data in California,
Oregon, and Washington which were assigned to the CMAQ grid by EPA staff based
on publicly available geographic coordinates, rather than coordinates specific to the
sites. Thus, these data have greater uncertainty (2014 WREA, Table 7-24).
Availability of biosite samplina data: Because samplina was discontinued in some
states prior to this analysis, we did not include data for many western states (Montana,
Idaho, Wyoming, Nevada, Utah, Colorado, Arizona, New Mexico, Oklahoma, and
portions of Texas).
Magnitude of potential influence of this element on
risk estimates was estimated to be medium (WREA,
Table 6-27).
Categorization of biosites
by moisture level
Soil moisture threshold for foliar iniurv: Low soil moisture reduces the potential for foliar
injury, but injury could still occur because plants must open their stomata even during
periods of drought (2014 WREA, Table 7-24).
Spatial resolution of soil moisture data: Some veaetation such as alona riverbanks mav
experience sufficient soil moisture during periods of drought to exhibit foliar injury. In
addition, we did not have soil moisture data for Alaska, Hawaii, Puerto Rico, or Guam
(2014 WREA, Table 7-24).
Time period for soil moisture data: Short-term estimates of soil moisture are hiahlv
variable over time, even from month to month within a single year; yet using averages
to address variability contributes to a potential temporal mismatch between soil
moisture and injury (2014 WREA, Table 7-24).
Drouaht cateaories: The soil moisture cateaories used to derive the foliar iniurv
benchmarks (i.e., wet, normal, and dry) are uncertain (2014 WREA, Table 7-24).
The 2014 WREA estimated this uncertainty to have
medium magnitude of impact on risk estimates (2014
WREA, Table 7-24).
The 2014 WREA estimated this uncertainty to have
medium magnitude of impact on risk estimates (2014
WREA, Table 7-24). More refined spatial data are not
known to be available.
The 2014 WREA estimated this uncertainty to have
low-medium magnitude of impact on risk estimates
(2014 WREA, Table 7-24).
The 2014 WREA estimated this uncertainty to have
unknown magnitude of impact on risk estimates (2014
WREA, Table 7-24).
5B-7
-------
> W C
OQ
3
o
3
3
r-K
P
hd
O
r-K
o>
o
r-K^
o"
3
O-
CZ2
r-K
S
o>
V3
£
CD
VI
a>
fa
>-t
o
3*
fa
fa
3
CE-
?T
Tl
fa
pr
2
O
fa
3
O-
W
3
>-t'
O
3
3
3
3
o
O
>-+>
>
~1
o
c
£L
Tl
fa
fa
3
3
3 3'
T3 OQ
fa fa
S s
(« &-
S £
§
X/l ^
o' e?
3 a
6*
Tl
C
OQ
ct;
c
H— •
X/l
o
r-K
p
to
r-K^
o
o*
VO
2
o
w
Tl
>
-k
to
VO
o
o
to
------- |