&EPA
United State
Environmental Protection
Agency
Policy Assessment for the Review of the
Ozone National Ambient Air Quality
Standards
Second External Review Draft
-------
DISCLAIMER
This draft document has been reviewed by the Office of Air Quality Planning and Standards
(OAQPS), U.S. Environmental Protection Agency (EPA), and approved for publication. This
OAQPS Policy Assessment contains preliminary conclusions of the staff of the OAQPS and does
not necessarily reflect the views of the Agency. This draft document is being circulated to
facilitate discussion with the Clean Air Scientific Advisory Committee to inform the EPA's
consideration of the ozone National Ambient Air Quality Standards.
This information is distributed for the purposes of pre-dissemination peer review under
applicable information quality guidelines. It has not been formally disseminated by EPA. It
does not represent and should not be construed to represent any Agency determination or policy.
Mention of trade names or commercial products is not intended to constitute endorsement or
recommendation for use.
-------
EPA-452/P-14-002
January 2014
Policy Assessment for the Review of the Ozone National Ambient Air
Quality Standards
Second External Review Draft
U.S. Environmental Protection Agency
Office of Air and Radiation
Office of Air Quality Planning and Standards
Health and Environmental Impacts Division
Ambient Standards Group
Research Triangle Park, North Carolina 27711
-------
This page left intentionally blank.
-------
1 TABLE OF CONTENTS
2 LISTS OF FIGURES vi
3 LIST OF TABLES x
4 EXECUTIVE SUMMARY ES-1
5 1 INTRODUCTION 1-1
6 1.1 PURPOSE 1-1
7 1.2 BACKGROUND 1-3
8 1.2.1 Legislative Requirements 1-3
9 1.2.2 History of O3 NAAQS Reviews 1-5
10 1.2.3 Current O3 NAAQS Review 1-10
11 1.3 GENERAL APPROACH FOR REVIEW OF THE STANDARDS 1-12
12 1.3.1 Approach for the Primary Standard 1-13
13 1.3.1.1 Approach Used in the Last Review 1-14
14 1.3.1.2 Approach for the Current Review 1-17
15 1.3.1.2.1 Consideration of the Scientific Evidence 1-20
16 1.3.1.2.2 Consideration of Exposure and Risk Estimates 1-24
17 1.3.1.2.3 Considerations Regarding Ambient Os Concentration Estimates
18 Attributable to Background Sources 1-25
19 1.3.2 Approach for the Secondary Standard 1-26
20 1.3.2.1 Approach Used in the Last Review 1-27
21 1.3.2.2 Approach forthe Current Review 1-32
22 1.3.2.2.1 Consideration of the Scientific Evidence 1-35
23 1.3.2.2.2 Consideration of Exposure and Risk Estimates and Air Quality
24 Analyses 1-38
25 1.3.2.2.3 Considerations Regarding Ambient Os Concentration Estimates
26 Attributable to Background Sources 1-40
27 1.3.3 Organization of this Document 1-40
28 1.4 REFERENCES 1-41
29 2 O3 MONITORING AND AIR QUALITY 2-1
30 2.1 O3 MONITORING 2-1
31 2.1.1 O3 Monitoring Network 2-1
-------
1 2.1.2 Recent O3 Monitoring Data and Trends 2-2
2 2.2 EMISSIONS AND ATMOSPHERIC CHEMISTRY 2-7
3 2.3 AIR QUALITY CONCENTRATIONS 2-9
4 2.4 BACKGROUND O3 2-11
5 2.4.1 Seasonal Mean Background O3 intheU.S 2-13
6 2.4.2 Seasonal Mean Background O3 in the U.S. as a Proportion of Total O3.2-15
7 2.4.3 Daily Distributions of Background O3 within the Seasonal Mean 2-16
8 2.4.4 Proportion of Background O3 in 12 Urban Case Study Areas 2-20
9 2.4.5 Influence of Background O3 on W126 levels 2-22
10 2.4.6 Estimated Magnitude of Individual Components of Background O3 2-24
11 2.4.7 Summary 2-26
12 2.5 REFERENCES 2-27
13 3 ADEQUACY OF THE CURRENT PRIMARY STANDARD 3-1
14 3.1 EVIDENCE-BASED CONSIDERATIONS 3-1
15 3.1.1 Modes of Action 3-2
16 3.1.2 Nature of Effects 3-5
17 3.1.2.1 Respiratory Effects - Short-term Exposures 3-6
18 3.1.2.3 Total Mortality - Short-term Exposures 3-46
19 3.1.2.4 Cardiovascular effects - Short-term Exposure 3-51
20 3.1.3 Adversity of Effects 3-54
21 3.1.4 Ozone Concentrations Associated With Health Effects 3-58
22 3.1.4.1 Concentrations in Controlled Human Exposure Studies and in
23 Epidemiologic Panel Studies 3-58
24 3.1.4.2 Concentrations in Epidemiologic Studies - Short-term Metrics .... 3-61
25 3.1.4.3 Concentrations in Epidemiologic Studies-"Long-term" Metrics. 3-75
26 3.1.5 Public Health Implications 3-78
27 3.1.5.1 At-Risk Populations 3-78
28 3.1.5.2 Size of At-Risk Populations and Lifestages in the United States ...3-81
29 3.1.5.3 Averting Behavior 3-83
30 3.2 AIR QUALITY-, EXPOSURE-, AND RISK-BASED CONSIDERATIONS
31 3-84
32 3.2.1 Consideration of the Adjusted Air Quality Used in Exposure and Risk
33 Assessments 3-84
34 3.2.2 Exposure-Based Considerations 3-87
ii
-------
1 3.2.3 Risk-Based Considerations 3-97
2 3.2.3.1 Risk of Lung Function Decrements 3-97
3 3.2.3.2 Estimated Health Risks Associated with Short-or Long-Term Os
4 Exposures, Based on Epidemiologic Studies 3-106
5 3.3 CASAC ADVICE 3-115
6 3.4 PRELIMINARY STAFF CONCLUSIONS ON ADEQUACY OF PRIMARY
7 STANDARD 3-117
8 3.5 REFERENCES 3-124
9 4 CONSIDERATION OF ALTERNATIVE PRIMARY STANDARDS 4-1
10 4.1 INDICATOR 4-1
11 4.2 AVERAGING TIME 4-2
12 4.3 FORM 4-5
13 4.4 LEVEL 4-8
14 4.4.1 Evidence-based Considerations 4-8
15 4.4.2 Air Quality-, Exposure-, and Risk-Based Considerations 4-19
16 4.4.2.1 Exposure-Based Considerations 4-19
17 4.4.2.2 Risk-Based Considerations: Lung Function 4-27
18 4.4.2.3 Risk-Based Considerations: Epidemiology-Based Mortality and
19 Morbidity 4-34
20 4.5 CASAC ADVICE 4-42
21 4.6 PRELIMINARY STAFF CONCLUSIONS ON ALTERNATIVE PRIMARY
22 STANDARDS FOR CONSIDERATION 4-43
23 4.7 KEY UNCERTAINTIES AND AREAS FOR FUTURE RESEARCH AND
24 DATA COLLECTION 4-57
25 4.8 SUMMARY OF PRELIMINARY STAFF CONCLUSIONS ON PRIMARY
26 STANDARD 4-60
27 4.9 REFERENCES 4-64
28 5 ADEQUACY OF THE CURRENT SECONDARY STANDARD 5-1
29 5.1 NATURE OF EFFECTS AND BIOLOGICALLY-RELEVANT
30 EXPOSURE METRIC 5-1
31 5.2 FOREST TREE GROWTH, PRODUCTIVITY AND CARBON
32 STORAGE 5-7
33 5.2.1 Evidence-based Considerations 5-7
34 5.2.2 Exposure/Risk-based Considerations 5-20
35 5.3 CROP YIELD LOSS 5-29
iii
-------
1 5.3.1 Evidence-based Considerations 5-29
2 5.3.2 Exposure/Risk-based Considerations 5-32
3 5.4 VISIBLE FOLIAR INJURY 5-35
4 5.4.1 Evidence-based Considerations 5-37
5 5.4.2 Exposure-/Risk-based Considerations 5-44
6 5.5 OTHER WELFARE EFFECTS 5-54
7 5.5.1 Forest Susceptibility to Insect Infestation 5-54
8 5.5.2 Fire Regulation 5-55
9 5.5.3 Ozone Effects on Climate 5-56
10 5.5.4 Additional Effects 5-57
11 5.6 CASAC ADVICE 5-57
12 5.7 PRELIMINARY STAFF CONCLUSIONS ON ADEQUACY OF
13 SECONDARY STANDARD 5-59
14 5.8 REFERENCES 5-65
15 6 CONSIDERATION OF ALTERNATIVE SECONDARY STANDARDS 6-1
16 6.1 INDICATOR 6-1
17 6.2 FORM AND AVERAGING TIME 6-2
18 6.3 LEVEL 6-16
19 6.4 CASAC ADVICE 6-34
20 6.5 PRELIMINARY STAFF CONCLUSIONS ON ALTERNATIVE
21 STANDARD 6-36
22 6.6 SUMMARY OF PRELIMINARY CONCLUSIONS ON THE
23 SECONDARY STANDARD 6-45
24 6.7 KEY UNCERTAINTIES AND AREAS FOR FUTURE RESEARCH
25 AND DATA COLLECTION 6-47
26 6.8 REFERENCES 6-49
27 APPENDICES
28 Appendix 2A. Supplemental Air Quality Modeling Analyses of Background Os 2A-1
29 Appendix 2B. Monitoring Data Analysis of Relationships Between Current Standard and
30 W126 Metric 2B-1
31 Appendix 2C. Inter-annual Variability in W126 Index Values: Annual and 3-Year Average
32 Metrics (2008-2010) 2C-1
33 Appendix 3 A. Modes of Action Summary 3A-1
IV
-------
1 Appendix 3B. Recent Studies of Respiratory-related Emergency Department Visits and
2 Hospital Admissions 3B-1
3 Appendix 3C. At-risk Populations 3C-1
4 Appendix 3D. Air Quality Data for Locations of Key Epidemiological Studies 3D-1
5 Appendix 5 A. Additional Detail on 2006 Ecological Screening Assessment 5 A-1
6 Appendix 6A. Additional Detail on 2006 Ecological Screening Assessment 6A-1
-------
1 LIST OF FIGURES
2 Figure 1-1. Overview of approach to reviewing the primary standard 1-19
3 Figure 1 -2. Overview of approach to reviewing the secondary standard 1-34
4 Figure 2-1. Map of U.S. ambient Os monitoring sites reporting data to EPA during the 2006-
5 2010 period 2-2
6 Figure 2-2. Trend in U.S. annual 4th highest daily maximum 8-hour Os concentrations in ppb,
7 2000 to 2012 2-4
8 Figure 2-3. Map of 8-hour Os design values in ppb for the 2009-2011 period 2-4
9 Figure 2-4. Map of 8-hour Os design values in ppb for the 2010-2012 period 2-5
10 Figure 2-5. Trend in U.S. annual W126 concentrations in ppm-hrs, 2000 to 2012 2-6
11 Figure 2-6. Map of 2009-2011 average annual W126 values in ppm-hrs 2-6
12 Figure 2-7. Map of 2010-2012 average annual W126 values in ppm-hrs 2-7
13 Figure 2-8. Map of 2007 CMAQ-estimated seasonal mean natural background Os levels from
14 zero-out modeling 2-14
15 Figure 2-9. Map of 2007 CMAQ-estimated seasonal mean North American background Os
16 levels from zero-out modeling 2-15
17 Figure 2-10. Map of 2007 CMAQ-estimated seasonal mean United States background Os levels
18 from zero-out modeling 2-15
19 Figure 2-11. Map of U.S. background percent contribution to seasonal mean Os based on 2007
20 CMAQ zero-out modeling 2-18
21 Figure 2-12. Map of apportionment-based U.S. background percent contribution to seasonal
22 mean Os based on 2007 CAMx source apportionment modeling 2-18
23 Figure 2-13. Distributions of absolute estimates of apportionment-based U.S. Background (all
24 site-days), binned by modeled MDA8 from the 2007 source apportionment
25 simulation 2-19
26 Figure 2-14. Distributions of the relative proportion of apportionment-based U.S. Background to
27 total Os (all site-days), binned by modeled MDA8 from the 2007 source
28 apportionment
29 simulation 2-19
30 Figure 2-15. Fractional influence of background sources to W126 levels in four sample
31 locations. Model estimates based on 2007 CMAQ zero-out modeling 2-23
32 Figure 2-16. Differences in seasonal mean Os between the NAB andNB scenarios 2-25
33 Figure 2-17. Percent contribution of U.S. anthropogenic emissions to total seasonal mean MDA8
34 Os levels, based on 2007 source apportionment modeling 2-26
35 Figure 3-1. Modes of action/possible pathways underlying the health effects resulting from
36 inhalation exposure to Os. (Adapted from U.S. EPA, 2013, Figure 5-8) 3-4
37 Figure 3-2. Percent increase in respiratory-related hospital admission and emergency
38 department visits in studies that presented all-year and/or seasonal results 3-35
vi
-------
1 Figure 3-3. Summary of mortality risk estimates for short-term Os and all-cause (nonaccidental)
2 mortality 3-49
3 Figure 3-4. Concentration-response function for asthma hospital admissions over the
4 distribution of area-wide averaged Os
5 concentrations 3-67
6 Figure 3-5. Concentration-response function for pediatric asthma emergency department visits
7 over the distribution of averaged, population-weighted 8-hour 63
8 concentrations 3-69
9 Figure 3-6. Exposure-Response relationship between risk of death from respiratory causes and
10 ambient 63 concentration study metric 3-77
11 Figure 3-7. Percent of children estimated to experience one or more exposures of concern at or
12 above 60, 70, 80 ppb with air quality adjusted to just meet the current standard
13 (averaged over 2006 to 2010) 3-90
14 Figure 3-8. Percent of children estimated to experience one or more exposures of concern at or
15 above 60, 70, 80 ppb with air quality adjusted to just meet the current standard
16 (worst-case year, 2006 to 2010) 3-91
17 Figure 3-9. Percent of children estimated to experience two or more exposures of concern at or
18 above 60, 70, 80 ppb with air quality adjusted to just meet the current standard
19 (Averaged over 2006 to 2010) 3-92
20 Figure 3-10. Percent of children estimated to experience two or more exposures of concern at or
21 above 60, 70, 80 ppb with air quality adjusted to just meet the current standard
22 (Worst-Case Year, 2006 to 2010) 3-93
23 Figure 3-11. Percent of school-aged children (5-18 yrs) estimated to experience one or more
24 days with FEVi decrements > 10, 15, or 20% with air quality adjusted to just meet
25 the current standard (Averaged over 2006 to 2010) 3-100
26 Figure 3-12. Percent of school-aged children (5-18 yrs) estimated to experience one or more
27 days with FEVi decrements > 10, 15, or 20% with air quality adjusted to just meet
28 the current standard (Worst-Case Year from 2006 to 2010) 3-101
29 Figure 3-13. Percent of school-aged children (5-18 yrs) estimated to experience two or more
30 days with FEVi decrements > 10, 15, or 20% with air quality adjusted to just meet
31 the current standard (Averaged over 2006 to 2010) 3-102
32 Figure 3-14. Percent of school-aged children (aged 5-18 yrs) estimated to experience two or
33 more days with FEVi decrements > 10, 15, or 20% with air quality adjusted to just
34 meet the current standard (Worst-Case Year from 2006 to 2010) 3-103
35 Figure 3-15. Percent of all-cause mortality associated with 63 for air quality adjusted to just
36 meet the current standard 3-109
37 Figure 3-16. Estimated (^-associated mortality attributable to days above various area-wide
38 average 63 concentrations, with air quality adjusted to just meet current standard
39 (2007 Model Adjustment Year) 3-110
vn
-------
1 Figure 3-17. Percent of baseline respiratory mortality estimated to be associated with long-term
2 O3 3-111
3 Figure 4-1. Percent of children estimated to experience one or more exposures of concern at or
4 above 60, 70, or 80 ppb for air quality adjusted to just meet the current and potential
5 alternative standards (averaged over 2006 to 2010) 4-21
6 Figure 4-2. Percent of children estimated to experience one or more exposures of concern at or
7 above 60, 70, or 80 ppb for air quality adjusted to just meet the current and potential
8 alternative standards (worst-case year from 2006 to 2010) 4-22
9 Figure 4-3. Percent of children estimated to experience two or more exposures of concern at or
10 above 60, 70, or 80 ppb for air quality adjusted to just meet the current and potential
11 alternative standards (averaged over 2006 to 2010) 4-23
12 Figure 4-4. Percent of children estimated to experience two or more exposures of concern at or
13 above 60, 70, or 80 ppb for air quality adjusted to just meet the current and potential
14 alternative standards (worst-case year from 2006 to 2010) 4-24
15 Figure 4-5. Percent of children estimated to experience one or more Os-induced lung function
16 decrements greater than 10, 15, or 20% for air quality adjusted to just meet the
17 current and potential alternative standards (averaged over 2006 to 2010) 4-28
18 Figure 4-6. Percent of children estimated to experience one or more Cb-induced lung function
19 decrements greater than 10, 15, or 20% for air quality adjusted to just meet the
20 current and potential alternative standards (worst-case year from 2006 to 2010) 4-29
21 Figure 4-7. Percent of children estimated to experience two or more Os-induced lung function
22 decrements greater than 10, 15, or 20% for air quality adjusted to just meet the
23 current and potential alternative standards (averaged over 2006 to 2010) 4-30
24 Figure 4-8. Percent of children estimated to experience two or more Cb-induced lung function
25 decrements greater than 10, 15, or 20% for air quality adjusted to just meet the
26 current and potential alternative standards (worst-case year from 2006 to 2010) 4-31
27 Figure 4-9. Estimates of Total Mortality Associated with Short-Term O?, Concentrations in
28 Urban Case Study Areas (Air Quality Adjusted to Current and Potential alternative
29 standard levels) - Total Risk 4-36
30 Figure 4-10. Estimates of Total Mortality Attributable to Days with 8-Hour Area-Wide 63
31 Concentrations at or above 20, 40, or 60 ppb, Summed Across Urban Case Study
32 Areas (Air Quality Adjusted to Current and Potential alternative standard levels)
33 4-37
34 Figure 4-11. Estimates of Respiratory Hospital Admissions Associated with Short-Term Oj,
35 Concentrations in Urban Case Study Areas (Air Quality Adjusted to Current and
36 Potential alternative standard levels) - Total Risk 4-38
37 Figure 4-12. Estimates of Respiratory Mortality Associated with long-term O?, Concentrations in
38 Urban Case Study Areas (Air Quality Adjusted to Current and Potential alternative
39 standard levels) - Total Risk 4-39
Vlll
-------
1 Figure 4-13. Estimates of Total Mortality Attributable to Days with 8-Hour Area-Wide Os
2 Concentrations at or above 20, 40, or 60 ppb - Risks Summed Across Urban Case
3 Study Areas and Expressed Relative to 75 ppb 4-47
4 Figure 5-1. Relative biomass loss in seedlings for 12 studied species in response to seasonal
5 ozone concentrations in terms of seasonal W126 index values 5-14
6 Figure 5-2. Relationship of tree seedling percent biomass loss with seasonal W126 index.. 5-15
7 Figure 5-3. Association of Os with cottonwood biomass in gradient downwind of New York
8 City 5-17
9 Figure 5-4. Relative biomass loss of Ponderosa Pine for air quality model-adjusted to just meet
10 the current standard 5-24
11 Figure 5-5. Cumulative proportion of sites with a) any foliar injury or b) elevated injury
12 present, by moisture category 5-48
IX
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
LIST OF TABLES
Table 2-1. Seasonal mean MDA8 Os (ppb), seasonal mean apportionment-based USB
contribution (ppb), and fractional apportionment-based USB contribution to total 63
(all site-days) in the 12 REA urban case study areas (%) ..................................... 2-21
Seasonal mean MDA8 63 (ppb), seasonal mean apportionment-based USB
contribution (ppb), and fractional apportionment-based USB contribution to total 63
(site-days > 60 ppb) in the 12 REA urban study areas (%) .................................. 2-21
Table 2-2.
Table 2-3. Fractional contribution of apportionment-based USB in the 12 REA urban study
areas (%), using the means and medians of daily MDA8 fractions (instead of
fractions of seasonal means) 2-21
Table 2-4. Seasonal mean MDA8 03 (ppb), seasonal mean USB contribution (ppb), and
fractional USB contribution to total Os (all site-days) in the 12 REA urban case
study areas
(%) 2-22
Table 3-1. Group mean results of controlled human exposure studies that have evaluated
exposures to ozone concentrations below 75 ppb in young, healthy adults 3-59
Table 3-2. Panel studies of lung function decrements with analyses restricted to 63
concentrations below 75 ppb 3-61
Table 3-3. U.S. and Canadian epidemiologic studies reporting 63 health effect associations in
locations that would have met the current standard during study periods 3-64
Table 3-4. Distributions of daily 8-hour maximum ozone concentrations from highest monitors
over range of 2-day moving averages from composite monitors (for study area
evaluated by Silverman and Ito, 2010) 3-68
Table 3-5. Distribution of daily 8-hour maximum ozone concentrations from highest monitors
over range of 3-day moving averages of population-weighted concentrations (for
study area evaluated by Strickland et al., 2010) 3-70
Table 3-6. Number of study cities with 4 highest 8-hour daily maximum concentrations
greater than 75 ppb, for various cut-point analyses presented in Bell et al. (2006)....
3-75
Table 3-7. Prevalence of asthma by age in the U.S 3-82
Table 4-1. Numbers of epidemiologic study locations likely to have met potential alternative
standards with levels of 70, 65, and 60 ppb 4-13
Table 4-2. Number of study cities with 3-year averages of 4* highest 8-hour daily max
concentrations greater than 70, 65, or 60 ppb, for various cut-point analyses
presented in Bell etal. (2006) 4-15
Table 4-3. Seasonal averages of 1-hour daily max Os concentrations in U.S. urban case study
areas for recent air quality and air quality adjusted to just meet the current and
potential alternative standards 4-18
Table 4-4. Summary of Estimated Exposures of Concern for Potential alternative standard
levels of 70, 65, and 60 ppb in Urban Case Study Areas 4-45
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
Table 4-5.
Table 5-1.
Table 5-2.
Table 5-3.
Table 5-4.
Table 5-5.
Table 5-6.
Table 5-7.
Table 5-8.
Table 6-1.
Table 6-2.
Table 6-3.
Table 6-4.
Table 6-5.
Summary of Estimated Lung Function Decrements for Potential alternative
standard levels of 70, 65, and 60 ppb in Urban Case Study
Areas 4-46
Ozone concentrations associated with effects on tulip poplar in southern
Appalachian Mountains (2001-2003) 5-16
Examples of counties containing Class I areas where recent air quality might be
expected to meet the current standard and where 3-yr W126 index values are above
15 ppm-hrs 5-19
Exposure, risk and ecosystem services analyses related to tree growth, productivity
and carbon storage 5-20
Summary of methodology by which national surface of average W126 index values
was derived for each air quality scenario 5-21
Exposure, risk and ecosystem services analyses related to crop yield 5-33
Visible foliar injury incidence in four National Wildlife Refuges 5-41
Exposure, risk and ecosystem services analyses related to visible foliar injury... 5-45
Benchmark criteria for Os exposure and relative soil moisture used in screening-
level assessment of parks 5-50
Seedling and crop growth or yield reductions estimated for Os exposure over a
season 6-19
Percent of assessed geographic area exceeding 2% weighted relative biomass loss
in WREA air quality scenarios 6-24
Number of Class I areas (of 145 assessed) with weighted relative biomass loss
greater than 2% 6-25
Estimated mean yield loss (and range across states) due to Os exposure for two
important crops 6-26
Estimated effect of Os-sensitive tree growth-related impacts on the ecosystem
services of air pollutant removal and carbon sequestration in five urban case study
areas 6-28
XI
-------
1 EXECUTIVE SUMMARY
2
3 This second draft Policy Assessment (PA) has been prepared by staff in the
4 Environmental Protection Agency's (EPA) Office of Air Quality Planning and Standards
5 (OAQPS) as part of the Agency's ongoing review of the primary (health-based) and secondary
6 (welfare-based) national ambient air quality standards (NAAQS) for ozone (63). It presents
7 analyses and preliminary staff conclusions regarding the policy implications of the key scientific
8 and technical information that informs this review. Preliminary staff conclusions are presented
9 regarding the adequacy of the current standards and, as appropriate, potential alternative
10 standards appropriate for consideration in this review. Staff analyses in this second draft PA are
11 based on the scientific and technical information, as well as uncertainties and limitations related
12 to this information, assessed in other EPA documents, including the scientific assessment
13 presented in the Integrated Science Assessment for Ozone, the second draft Health Risk and
14 Exposure Assessment for Ozone and the second draft Welfare Risk and Exposure Assessment for
15 Ozone. The final PA is intended to "bridge the gap" between the relevant scientific evidence and
16 technical information and the judgments required of the EPA Administrator in determining
17 whether to retain or revise the current standards. Development of the PA is also intended to
18 facilitate advice and recommendations on the standards to the Administrator from an
19 independent scientific review committee, the Clean Air Scientific Advisory Committee
20 (CASAC), as provided for in the Clean Air Act (CAA).
21 The overarching questions in this review, as in all NAAQS reviews, regard the support
22 provided by the currently available scientific evidence and exposure/risk-based information for
23 the adequacy of the current standards and the extent to which the scientific evidence and
24 technical information provides support for concluding that consideration of alternative standards
25 may be appropriate. Comments and recommendations from CASAC and public comments based
26 on review of this draft PA will inform final staff conclusions and the presentation of information
27 in the final PA.
28 Health Effects and Review of the Primary Standard
29 The longstanding and comprehensive evidence base, stronger today than in the last
30 review, documents the effects of Os in ambient air on health. In particular, Os affects the
31 respiratory system, posing greatest hazard to those with respiratory disease and those with
32 highest exposures, including children with asthma. The evidence indicates that higher exposures
33 and repeated occurrence of exposures lead to more severe effects, including increased
34 susceptibility to other respiratory stressors, and that higher exposures lead to greater prevalence
35 of effects among the exposed population. Based on the staff evaluation presented in this draft
ES-1
-------
1 document, staff preliminarily concludes that the currently available evidence and exposure and
2 risk information call into question the adequacy of the current primary standard and that
3 consideration should be given to revising the standard to provide increased public health
4 protection. With regard to potential alternative standards, staff concludes it is appropriate to
5 consider standards with the same indicator, averaging time and form as the current standard with
6 alternative levels within the range from 70 ppb to 60 ppb.
7 In drawing these preliminary conclusions, staff additionally notes that the final decision
8 on the adequacy of the current standard and consideration of potential alternative standards is
9 largely a public health policy judgment to be made by the Administrator, drawing upon the
10 scientific information as well as judgments about how to consider the range and magnitude of
11 uncertainties that are inherent in the scientific evidence and technical analyses.
12 Welfare Effects and Review of the Secondary Standard
13 The longstanding evidence base, strengthened since the last review, documents the
14 welfare-related effects of O^ in ambient air. In particular, O^ affects vegetation and poses risk of
15 related effects on terrestrial ecosystems. Based on the staff evaluation presented in this draft
16 document, staff preliminarily concludes that the currently available evidence and exposure and
17 risk information call into question the adequacy of the current secondary standard and that
18 consideration should be given to revising the standard to provide increased public welfare
19 protection. In considering the level of protection achieved by potential alternative standards,
20 staff preliminarily concludes it is appropriate for the Administrator to judge 63 welfare impacts
21 using the W126-based cumulative seasonal index, defined as an index of the sum of weighted
22 hourly concentrations, cumulated over 12 hours per day (8 am to 8 pm) during the consecutive
23 three-month period within the Os season with the maximum index value. With regard to
24 potential alternative standards, staff preliminarily concludes it is appropriate to consider
25 standards in terms of the W126-based cumulative seasonal metric with a form averaged across
26 three consecutive years and levels extending somewhat above 15 ppm-hrs (e.g., to 17 ppm-hrs)
27 down to 7 ppm-hrs.
28 In drawing these preliminary conclusions, staff additionally notes that the final decision
29 on the adequacy of the current standard and consideration of potential alternative standards is
30 largely a public welfare policy judgment to be made by the Administrator, drawing upon the
31 scientific information as well as judgments about how to consider the range and magnitude of
32 uncertainties that are inherent in the scientific evidence and technical analyses.
ES-2
-------
1 1 INTRODUCTION
2 1.1 PURPOSE
3 The U.S. Environmental Protection Agency (EPA) is presently conducting a review of
4 the primary (health-based) and secondary (welfare-based) national ambient air quality standards
5 (NAAQS) for ozone (Os). The overall plan for this review was presented in the Integrated
6 Review Plan for the O3 National Ambient Air Quality Standards (IRP, U. S. EPA, 2011 a). The
7 IRP also identified key policy-relevant issues to be addressed in this review and discussed the
8 key documents that generally inform NAAQS reviews, including an Integrated Science
9 Assessment (ISA), Risk and Exposure Assessments (REAs), and a Policy Assessment (PA). The
10 PA is prepared by the staff in EPA's Office of Air Quality Planning and Standards (OAQPS). It
11 presents a staff evaluation of the policy implications of the key scientific and technical
12 information in the ISA and REAs for EPA's consideration.l The PA provides a transparent
13 evaluation, and staff conclusions, regarding policy considerations related to reaching judgments
14 about the adequacy of the current standards, and if revision is considered, what revisions may be
15 appropriate to consider.
16 When final, the PA is intended to help "bridge the gap" between the Agency's scientific
17 assessments presented in the ISA and REAs, and the judgments required of the EPA
18 Administrator in determining whether it is appropriate to retain or revise the NAAQS.2 In
19 evaluating the adequacy of the current standard and whether it is appropriate to consider
20 potential alternative standards, the PA focuses on information that is most pertinent to evaluating
21 the basic elements of the NAAQS: indicator,3 averaging time, form,4 and level. These
22 elements, which together serve to define each standard, must be considered collectively in
23 evaluating the health and welfare protection afforded by the Os standards. The PA integrates and
24 interprets the information from the ISA and REAs to frame policy options for consideration by
25 the Administrator. In so doing, the PA recognizes that the selection of a specific approach to
26 reaching final decisions on the primary and secondary Oj standards will reflect the judgments of
27 the Administrator.
1 The terms "staff" and "we" through this document refer to personnel in the EPA's Office of Air Quality Planning
and Standards (OAQPS).
2 American Farm Bureau Federation v. EPA. 559 F. 3d 512, 521 (D.C. Cir. 2009); Natural Resources Defense
Council v. EPA. 902 F. 2d 962, 967-68, 970 (D.C. Cir. 1990).
3 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 for photochemical oxidants is ozone.
4 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 current 8-hour O3 NAAQS is the 3-
year average of the annual fourth-highest daily maximum 8-hour average.
1-1
-------
1 The development of the PA is also intended to facilitate advice to the Agency and
2 recommendations to the Administrator from an independent scientific review committee, the
3 Clean Air Scientific Advisory Committee (CASAC), as provided for in the Clean Air Act. As
4 discussed below in section 1.2.1, the CASAC is to advise not only on the Agency's assessment
5 of the relevant scientific information, but also on the adequacy of the existing standards, and to
6 make recommendations as to any revisions of the standards that may be appropriate. The EPA
7 facilitates CASAC advice and recommendations, as well as public input and comment, by
8 requesting CASAC review and public comment on one or more drafts of the PA.
9 The decision whether to prepare one or more drafts of the PA is influenced by
10 preliminary staff conclusions and associated CASAC advice and public comment, among other
11 factors. Typically, as in this review, staff prepares a second draft PA where the available
12 information calls into question the adequacy of the current standard and analyses of potential
13 alternative standards are developed taking into consideration CASAC advice and public
14 comment. In such cases, a second draft PA includes preliminary staff conclusions regarding
15 potential alternative standards and undergoes CASAC review and public comment prior to
16 preparation of the final PA.5
17 In this second draft of the PA for the review of the Os NAAQS, we consider the scientific
18 and technical information available in this review as assessed in the Integrated Science
19 Assessment for 0$ and Related Photochemical Oxidants (ISA, U.S. EPA, 2013), prepared by
20 EPA's National Center for Environmental Assessment (NCEA), and the second drafts of the
21 quantitative human exposure and health risk assessment and welfare risk assessment documents
22 (HREA, U.S. EPA, 2014a; WREA, U.S. EPA, 2014b). The evaluation and preliminary staff
23 conclusions presented in this second draft PA have been informed by comments and advice
24 received from CASAC in their review of the first draft PA and of the other draft Agency
25 documents prepared for this NAAQS review. Review and comments on this second draft PA
26 from CASAC and the public will inform the final evaluation and staff conclusions in the final
27 PA.
28 Beyond informing the EPA Administrator and facilitating the advice and
29 recommendations of CASAC and the public, the PA is also intended to be a useful reference to
30 all parties interested in the NAAQS review. In these roles, it is intended to serve as a single
31 source of the most policy-relevant information that informs the Agency's review of the NAAQS,
32 and it is written to be understandable to a broad audience.
5 When such analyses are not undertaken, a second draft PA may not be warranted.
1-2
-------
1 1.2 BACKGROUND
2 1.2.1 Legislative Requirements
3 Two sections of the Clean Air Act (CAA) govern the establishment and revision of the
4 NAAQS. Section 108 (42 U.S.C. section 7408) directs the Administrator to identify and list
5 certain air pollutants and then to issue air quality criteria for those pollutants. The Administrator
6 is to list those air pollutants that in her "judgment, cause or contribute to air pollution which may
7 reasonably be anticipated to endanger public health or welfare;" "the presence of which in the
8 ambient air results from numerous or diverse mobile or stationary sources;" and "for which . . .
9 [the Administrator] plans to issue air quality criteria...." Air quality criteria are intended to
10 "accurately reflect the latest scientific knowledge useful in indicating the kind and extent of all
11 identifiable effects on public health or welfare which may be expected from the presence of [a]
12 pollutant in the ambient air . . ." 42 U.S.C. § 7408(b). Section 109 (42 U.S.C. 7409) directs the
13 Administrator to propose and promulgate "primary" and "secondary" NAAQS for pollutants for
14 which air quality criteria are issued. Section 109(b)(l) defines a primary standard as one "the
15 attainment and maintenance of which in the judgment of the Administrator, based on such
16 criteria and allowing an adequate margin of safety, are requisite to protect the public health."6
17 A secondary standard, as defined in section 109(b)(2), must "specify a level of air quality the
18 attainment and maintenance of which, in the judgment of the Administrator, based on such
19 criteria, is requisite to protect the public welfare from any known or anticipated adverse effects
20 associated with the presence of [the] pollutant in the ambient air." 7
21 The requirement that primary standards provide an adequate margin of safety was
22 intended to address uncertainties associated with inconclusive scientific and technical
23 information available at the time of standard setting. It was also intended to provide a reasonable
24 degree of protection against hazards that research has not yet identified. See Lead Industries
25 Association v. EPA. 647 F.2d 1130, 1154(D.C. Cir 1980): American Petroleum Institute v.
26 Costle. 665F.2d 1176, 1186(D.C. Cir. 1981): American Farm Bureau Federation v. EPA. 559 F.
27 3d 512, 533 (D.C. Cir. 2009); Association of Battery Recvclers v. EPA. 604 F. 3d 613, 617-18
28 (D.C. Cir. 2010). Both kinds of uncertainties are components of the risk associated with pollution
29 at levels below those at which human health effects can be said to occur with reasonable
6 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).
7 Welfare effects as defined in section 302(h) (42 U.S.C. § 7602(h)) include, but are not limited to, "effects on soils,
water, crops, vegetation, man-made materials, animals, wildlife, weather, visibility and climate, damage to and
deterioration of property, and hazards to transportation, as well as effects on economic values and on personal
comfort and well-being."
1-3
-------
1 scientific certainty. Thus, in selecting primary standards that provide an adequate margin of
2 safety, the Administrator is seeking not only to prevent pollution levels that have been
3 demonstrated to be harmful but also to prevent lower pollutant levels that may pose an
4 unacceptable risk of harm, even if the risk is not precisely identified as to nature or degree. The
5 CAA does not require the Administrator to establish a primary NAAQS at a zero-risk level or at
6 background concentration levels, see Lead Industries v. EPA, 647 F.2d at 1156 n.51; State of
7 Mississippi v. EPA. 723 F. 3d 246, 255, 262-63 (D.C. Cir. 2013), but rather at a level that
8 reduces risk sufficiently so as to protect public health with an adequate margin of safety.
9 In addressing the requirement for an adequate margin of safety, the EPA considers such
10 factors as the nature and severity of the health effects, the size of sensitive population(s)8 at risk,
11 and the kind and degree of the uncertainties that must be addressed. The selection of any
12 particular approach for providing an adequate margin of safety is a policy choice left specifically
13 to the Administrator's judgment. See Lead Industries Association v. EPA, 647 F.2d at 1161-62;
14 State of Mississippi, 723 F. 3d at 265.
15 In setting primary and secondary standards that are "requisite" to protect public health
16 and welfare, respectively, as provided in section 109(b), EPA's task is to establish standards that
17 are neither more nor less stringent than necessary for these purposes. In so doing, the EPA may
18 not consider the costs of implementing the standards. See generally, Whitman v. American
19 Trucking Associations, 531 U.S. 457, 465-472, 475-76 (2001). Likewise, "[attainability and
20 technological feasibility are not relevant considerations in the promulgation of national ambient
21 air quality standards." American Petroleum Institute v. Costle, 665 F. 2d at 1185.
22 Section 109(d)(l) requires that "not later than December 31, 1980, and at 5-year intervals
23 thereafter, the Administrator shall complete a thorough review of the criteria published under
24 section 108 and the national ambient air quality standards . . . and shall make such revisions in
25 such criteria and standards and promulgate such new standards as may be appropriate
26 Section 109(d)(2) requires that an independent scientific review committee "shall complete a
27 review of the criteria . . . and the national primary and secondary ambient air quality standards . .
28 . and shall recommend to the Administrator any new . . . standards and revisions of existing
29 criteria and standards as may be appropriate . . . ." Since the early 1980's, the Clean Air
30 Scientific Advisory Committee (CASAC) has performed this independent review function. 9
8 As used here and similarly throughout this document, the term population refers to persons having a quality or
characteristic in common, including a specific pre-existing illness or a specific age or life stage.
9 Lists of CASAC members and of members of the CASAC Ozone Review Panel are available at:
http://yosemite.epa.gov/sab/sabpeople.nsf/WebCommitteesSubCommittees/Ozone%20Review%20Panel.
1-4
-------
1 1.2.2 History of O3 NAAQS Reviews
2 Table 1-1 summarizes the O3 NAAQS that the EPA has promulgated to date. In each
3 review, the EPA set the secondary standard at a level identical to the primary standard. These
4 reviews are briefly described below.
5 Table 1-1. Summary of primary and secondary Os NAAQS promulgated during the
6 period from 1971 to 2008.
Final Rule
1971 (36 FR 81 86)
1979 (44 FR 8202)
1 993 (58 FR 13008)
1997 (62 FR 38856)
2008 (73 FR 16483)
Indicator
Total photochemical
oxidants
03
Averaging Time
1 hour
1 hour
Level (ppm)
0.08
0.12
Form
Not to be exceeded more than
one hour per year
Attainment is defined when the
expected number of days per
calendar year, with maximum
hourly average concentration
greater than 0.1 2 ppm, is
equal to or less than 1
The EPA decided that revisions to the standards were not warranted at the time.
03
03
8 hours
8 hours
0.08
0.075
Annual fourth-highest daily
maximum 8-hour
concentration, averaged over 3
years
Form of the standards
remained unchanged relative
to the 1997 standard
9
10
11
12
13
14
15
16
17
18
The EPA first established primary and secondary NAAQS for photochemical oxidants in
1971 (36 FR 8186, April 30, 1971). The EPA set both primary and secondary standards at a level
of 0.08 parts per million (ppm), 1-hr average, total photochemical oxidants, not to be exceeded
more than one hour per year. The EPA based the standards on scientific information contained in
the 1970 Air Quality Criteria for Photochemical Oxidants (U.S. DHEW, 1970). We initiated the
first periodic review of the NAAQS for photochemical oxidants in 1977. Based on the 1978 Air
Quality Criteria for Ozone and Other Photochemical Oxidants (U.S. EPA, 1978), the EPA
published proposed revisions to the original NAAQS in 1978 (43 FR 16962) and final revisions
in 1979 (44 FR 8202). At that time, the EPA revised the level of the primary and secondary
standards from 0.08 to 0.12 ppm and changed the indicator from photochemical oxidants to Os,
and the form of the standards from a deterministic to a statistical form. This statistical form
1-5
-------
1 defined attainment of the standards as occurring when the expected number of days per calendar
2 year with maximum hourly average concentration greater than 0.12 ppm equaled one or less.
3 Following the final decision in the 1979 review, the City of Houston challenged the
4 Administrator's decision arguing that the standard was arbitrary and capricious because natural
5 Os concentrations and other physical phenomena in the Houston area made the standard
6 unattainable in that area. The U.S. Court of Appeals for the District of Columbia circuit (D.C.
7 Circuit) rejected this argument, stating (as noted above) that attainability and technological
8 feasibility are not relevant considerations in the promulgation of the NAAQS. The court also
9 noted that the EPA need not tailor the NAAQS to fit each region or locale, pointing out that
10 Congress was aware of the difficulty in meeting standards in some locations and had addressed
11 this difficulty through various compliance related provisions in the Act. See API v. Costle, 665
12 F.2d 1176, 1184-6 (D.C. Cir. 1981). In 1982, we announced plans to revise the 1978 Air Quality
13 Criteria document (47 FR 11561), and in 1983, we initiated the second periodic review of the O3
14 NAAQS (48 FR 38009). We subsequently published the 1986 Air Quality Criteria for Ozone
15 and Other Photochemical Oxidants (U. S. EPA, 1986) and the 1989 Staff Paper (U. S. EPA,
16 1989). Following publication of the 1986 Air Quality Criteria Document (AQCD), a number of
17 scientific abstracts and articles were published that appeared to be of sufficient importance
18 concerning potential health and welfare effects of O3 to warrant preparation of a Supplement
19 (U.S. EPA, 1992). On August 10, 1992, under the terms of a court order, the EPA published a
20 proposed decision to retain the existing primary and secondary standards. (57 FR 35542). The
21 notice explained that the proposed decision would complete EPA's review of information on
22 health and welfare effects of O3 assembled over a 7-year period and contained in the 1986
23 AQCD and its 1992 Supplement. The proposal also announced EPA's intention to proceed as
24 rapidly as possible with the next review of the air quality criteria and standards for O3 in light of
25 emerging evidence of health effects related to 6- to 8-hour O3 exposures. On March 9, 1993, the
26 EPA concluded the review by affirming its proposed decision to retain the existing primary and
27 secondary standards. (58 FR 13008).
28 In August 1992, we announced plans to initiate the third periodic review of the air quality
29 criteria and O3 NAAQS (57 FR 35542). In December 1996, the EPA proposed to replace the then
30 existing 1-hour primary and secondary standards with 8-hour average O3 standards set at a level
31 of 0.08 ppm (equivalent to 0.084 ppm using standard rounding conventions) (61 FR 65716). The
32 EPA also proposed to establish a new distinct secondary standard using a biologically-based
33 cumulative, seasonal form. The EPA completed this review on July 18, 1997 (62 FR 38856) by
34 setting the primary standard at a level of 0.08 ppm, based on the annual fourth-highest daily
35 maximum 8-hr average concentration, averaged over three years, and setting the secondary
36 standard identical to the revised primary standard. In reaching this decision, the EPA identified
1-6
-------
1 several reasons supporting its decision to reject a potential alternate standard set at 0.07 ppm.
2 Most importantly, the EPA pointed out the scientific uncertainty at lower concentrations and
3 placed significant weight on the fact that no CAS AC panel member supported a standard level
4 set lower than 0.08 ppm (62 FR 38868). In addition to noting the uncertainties in the health
5 evidence for exposure concentrations below 0.08 ppm and the advice of CASAC, the EPA noted
6 that a standard set at a level of 0.07 ppm would be closer to peak background concentrations that
7 infrequently occur in some areas due to nonanthropogenic sources of 63 precursors (62 FR
8 38856, 38868; July 18, 1997).
9 On May 14, 1999, in response to challenges by industry and others to EPA's 1997
10 decision, the U.S. Court of Appeals for the District of Columbia Circuit remanded the 63
11 NAAQS to the EPA, finding that section 109 of the Act, as interpreted by the EPA, effected an
12 unconstitutional delegation of legislative authority. American Trucking Assoc. vs. EPA, 175
13 F.3d 1027, 1034-1040(D.C. Cir. 1999) ("ATA I"). In addition, the court directed that, in
14 responding to the remand, the EPA should consider the potential beneficial health effects of 03
15 pollution in shielding the public from the effects of solar ultraviolet (UV) radiation, as well as
16 adverse health effects. Id. At 1051-53. In 1999, the EPA petitioned for rehearing en bane on
17 several issues related to that decision. The court granted the request for rehearing in part and
18 denied it in part, but declined to review its ruling with regard to the potential beneficial effects of
19 O3 pollution. 195 F3d 4, 10 (D.C Cir., 1999) ("ATA II"). On January 27, 2000, the EPA
20 petitioned the U.S. Supreme Court for certiorari on the constitutional issue (and two other
21 issues), but did not request review of the ruling regarding the potential beneficial health effects
22 of 03. On February 27, 2001, the U.S. Supreme Court unanimously reversed the judgment of the
23 D.C. Circuit on the constitutional issue. Whitman v. American Trucking Assoc., 531 U. S. 457,
24 472-74 (2001) (holding that section 109 of the CAA does not delegate legislative power to the
25 EPA in contravention of the Constitution). The Court remanded the case to the D.C. Circuit to
26 consider challenges to the 03 NAAQS that had not been addressed by that court's earlier
27 decisions. On March 26, 2002, the D.C. Circuit issued its final decision on remand, finding the
28 1997 Os NAAQS to be "neither arbitrary nor capricious," and so denying the remaining petitions
29 for review. American Trucking Associations, Inc. v EPA, 283 F.3d 355, 379 (D.C Cir.,
30 2002)("ATA III").
31 Specifically, in ATA III, the D.C. Circuit upheld EPA's decision on the 1997 Oj standard
32 as the product of reasoned decision-making. The Court made clear that the most important
33 support for EPA's decision was the health evidence and the concerns it raised about setting a
34 standard level below 0.08 ppm. ("the record is replete with references to studies demonstrating
35 the inadequacies of the old one-hour standard", as well as extensive information supporting the
36 change to an 8-hour averaging time). 283 F 3d at 378. The Court also pointed to the significant
1-7
-------
1 weight that the EPA properly placed on the advice it received from CASAC. Id. at 379. The
2 court further noted that "although relative proximity to peak background ozone concentrations
3 did not, in itself, necessitate a level of 0.08, EPA could consider that factor when choosing
4 among the three alternative levels." Id.
5 Independently of the litigation, the EPA also responded to the Court's remand to
6 consider the potential beneficial health effects of O^ pollution in shielding the public from effects
7 of solar (ultraviolet or UV-B) radiation. The EPA provisionally determined that that the
8 information linking changes in patterns of ground-level Os concentrations to changes in relevant
9 patterns of exposures to ultraviolet (UV-B) radiation of concern to public health was too
10 uncertain, at that time, to warrant any relaxation in 1997 63 NAAQS. The EPA also expressed
11 the view that any plausible changes in UV-B radiation exposures from changes in patterns of
12 ground-level Os concentrations would likely be very small from a public health perspective. In
13 view of these findings, the EPA proposed to leave the 1997 8-hour NAAQS unchanged (66 FR
14 57268, Nov. 14, 2001). After considering public comment on the proposed decision, the EPA
15 published its final response to this remand on January 6, 2003, re-affirming the 8-hour 63
16 NAAQS set in 1997 (68 FR 614).
17 The EPA initiated the fourth periodic review of the air quality criteria and 63 standards in
18 September 2000 with a call for information (65 FR 57810). The schedule for completion of that
19 review was ultimately governed by a consent decree resolving a lawsuit filed in March 2003 by
20 plaintiffs representing national environmental and public health organizations, who maintained
21 that EPA was in breach of a mandatory legal duty to complete review of the Os NAAQS within a
22 statutorily-mandated deadline. On July 11, 2007, the EPA proposed to revise the level of the
23 primary standard within a range of 0.075 to 0.070 ppm. (72 FR 37818). Documents supporting
24 this proposed decision included the Air Quality Criteria for Ozone and Other Photochemical
25 Oxidants (U.S. EPA, 2006) and the Staff Paper (U.S EPA, 2007a) and related technical support
26 documents. The EPA also proposed two options for revising the secondary standard: (1) replace
27 the current standard with a cumulative, seasonal standard, expressed as an index of the annual
28 sum of weighted hourly concentrations cumulated over 12 daylight hours during the consecutive
29 3-month period within the Os season with the maximum index value, set at a level within the
30 range of 7 to 21 ppm-hrs, and (2) set the secondary standard identical to the proposed primary
31 standard. The EPA completed the review with publication of a final decision on March 27, 2008
32 (73 FR 16436). In that final rule, the EPA revised the NAAQS by lowering the level of the 8-
33 hour primary Os standard from 0.08 ppm to 0.075 ppm, not otherwise revising the primary
34 standard, and adopting a secondary standard identical to the revised primary standard. In May
35 2008, state, public health, environmental, and industry petitioners filed suit challenging EPA's
36 final decision on the 2008 Os standards. On September 16, 2009, the EPA announced its
1-8
-------
1 intention to reconsider the 2008 O^ standards, and initiated a rulemaking to do so. At EPA's
2 request, the Court held the consolidated cases in abeyance pending EPA's reconsideration of the
3 2008 decision.
4 On January 19, 2010 (75 FR 2938), the EPA issued a notice of proposed rulemaking to
5 reconsider the 2008 final decision. In that notice, the EPA proposed that further revisions of the
6 primary and secondary standards were necessary to provide a requisite level of protection to
7 public health and welfare. The EPA proposed to decrease the level of the 2008 8-hour primary
8 standard from 0.075 ppm to a level within the range of 0.060 to 0.070 ppm, and to change the
9 secondary standard to a new cumulative, seasonal standard expressed as an annual index of the
10 sum of weighted hourly concentrations, cumulated over 12 hours per day (8 am to 8 pm), during
11 the consecutive 3-month period within the O?, season, with a maximum index value set at a level
12 within the range of 7 to 15 ppm-hours. The Agency also solicited CASAC review of the
13 proposed rule on January 25, 2010 and solicited additional CASAC advice on January 26, 2011.
14 After considering comments from CASAC and the public, the EPA prepared a draft final rule,
15 which was submitted for interagency review pursuant to Executive Order 12866. On September
16 2, 2011, consistent with the direction of the President, the Administrator of the Office of
17 Information and Regulatory Affairs ("OIRA"), Office of Management and Budget ("OMB"),
18 returned the draft final rule to the EPA for further consideration. In view of this return and the
19 timing of the Agency's ongoing periodic review of the O?, NAAQS required under Clean Air Act
20 section 109 (as announced on September 29, 2008), the EPA decided to coordinate further
21 proceedings on its voluntary rulemaking on reconsideration with that ongoing periodic review,
22 by deferring the completion of its voluntary rulemaking on reconsideration until it completes its
23 statutorily-required periodic review.
24 In light of EPA's decision to consolidate the reconsideration with the current review, the
25 Court proceeded with the litigation on the 2008 final decision. On July 23, 2013, the D.C. Circuit
26 Court of Appeals upheld EPA's 2008 primary Oj standard, but remanded the 2008 secondary
27 standard to the EPA. State of Mississippi v. EPA. 723 F. 3d 246. With respect to the primary
28 standard, the court first held that the EPA reasonably determined that the existing standard was
29 not requisite to protect public health with an adequate margin of safety, and consequently
30 required revision. Specifically, the court noted that there were "numerous epidemiological
31 studies linking health effects to exposure to ozone levels below 0.08 ppm and clinical human
32 exposure studies finding a causal relationship between health effects and exposure to ozone
33 levels at and below 0.08 ppm". Id. at 257. The court also specifically endorsed the weight of
34 evidence approach utilized by EPA in its deliberations. Id. at 256.
35 The court went on to reject arguments that EPA should have adopted a more stringent
36 primary standard. Dismissing arguments that a single clinical study (properly interpreted by
1-9
-------
1 EPA) to show effects at 0.06 ppm necessitated a standard level lower than that selected, the court
2 noted that this was a single, limited study. Id. at 262. With respect to the epidemiologic evidence,
3 the court accepted EPA's argument that there could be legitimate uncertainty that a causal
4 relationship between Oj, and 8-hour exposures less than 0.075 ppm exists, so that associations at
5 lower levels reported in epidemiologic studies did not necessitate a more stringent standard. Id.
6 at 264-65.10
7 The court also rejected arguments that an 8-hour primary standard of 0.075 ppm failed to
8 provide an adequate margin of safety, noting that margin of safety considerations involved policy
9 judgments by the agency, and that by setting a standard "appreciably below" the level of the
10 current standard (0.08 ppm), the agency had made a reasonable policy choice . Id. Finally, the
11 court rejected arguments that EPA's decision was inconsistent with CAS AC's scientific
12 recommendations because CASAC had been insufficiently clear in its recommendations whether
13 it was providing scientific or policy recommendations, and EPA had reasonably addressed
14 CASAC's policy recommendations. Id. at 269-70.
15 With respect to the secondary standard, the court held that because EPA had failed to
16 identify a level of air quality requisite to protect public welfare, EPA's comparison between the
17 primary and secondary standards for determining if requisite protection for public welfare was
18 afforded by the primary standard was inherently arbitrary. The court thus rejected EPA's
19 determination that the revised 8-hour primary standard afforded requisite protection of public
20 welfare, and remanded the standard to EPA. Id. at 272-73.
21 1.2.3 Current O3 NAAQS Review
22 On September 29, 2008, the EPA announced the initiation of a new periodic review of
23 the air quality criteria for O^ and related photochemical oxidants and issued a call for
24 information in the Federal Register (73 FR 56581, Sept. 29, 2008). A wide range of external
25 experts, as well as EPA staff, representing a variety of areas of expertise (e.g., epidemiology,
26 human and animal toxicology, statistics, risk/exposure analysis, atmospheric science, ecology,
27 biology, plant science, ecosystem services) participated in a workshop. This workshop was held
28 on October 28-29, 2008 in Research Triangle Park, NC. The workshop provided an opportunity
29 for a public discussion of the key policy-relevant issues around which the EPA would structure
30 this Os NAAQS review and the most meaningful new science that would be available to inform
31 our understanding of these issues.
10 The court cautioned, however, that "perhaps more [clinical] studies like the Adams studies will yet reveal that the
0.060 ppm level produces significant adverse decrements that simply cannot be attributed to normal variation in lung
function", and further cautioned that "agencies may not merely recite the terms 'substantial uncertainty' as a
justification for their actions'". Id. at 262, 269 (internal citations omitted).
1-10
-------
1 Based in part on the workshop discussions, the EPA developed a draft IRP outlining the
2 schedule, process, and key policy-relevant questions that would guide the evaluation of the air
3 quality criteria for O?, and the review of the primary and secondary O^ NAAQS. A draft of the
4 integrated review plan was released for public review and comment in September 2009. This IRP
5 was the subject of a consultation with the CASAC on November 13, 2009 (74 FR 54562;
6 October 22, 2009).u We considered comments received from that consultation and from the
7 public in finalizing the plan and in beginning the review of the air quality criteria. The EPA's
8 overall plan and schedule for this review is presented in the Integrated Review Plan for the
9 Ozone National Ambient Air Quality Standards.12
10 As part of the process of preparing the 63 ISA, NCEA hosted a peer review workshop in
11 October 29-30, 2008 (73 FR 56581, September 29, 2008) on preliminary drafts of key ISA
12 chapters. The CASAC and the public reviewed the first external review draft ISA (U.S. EPA,
13 201 Ib; 76 FR 10893, February 28, 2011) at a meeting held in May 19-20, 2011 (76 FR 23809;
14 April 28, 2011). Based on CASAC and public comments, NCEA prepared a second draft ISA
15 (U.S. EPA, 201 Ic; 76 FR 60820, September 30, 2011). CASAC and the public reviewed this
16 draft at a January 9-10, 2012 (76 FR 236, December 8, 2011) meeting. Based on CASAC and
17 public comments, NCEA prepared a third draft ISA (U.S. EPA 2012a; 77 FR 36534; June 19,
18 2012), which was reviewed at a CASAC meeting in September 2012. The final ISA was released
19 in February 2013.
20 The EPA presented its plans for conducting the Risk and Exposure Assessments (REAs)
21 that build on the scientific evidence presented in the ISA, in two planning documents titled
22 Ozone National Ambient Air Quality Standards: Scope and Methods Plan for Health Risk and
23 Exposure Assessment and Ozone National Ambient Air Quality Standards: Scope and Methods
24 Plan for Welfare Risk and Exposure Assessment (henceforth, Scope and Methods Plans).13
25 These planning documents outlined the scope and approaches that staff planned to use in
26 conducting quantitative assessments, as well as, key issues that would be addressed as part of the
27 assessments. We released these documents for public comment in April 2011, and consulted with
28 CASAC on May 19-20, 2011 (76 FR 23809; April 28, 2011). In designing and conducting the
29 initial health risk and welfare risk assessments, we considered CASAC comments (Samet 2011)
30 on the Scope and Methods Plans and also considered public comments. In May 2012, we issued
11 See http ://vo Semite, epa. gov/sab/sabproduct. nsfAVebProj ectsbvTopicC AS AC! OpenView for more information on
CASAC activities related to the current O3 NAAQS review.
12 EPA 452/R-11-006; April 2011; Available:
http://www.epa.gov/ttn/naaqs/standards/ozone/data/2011 04 OzoneIRP.pdf
13 EPA-452/P-11-001 and -002; April 2011; Available:
http://www.epa.gOv/ttn/naaqs/standards/ozone/s o3 2008_pd.html
1-11
-------
1 a memo titled, Updates to Information Presented in the Scope and Methods Plans for the Ozone
2 NAAQS Health and Welfare Risk and Exposure Assessments., that described changes to elements
3 of the scope and methods plans and provided a brief explanation of each change and the reason
4 for it.
5 In July 2012, EPA made the first drafts of the Health and Welfare REAs available for
6 CASAC review and public comment (77 FR 42495, July 19, 3023). The first draft PA was made
7 available for CASAC review and public comment in August 2012. These documents were
8 reviewed by CASAC Os Panel at a public meeting in September 2012. The second draft REAs
9 and PA have been prepared in consideration of CASAC (Frey and Samet, 2012a, 2012b) and
10 public comment and will be reviewed by the CASAC 63 Panel at a public meeting in March
11 2014.
12 1.3 GENERAL APPROACH FOR REVIEW OF THE STANDARDS
13 As described in section 1.1 above, the final PA will present a transparent evaluation and
14 staff conclusions regarding policy considerations related to reaching judgments about the
15 adequacy of the current standards and what, if any, revisions may be appropriate to consider.
16 Preliminary staff considerations and conclusions in this document are based on the available
17 body of scientific evidence assessed in the ISA (U.S. EPA, 2013), exposure and risk analyses
18 presented in the 2nd draft REAs (U.S. EPA, 2014a, b), advice and recommendations from
19 CASAC on the first draft PA and other draft EPA documents in this review, as well as on public
20 comments. When final, this evaluation and associated conclusions on the range of policy options
21 that, in staffs view, could be supported by the available scientific evidence and exposure/risk
22 information will inform the Administrator's decisions as to whether the existing primary and/or
23 secondary 63 standards should be revised and, if so, what revised standard or standards is/are
24 appropriate.
25 Staffs considerations and conclusions related to the current and alternative primary and
26 secondary 63 standards are framed by a series of key policy-relevant questions, expanding upon
27 those presented in the IRP at the outset of this review (U.S. EPA, 201 la). Answers to these
28 questions in the final PA will inform the Administrator's decisions as to whether, and if so how,
29 to revise the current 63 standards. The first overarching question is as follows.
30 • Do the currently available scientific evidence and exposure/risk information, as
31 reflected in the ISA and REAs, support or call into question the adequacy of the
32 protection afforded by the current Os standards?
33 If the answer to this question, which is informed by staffs consideration of more specific
34 questions related to the primary and secondary standards, suggests that revision of the current
1-12
-------
1 standards may be appropriate, then staff further considers the currently available evidence and
2 information with regard to the following question.
3 • What range of potential alternative standards is appropriate to consider based on
4 the scientific evidence, air quality analyses, and exposure/risk-based
5 information?
6 The general approaches for consideration of these overarching questions in review of the primary
7 and secondary standards are described separately in sections 1.3.1 and 1.3.2 below.
8 1.3.1 Approach for the Primary Standard
9 Staffs approach in this review of the current primary Os standard takes into
10 consideration the approaches used in previous 63 NAAQS reviews. The past and current
11 approaches described below are both based, fundamentally, on using EPA's assessment of the
12 current scientific evidence and associated quantitative analyses to inform the Administrator's
13 judgment regarding a primary standard for 63 that is "requisite" (i.e., neither more nor less
14 stringent than necessary) to protect public health with an adequate margin of safety.
15 In reaching conclusions on options for the Administrator's consideration, we note that the
16 final decision to retain or revise the current primary 63 standard is a public health policy
17 judgment to be made by the Administrator. This final decision by the Administrator will draw
18 upon the available scientific evidence for Os-attributable health effects, and on analyses of
19 population exposures and health risks, including judgments about the appropriate weight to
20 assign the range of uncertainties inherent in the evidence and analyses. Our general approach to
21 informing these judgments, discussed more fully below, recognizes that the available health
22 effects evidence reflects a continuum from relatively higher Oi concentrations, at which
23 scientists generally agree that health effects are likely to occur, through lower concentrations, at
24 which the likelihood and magnitude of a response become increasingly uncertain. Therefore, in
25 developing conclusions in this second draft PA, we are mindful that the Administrator's ultimate
26 judgments on the primary standard will most appropriately reflect an interpretation of the
27 available scientific evidence and exposure/risk information that neither overstates nor understates
28 the strengths and limitations of that evidence and information. This approach is consistent with
29 the requirements of sections 108 and 109 of the Act, as well as with how the EPA and the courts
30 have historically interpreted the Act.
31 Section 1.3.1.1 below provides an overview of the general approach taken in the last
32 review of the primary Os NAAQS (i.e., the 2008 review), and a summary of the rationale for the
33 decision on the level of the standard in that review (73 FR 16436). Section 1.3.1.2 presents our
34 approach in the current review, including our approach to considering the health evidence and
1-13
-------
1 exposure/risk information, and considerations regarding ambient Os concentrations attributable
2 to background sources.
3 1.3.1.1 Approach Used in the Last Review
4 In the 2008 review of the O^ NAAQS, the Administrator relied upon consideration of the
5 available scientific evidence and exposure/risk information, the advice and recommendations of
6 CASAC, and comments from the public. Based on this, he revised the level of the 8-hour
7 primary Oj, standard from 0.08 ppm14 to 0.075 ppm (75 ppb15). In reaching a decision to revise
8 the 1997 8-hour primary 63 standard, the Administrator noted that much new evidence had
9 become available since the 1997 review. He noted that this body of scientific evidence was very
10 robust and provided consistent and coherent evidence of an array of Ch-related respiratory
11 morbidity effects, and possibly cardiovascular-related morbidity, as well as total nonaccidental
12 and cardiorespiratory mortality. The Administrator specifically observed that (1) the evidence of
13 a range of respiratory-related morbidity effects had been considerably strengthened; (2) newly
14 available evidence from controlled human exposure and epidemiologic studies identified people
15 with asthma as an important susceptible population for which estimates of respiratory effects in
16 the general population likely underestimate the magnitude or importance of these effects; (3)
17 newly available evidence about mechanisms of toxicity more completely explained the
18 biological plausibility of Cb-induced respiratory effects and was beginning to suggest
19 mechanisms that may link 63 exposure to cardiovascular effects; and (4) there was relatively
20 strong evidence for associations between short-term 03 concentrations and total nonaccidental
21 and cardiopulmonary mortality. The Administrator believed that this very robust body of
22 evidence enhanced our understanding of Os- related effects and provided increased confidence
23 that various respiratory morbidity effects and other effects marked by indicators of respiratory
24 morbidity are causally related to 63 exposures, and that the evidence was highly suggestive that
25 Os exposures during the warm Os season contribute to premature mortality.16
26 The Administrator also noted important new health evidence reporting a broad array of
27 adverse effects following short-term exposures to 03 concentrations below the level of the 1997
28 standard, and concerns for such or related effects in at-risk populations,17 including people with
14 Due to rounding convention, the 1997 standard level of 0.08 ppm corresponded to 0.084 ppm (84 ppb).
15 The level of the O3 standard is specified as 0.075 ppm rather than 75 ppb. However, in this draft PA we refer to
ppb, which is most often used in the scientific literature and in the ISA, in order to avoid the confusion that could
result from switching units when discussing the evidence in relation to the standard level.
16 73 FR 16470-16471 (March 27, 2008)
17 Here, as in the ISA, the term "at-risk population" is used to encompass populations or lifestages that have a
greater likelihood of experiencing health effects related to exposure to an air pollutant due to a variety of factors;
other terms used in the literature include susceptible, vulnerable, and sensitive. These factors may be intrinsic, such
1-14
-------
1 asthma or other lung diseases, older adults with increased susceptibility, and those who are likely
2 to be vulnerable as a result of spending a lot of time outdoors engaged in physical activity (e.g.,
3 especially active children and outdoor workers).
4 He specifically noted new scientific evidence, which built upon existing evidence,
5 demonstrating Os-induced lung function effects and respiratory symptoms in some healthy
6 individuals following exposures down to 80 ppb. He also noted very limited new evidence
7 demonstrating such effects at exposure concentrations well below 80 ppb. In addition, the
8 Administrator noted (1) epidemiologic evidence of statistically significant associations with Os-
9 related health effects in areas that likely would have met the then-current standard; (2)
10 epidemiologic studies conducted in areas that likely would have violated the existing standard
11 but which nonetheless reported statistically significant associations that generally extended down
12 to ambient 63 concentrations below the level of that standard; (3) the few studies that had
13 reported statistically significant associations with respiratory morbidity outcomes and mortality
14 in subsets of data that included only days with ambient Os concentrations below the level of the
15 existing standard; and (4) controlled human exposure studies, together with animal toxicological
16 studies, that provided considerable support for the biological plausibility of the respiratory
17 morbidity associations observed in the epidemiologic studies. Based on the available evidence,
18 the Administrator agreed with the CAS AC and the maj ority of public commenters that the
19 existing standard was not requisite to protect public health with an adequate margin of safety (FR
20 73 16471).
21 Beyond this focus on the available health evidence, the Administrator also considered
22 estimates of Os exposures and health risks based on analyses where air quality was adjusted to
23 simulate just meeting the existing and potential alternative standards. For the various air quality
24 simulations, he specifically considered the pattern of estimated reductions in Os exposures across
25 health benchmark concentrations of 80, 70, and 60 ppb. The 80 ppb benchmark reflected an
26 exposure concentration for which there was strong evidence for respiratory effects in healthy
27 people, including airway inflammation, respiratory symptoms, airway hyperresponsiveness, and
28 impaired lung host defense (U.S. EPA, 2007, section 4.7). The 60 ppb benchmark reflected an
as genetic factors, lifestage, or the presence of preexisting diseases, or they may be extrinsic, such as socioeconomic
status (SES), activity pattern and exercise level, or increased pollutant exposures (U.S. EPA 2013, p. Ixx, 8-1, 8-2).
The courts and the Act's legislative history refer to these at-risk subpopulations as "susceptible" or "sensitive"
populations. See, e.g., American Lung Ass'n v. EPA. 134 F. 3d 388, 389 (D.C. Cir. 1998) ("NAAQS must protect
not only average health individuals, but also 'sensitive citizens' - children, for example, or people with asthma,
emphysema, or other conditions rendering them particularly vulnerable to air pollution" (quoting S. Rep. No. 91 -
1196 at 10).
1-15
-------
1 exposure concentration for which the Administrator judged the evidence of such effects to be
2 very limited (73 FR 16471).
3 The Administrator took note of the magnitudes of estimated health risks for a range of
4 health effects, including moderate and large lung function decrements, respiratory symptoms,
5 respiratory-related hospital admissions, and nonaccidental and cardiorespiratory mortality. He
6 recognized that these quantitative risk estimates for a limited number of specific health effects
7 were indicative of a much broader array of (Vrelated effects, including various indicators of
8 morbidity in at-risk populations that we could not analyze in the risk assessment (e.g., school
9 absences, increased medication use, emergency department visits). The Administrator concluded
10 that quantitative exposure and risk estimates, as well as the broader array of Os-related health
11 endpoints that could not be quantified, provided additional support for the evidence-based
12 conclusion that the existing standard needed to be revised (73 FR 16472).
13 Based on the above considerations, and consistent with CASAC's unanimous conclusion
14 that there was no scientific justification for retaining the existing standard, the Administrator
15 concluded that the primary 63 standard set in 1997 was not sufficient and thus not requisite to
16 protect public health with an adequate margin of safety. He further concluded that revision of
17 this standard was needed to provide increased public health protection (73 FR 16472).
18 Throughout the 2008 review, CAS AC supported a standard level in the range of 60 to 70
19 ppb (without change to the form, indicator, or averaging time). In a letter to the Administrator on
20 the second draft Staff Paper, CAS AC unanimously recommended "that the current primary
21 ozone standard be revised and that the level that should be considered for the revised standard be
22 from 0.060 to 0.070 ppm" (60 to 70 ppb) (Henderson, 2006, p. 5). This recommendation, based
23 in part on the placement of more weight on the evidence for effects following exposures to 60
24 ppb Os, followed from the CASAC's more general recommendation that the 1997 standard
25 needed to be made substantially more protective of human health, particularly for at-risk
26 populations. In a subsequent letter sent specifically to offer advice to aid the Administrator and
27 Agency staff in developing the 2007 O^ proposal, CAS AC reiterated that Panel members "were
28 unanimous in recommending that the level of the current primary ozone standard should be
29 lowered from 0.08 ppm to no greater than 0.070 ppm" (Henderson, 2007, p. 2).18
30 After considering CASACs comments, the Administrator judged that the appropriate
31 balance to draw, based on the entire body of evidence and information available in the 2008
18 The D.C. Circuit, in its review of the 2008 primary standard, stated that it was unclear whether CASAC's advice
reflected issues of pure science or issues of science and policy. That is, the court was unable to determine whether
CASAC's conclusion in its 2007 letter that the standard be set no higher than 70 ppb "was based on its scientific
judgment that adverse effects would occur at that level or instead based on its more qualitative judgment that the
range it proposed would be more appropriately protective of human health with an adequate margin of safety."
Mississippi. 723 F. 3d at 269.
1-16
-------
1 review, was a standard set at a level of 75 ppb (and leaving all other elements of the NAAQS
2 unchanged). In making this decision, the Administrator placed primary emphasis on the body of
3 available scientific evidence, while viewing the results of exposure and risk assessments as
4 providing supporting information. Specifically, the Administrator judged that a standard set at
5 75 ppb would be appreciably below 80 ppb, the level in controlled human exposure studies at
6 which adverse effects had been demonstrated at the time, and would provide a significant
7 increase in protection compared to the then-current standard. Based on results of the exposure
8 assessment, he also noted that exposures to Os concentrations at and above a benchmark level of
9 80 ppb would be essentially eliminated with a standard level of 75 ppb, and that exposures at and
10 above a 70 ppb benchmark level would be substantially reduced or eliminated for the vast
11 majority of people in at-risk groups. In addition, the Administrator concluded that the body of
12 evidence did not support setting a lower standard level, specifically judging that the available
13 evidence for effects following exposures to Os concentrations of 60 ppb was "too limited to
14 support a primary focus at this level" (75 FR 2938). With respect to the epidemiologic evidence,
15 the Administrator stated that a standard set at a level lower than 75 ppb "would only result in
16 significant further public health protection if, in fact, there is a continuum of health risks in areas
17 with 8-hour average 63 concentrations that are well below the concentrations observed in the key
18 controlled human exposure studies and if the reported associations observed in the
19 epidemiological studies are, in fact, causally related to O?, at those lower levels" (73 FR 16483).
20 In making his final decision about the level of the primary 63 standard, the Administrator
21 noted that the level of 75 ppb was above the range recommended by CASAC (i.e., 70 to 60 ppb).
22 He concluded that "CASAC's recommendation appeared to be a mixture of scientific and policy
23 considerations" (75 FR 2992). The Administrator reached a different policy judgment than the
24 CASAC Panel, placing less weight than CASAC on the available controlled human exposure
25 studies reporting effects following exposures to 60 ppb 63 and less weight on the results from
26 exposure and risk assessments, particularly on estimates of exposures to 63 concentrations at or
27 above 60 ppb (73 FR 16482-3).
28 1.3.1.2 Approach for the Current Review
29 To identify the range of options appropriate for the Administrator to consider in the
30 current review, we apply an approach that builds upon the general approach used in the last
31 review (and in the 2010 reconsideration proposal) and that reflects the broader body of scientific
32 evidence, updated exposure/risk information, and advances in 63 air quality modeling now
33 available. As summarized above, the Administrator's decisions in the prior review were based on
34 an integration of information on health effects associated with exposure to 63, judgments on the
1-17
-------
1 adversity and public health significance of key health effects, and expert and policy judgments as
2 to when the standard is requisite to protect public health with an adequate margin of safety.
3 Staffs preliminary conclusions on the primary Os standard reflect our consideration of
4 the available scientific evidence, exposure/risk information, and air quality modeling
5 information, within the context of the overarching questions related to: (1) the adequacy of the
6 current primary Os standard to protect against effects associated with both short- and long-term
7 exposures and (2) potential alternative standards, if any, that are appropriate to consider in this
8 review. In addressing these broad questions, we organize the discussions in chapters 3 and 4 of
9 this document around a series of more specific questions reflecting different aspects of each
10 overarching question. When evaluating the health protection afforded by the current or potential
11 alternative standards, we take into account the four basic elements of the NAAQS: the indicator,
12 averaging time, form, and level.
13 Figure 1-1 below provides an overview of our approach in this review. We believe that
14 the general approach summarized in this section, and outlined in Figure 1-1, provides a
15 comprehensive basis to help inform the judgments required of the Administrator in reaching
16 decisions about the current and potential alternative primary Os standards. In the subsections
17 below, we describe our general approaches to considering the scientific evidence (evidence-
18 based considerations) and to considering the human exposure- and health risk information
19 (exposure- and risk-based considerations). We also recognize considerations related to ambient
20 Oj, attributable to background sources.
1-18
-------
Adequacy of Current 8-Hour Primary 03 Standard
1
2
Evidence-Based Considerations
/•ISA weight-of-evidence conclusions for health effects and at-risk
populations
/'Controlled human exposure and animal toxicology studies:
Nature, magnitude, likely adversity of effects; consistency across
studies?
/•Epidemiologic studies: Statistical precision; confounding; link
effects to 03 air quality in specific locations?
/-Health effect associations reported in locations meeting
current standard?
.'•Confidence in concentration-response relationships over
distributions of ambient concentrations, including
concentrations below level of current standard?
^-Uncertainties in evidence for 03-attriubutable effects across
distributions of exposure concentrations and ambient
concentrations
g
>cts across
nt
i
Exposure-/Risk-Based
Considerations
/'Nature, magnitude, and
importance of estimated exposures
and risks associated with current
03 standard
r Focus on at-risk populations
/-Uncertainties in the exposure
and risk estimates
Does information call
into question
adequacy of current
8-hour Primary 03
standard?
Consider retaining
currentS-hour 03
standard
lYES
Consider Potential Alternative Standards
J
Indicator
^Support for retaining
03?
Averaging Time
/•Support for 8-hour
only?
/-Support for longer-term
averaging time?
Form
/•Support for retaining annual
4th highest?
Level
/'Evidence-based considerations: Consider controlled human exposure studies and
epidemiologic studies, including uncertainties
/•Exposure- and risk-based considerations: Consider exposure and risk reductions for
alternative 03 standards, exposures and risks estimated to remain upon meeting
alternatives, and uncertainties and limitations in exposure/risk estimates
I
Identify range of potential alternative standards for
consideration
Figure 1-1. Overview of approach to reviewing the primary standard.
1-19
-------
1 1.3.1.2.1 Consideration of the Scientific Evidence
2 Our approach in this review draws upon an integrative synthesis of the entire body of
3 available scientific evidence for Os-related health effects, including the evidence newly available
4 in the current review and the evidence from previous reviews, as presented in the ISA (U.S.
5 EPA, 2013). Our approach to considering the scientific evidence is based fundamentally on using
6 information from controlled human exposure and epidemiologic studies, supplemented by
7 information from animal toxicology studies. Such evidence informs our consideration of the
8 health endpoints and at-risk populations on which to focus the current review, and our
9 consideration of the Os concentrations at which various health effects can occur.
10 Since the 2008 review of the O^ NAAQS, the Agency has developed formal frameworks
11 for characterizing the strength of the scientific evidence with regard to health effects associated
12 with exposures to Os in ambient air and factors that may increase risk in some populations or
13 lifestages (U.S. EPA, 2013, Preamble; Chapter 8). These frameworks provide the basis for
14 robust, consistent, and transparent processes for evaluating the scientific evidence, including
15 uncertainties in the evidence, and for drawing weight-of-evidence conclusions on air pollution-
16 related health effects and at-risk populations.
17 With regard to characterization of health effects, the ISA uses a five-level hierarchy to
18 classify the overall weight-of-evidence into one of the following categories: causal relationship,
19 likely to be a causal relationship, suggestive of a causal relationship, inadequate to infer a causal
20 relationship, and not likely to be a causal relationship (U.S. EPA, 2013, Preamble Table II). In
21 this PA, we place the greatest weight on the evidence for health effects that have been judged in
22 the ISA to be caused by, or likely to be caused by, Os exposures. Our consideration of the
23 available evidence for such effects is presented below in Chapter 3 (consideration of the
24 adequacy of the current standard) and in Chapter 4 (consideration of potential alternative
25 standards).
26 As discussed below, we further consider the evidence base assessed in the ISA with
27 regard to the types and levels of exposure at which health effects are indicated. This further
28 consideration of the evidence, which directly informs EPA's conclusions regarding the adequacy
29 of current or potential alternative standards in providing requisite public health protection, differs
30 from consideration of the evidence in the ISA with regard to overarching determinations of
31 causality. Therefore, studies that inform determinations of causality may or may not be
32 concluded to be informative with regard to the adequacy of the current or potential alternative
33 standards.
34 As with health endpoints, the ISA's characterization of the weight-of-evidence for
35 potential at-risk populations is based on the evaluation and synthesis of evidence from across
36 scientific disciplines. The ISA characterizes the evidence for a number of "factors" that have the
1-20
-------
1 potential to place populations at increased risk for Os-related effects. The categories considered
2 in evaluating the evidence for these potential at-risk factors are "adequate evidence," "suggestive
3 evidence," "inadequate evidence," and "evidence of no effect." These categories are discussed
4 in more detail in the ISA (U.S. EPA, 2013, chapter 8, Table 8-1). In this draft PA, we focus our
5 consideration of potential at-risk populations on those factors for which the ISA judges there is
6 "adequate" evidence (U.S. EPA, 2013, Table 8-5). At-risk populations are discussed in more
7 detail in section 3.2.1, below.
8 Using the available scientific evidence to inform conclusions on the adequacy of the
9 current primary Oj, standard, and on potential alternative standards appropriate for consideration,
10 is complicated by the recognition that a population-level threshold in exposure or ambient 63
11 concentrations has not been identified, below which it can be concluded with confidence that O^-
12 attributable effects do not occur in exposed populations (U.S. EPA, 2013, section 2.5.4.4). In the
13 absence of a discernible threshold, our general approach to considering the available Os health
14 evidence involves characterizing our confidence in the extent to which Os-attributable effects
15 occur, and the extent to which such effects are adverse, over the ranges of 63 exposure
16 concentrations evaluated in controlled human exposure studies and over the distributions of
17 ambient 63 concentrations in locations where epidemiologic studies have been conducted. As
18 noted above, we recognize that the available health effects evidence reflects a continuum from
19 relatively high O^ concentrations, at which scientists generally agree that adverse health effects
20 are likely to occur, through lower concentrations, at which the likelihood and magnitude of a
21 response become increasingly uncertain. Aspects of our approach particular to evidence from
22 controlled human exposure and epidemiologic studies, respectively, are discussed below.
23 Controlled Human Exposure Studies
24 Controlled human exposure studies provide direct evidence of relationships between
25 pollutant exposures and human health effects (U.S. EPA, 2013, p.lx). Such studies are
26 particularly useful in defining the specific conditions under which pollutant exposures can result
27 in health impacts, including the exposure concentrations, durations, and ventilation rates under
28 which effects can occur. As discussed in the ISA, controlled human exposure studies provide
29 clear and compelling evidence for an array of human health effects that are directly attributable
30 to acute exposures to O^per se (i.e., as opposed to 03 and other photochemical oxidants, for
31 which 63 is an indicator, or other co-occurring pollutants) (U.S. EPA, 2013, Chapter 6).
32 Together with animal toxicological studies, which can provide information about more serious
33 health outcomes as well as the effects of long-term exposures and mode of action, controlled
34 human exposure studies also help to provide biological plausibility for health effects observed in
35 epidemiologic studies.
1-21
-------
1 In this draft PA, we consider the evidence from controlled human exposure studies in two
2 ways. First, we consider the extent to which controlled human exposure studies provide evidence
3 for health effects following exposures to different O^ concentrations, down to the lowest-
4 observed-effects levels in those studies. Second, we use such studies to inform our evaluation of
5 the extent to which we have confidence in health effect associations reported in epidemiologic
6 studies down through lower ambient O^ concentrations, where the likelihood and magnitude of
7 Os-attributable effects become increasingly uncertain.
8 We consider the range of Os exposure concentrations evaluated in controlled human
9 exposure studies, including concentrations near or below the level of the current standard. We
10 consider both group mean responses, which provide insight into the extent to which observed
11 changes are due to Os exposures rather than to chance alone, and inter-individual variability in
12 responses, which provides insight into the fraction of the population that might be affected by
13 such Os exposures (U.S. EPA, 2013, section 6.2.1.1). When considering the relative weight to
14 place on various controlled human exposure studies, we consider the exposure conditions
15 evaluated (e.g., exercising versus resting, exposure duration); the nature, magnitude, and likely
16 adversity of effects over the range of reported Os exposure concentrations; the statistical
17 precision of reported effects; and the consistency of results across studies for a given health
18 endpoint and exposure concentration. In addition, because controlled human exposure studies
19 typically involve healthy individuals and do not evaluate the most sensitive individuals in the
20 population (U.S. EPA, 2013, Preamble p. Ix), when considering the implications of these studies
21 for our evaluation of the current and potential alternative standards, we also consider the extent
22 to which reported effects are likely to reflect the magnitude and/or severity of effects in at-risk
23 groups.
24 Epidemiologic Studies
25 We also consider epidemiologic studies of short- and long-term 63 concentrations in
26 ambient air. Epidemiologic studies provide information on associations between variability in
27 ambient O?, concentrations and variability in various health outcomes, including lung function
28 decrements, respiratory symptoms, school absences, hospital admissions, emergency department
29 visits, and premature mortality (U.S. EPA, 2013, Chapters 6 and 7). Epidemiologic studies can
30 inform our understanding of the effects in the study population (which may include at-risk
31 groups) of real-world exposures to the range of 63 concentrations in ambient air.
32 Available studies have generally not indicated a discernible population threshold, below
33 which Os is no longer associated with health effects (U.S. EPA, section 2.5.4.4). However, the
34 currently available epidemiologic evidence indicates decreased confidence in reported
35 concentration-response relationships for O^ concentrations at the lower ends of ambient
36 distributions (U.S. EPA, section 2.5.4.4). Therefore, our general approach to considering the
1-22
-------
1 results of epidemiologic studies within the context of the current and potential alternative
2 standards focuses on characterizing the range of ambient 63 concentrations over which we have
3 the most confidence in (^-associated health effects, and the concentrations below which our
4 confidence in such health effect associations becomes appreciably lower. In doing so, we
5 consider the statistical precision of Os health effect associations reported in study locations with
6 various ambient Os concentrations; confidence intervals around concentration-response functions
7 reported over distributions of ambient 63 (where available); and the extent to which the
8 biological plausibility of associations at various ambient Os concentrations is supported by
9 evidence from controlled human exposure and/or animal toxicological studies.
10 We consider both multi-city and single-city studies assessed in the ISA, each of which
11 have strengths and limitations. Multi-city studies evaluate large populations and provide greater
12 statistical power than single-city studies. Multi-city studies also reflect Os-associated health
13 impacts across a range of diverse locations, providing spatial coverage for different regions
14 across the country and reflecting differences in exposure-related factors that could impact O^
15 risks. In addition, compared to single-city studies, multi-city studies are not prone to publication
16 bias and they afford the possibility of generalizing to the broader national population (U.S. EPA,
17 2004, p. 8-30). In contrast, while single-city studies are more limited than multicity studies in
18 terms of statistical power and geographic coverage, conclusions regarding the extent to which air
19 quality met the current or potential alternative standards in the cities for which associations have
20 been reported can be made with greater certainty for single-city studies (compared to multicity
21 studies reporting only multicity effect estimates) because the associations are reported for city-
22 specific analyses (U.S. EPA, 201 Id, section 2.3.4.1).19 In some cases, single-city studies can
23 also provide evidence for locations or population-specific characteristics not reflected in
24 multicity studies (U.S. EPA, 2013, section 6.2.7.1). Therefore, when considering available
25 epidemiologic studies we evaluate both multi-city and single-city studies, recognizing the
26 strengths and limitations of each.
27 In placing emphasis on specific epidemiologic studies, we focus on studies conducted in
28 the U.S. and Canada. Such studies reflect air quality and exposure patterns that are likely more
29 typical of the U.S. population than the air quality and exposure patterns reflected in studies
90
30 conducted outside the U.S. and Canada. We also focus on studies reporting associations with
31 effects judged in the ISA to be robust to confounding by other factors, including co-occurring air
32 pollutants.
19 Though in some cases multicity studies present single-city effect estimates in addition to multi-city estimates.
20 All studies, including other international studies inform the causal determinations in the ISA.
1-23
-------
1 1.3.1.2.2 Consideration of Exposure and Risk Estimates
2 To put judgments about (Vrelated health effects into a broader public health context, we
3 consider exposure and risk estimates from the second draft HREA, which develops and applies
4 models to estimate human exposures to 63 and (Vrelated health risks in urban case study areas
5 across the United States (U.S. EPA, 2014). The second draft HREA estimates exposures of
6 concern, based on interpreting quantitative exposure estimates within the context of controlled
7 human exposure study results; lung function risks, based on applying exposure-response
8 relationships from controlled human exposure studies to quantitative estimates of exposures; and
9 epidemiologic-based risk estimates, based on applying concentration-response relationships
10 drawn from epidemiologic studies to adjusted air quality. Each of these types of assessments is
11 discussed briefly below.
12 As in the 2008 review, the second draft HREA estimates exposures at or above
13 benchmark concentrations of 60, 70, and 80 ppb, reflecting exposure concentrations of concern
91
14 based on the available health evidence. Estimates of exposures at or above discrete benchmark
15 concentrations provide perspective on the public health risks of Cb-related health effects that
16 have been demonstrated in controlled human exposure and toxicological studies but that, because
17 of a lack of exposure-response information from those studies, cannot be assessed using a
18 quantitative risk assessment. Though this analysis is conducted using discrete benchmark
19 concentrations, health-relevant exposures are more appropriately viewed as a continuum with
20 greater confidence and less uncertainty about the existence of health effects at higher 63
21 exposure concentrations and less confidence and greater uncertainty at lower exposure
22 concentrations. We recognize that there is no sharp breakpoint within the exposure-response
23 relationship for exposure concentrations at and above 80 ppb down to 60 ppb.
24 The second draft HREA also generates quantitative estimates of Os health risks for air
25 quality adjusted from recent conditions to those just meeting the current and potential alternative
26 standards. As noted above, one approach to estimating 63 health risks is to combine modeled
27 exposure estimates with exposure-response relationships derived from controlled human
28 exposure studies of Cb-induced health effects. The second draft HREA uses this approach to
29 estimate the occurrence of (Vinduced lung function decrements in simulated at-risk populations.
30 The available exposure-response information does not support this approach for other endpoints
31 evaluated in controlled human exposure studies (U.S. EPA, 2014a, section 2.3).
32 Another approach to estimating Os-associated health risks is to apply concentration-
33 response relationships derived from short- and/or long-term epidemiologic studies to air quality
21 For example, see 75 FR 2945-2946 (January 19, 2010) and 73 FR 16441-16442 (March 27, 2008) discussing
"exposures of concern".
1-24
-------
1 adjusted to just meet current and potential alternative standards. The concentration-response
2 relationships drawn from epidemiologic studies are based on population exposure surrogates,
3 such as 8-hour concentrations averaged across monitors and over more than one day
4 (incorporation of lag) (U.S. EPA, 2013, Chapter 6). The second draft HREA presents
5 epidemiologic-based risk estimates for Os-associated mortality, hospital admissions, emergency
6 department visits, and respiratory symptoms (U.S. EPA, 2014a, section 2.3). These estimates are
7 derived from the full distribution of ambient 63 concentrations estimated for the study
99
8 locations. In addition, the second draft HREA estimates mortality risks attributable to various
9 portions of those distributions (U.S. EPA, 2014a). In this second draft PA we consider risk
10 estimates based on the full distributions of ambient 63 concentrations, and estimates of the risk
11 associated with various portions of those ambient distributions. In doing so, we take note of the
12 ISA conclusions regarding confidence in linear concentration-response relationships over
13 distributions of ambient concentrations, and of the extent to which health effect associations at
14 various ambient O^ concentrations are supported by the evidence from experimental studies for
15 effects following specific 63 exposures.
16 1.3.1.2.3 Considerations Regarding Ambient Os Concentration Estimates
17 Attributable to Background Sources
18 As noted above, our approach in this review utilizes recent advances in modeling
19 techniques to estimate the contributions of U.S. anthropogenic, international anthropogenic, and
20 natural sources to ambient 03 (discussed in detail in Chapter 2 of this draft PA). Such model
21 estimates can provide insights into the extent to which different types of background emissions
22 sources contribute to total ambient Os concentrations. Consideration of this issue in the current
23 review is informed by the approaches taken in previous reviews, as well as by court decisions.
24 In 1979, the EPA set a 1-hour Oi standard with a level of 0.12 ppm. Following the final
25 decision in that review, the City of Houston argued that the standard was arbitrary and capricious
26 because natural 63 concentrations and other physical phenomena in the Houston area made the
27 standard unattainable in that area. The D.C. Circuit rejected this argument, stating that
28 attainability and technological feasibility are not relevant considerations in the promulgation of
29 the NAAQS. The Court also noted that the EPA need not tailor the NAAQS to fit each region or
30 locale, pointing out that Congress was aware of the difficulty in meeting standards in some
In previous reviews, including the 2008 review and reconsideration, such risks were separately estimated for O3
concentrations characterized as above policy-relevant background concentrations. Policy-relevant background
concentrations were defined as the distribution of ozone concentrations attributable to sources other than
anthropogenic emissions of ozone precursor emissions (e.g., VOC, CO, NOx) in the U.S., Canada, and Mexico. The
decision to estimate total risk across the full range of O3 concentrations reflects current OAQPS views and
consideration of advice from CASAC (Frey and Samet, 2012b).
1-25
-------
1 locations and had addressed this difficulty through various compliance related provisions in the
2 Act. See API v. Costle. 665 F.2d 1176. 1184-6 (D.C. Cir. 1981).
3 More recently, in the 1997 review of the O^ NAAQS, the Administrator set an 8-hour
4 standard with a level of 0.08 ppm (84 ppb). In reaching this decision, the EPA identified several
5 reasons supporting its decision to reject a more stringent standard of 0.07 ppm. Most
6 importantly, the EPA pointed out the scientific uncertainty at lower concentrations and placed
7 significant weight on the fact that no CASAC panel member supported a standard level set lower
8 than 0.08 ppm (62 FR 38868). In addition to noting the uncertainties in the health evidence for
9 exposure concentrations below 0.08 ppm and the advice of CASAC, the EPA noted that a
10 standard set at a level of 0.07 ppm would be closer to peak background concentrations that
11 infrequently occur in some areas due to nonanthropogenic sources of Os precursors (62 FR
12 38856, 38868; July 18, 1997).
13 In subsequent litigation, the D.C. Circuit upheld the EPA's decision as the product of
14 reasoned decision-making. The Court made clear that the most important support for the EPA's
15 decision was the health evidence and the concerns it raised about setting a standard level below
16 0.08 ppm. The Court also pointed to the significant weight that the EPA properly placed on the
17 advice it received from CASAC. Finally (as noted in section 1.2.2 above), the Court noted that
18 the EPA could also consider relative proximity to peak natural background 63 when evaluating
19 alternative standards. See ATA III. 283 F.3d at 379 (D.C. Cir. 2002).
20 These cases provide a framework for considering the contributions of U.S.
21 anthropogenic, international anthropogenic, and natural sources, within the context of
22 considering the health evidence and CASAC advice, when evaluating various potential
23 alternative standards.
24 1.3.2 Approach for the Secondary Standard
25 Staffs approach in this review of the current secondary standard takes into consideration
26 aspects of the approaches used in past 63 NAAQS reviews. The past and current approaches,
27 generally described below, are both based fundamentally on using EPA's assessment of the
28 current scientific evidence and associated quantitative analyses to inform the Administrator's
29 judgment regarding a secondary standard for 63 that is requisite (i.e., neither more nor less
30 stringent than necessary) to protect public welfare.
31 In reaching conclusions on options for the Administrator's consideration, we note that the
32 final decision to retain or revise the current secondary 63 standard is a public welfare policy
33 judgment to be made by the Administrator. This final decision will draw upon the available
34 scientific evidence for Os-attributable welfare effects and on analyses of vegetation and
35 ecosystem exposures and public welfare risks based on impacts to vegetation, ecosystems and
1-26
-------
1 their associated services, including judgments about the appropriate weight to place on the range
2 of uncertainties inherent in the evidence and analyses. In determining the requisite level of
3 protection for crops and trees, the Administrator will need to weigh the importance of the
4 predicted risks of these effects in the overall context of public welfare protection, along with a
5 determination as to the appropriate weight to place on the associated uncertainties and limitations
6 of this information. Our general approach to informing these judgments, discussed more fully
7 below, recognizes that the available welfare effects evidence reflects a continuum from relatively
8 high OT, concentrations at which scientists generally agree that welfare effects are likely to occur,
9 through lower concentrations at which the likelihood and magnitude of a response become
10 increasingly uncertain. Therefore, in developing conclusions in this second draft PA, we are
11 mindful that the Administrator's ultimate judgments on the secondary standard will most
12 appropriately reflect an interpretation of the available scientific evidence and exposure/risk
13 information that neither overstates nor understates the strengths and limitations of that evidence
14 and information.
15 Section 1.3.2.1 below provides an overview of the general approach taken in the last
16 review of the secondary standard for Os (i.e., the 2008 review), and a summary of the rationale
17 for the decision on the standard in that review (73 FR 16436). Section 1.3.2.2 presents our
18 approach in the current review, including our approach to considering the vegetation effects
19 evidence and exposure/risk information, and considerations regarding ambient O^ concentrations
20 attributable to background sources.
21 1.3.2.1 Approach Used in the Last Review
22 In the 2008 review of the secondary NAAQS for Os, the Administrator relied upon
23 consideration of the available scientific evidence and exposure/risk information, information
24 regarding biologically-relevant exposure indices, air quality information regarding the degree of
25 overlap between different exposure index forms, the advice and recommendations of CAS AC,
26 considerations regarding adversity, and comments from the public. Based on all of this, he
27 revised the level of the secondary 63 standard from 0.08 ppm23 to 0.075 ppm (75 ppb24).
28 In reaching a decision to revise the 1997 8-hour secondary standard, the Administrator
29 found, after carefully considering the public comments, that the fundamental scientific
30 conclusions on the effects of Os on vegetation and sensitive ecosystems reached in the 2006
31 Criteria Document and 2007 Staff Paper, as discussed in section IV. A of the final rule remained
23 Due to rounding convention, the 1997 standard level of 0.08 ppm corresponded to 0.084 ppm (84 ppb).
24 The level of the O3 standard is specified as 0.075 ppm rather than 75 ppb. However, in this draft PA we refer to
ppb, which is most often used in the scientific literature and in the ISA, n order to avoid the confusion that could
result from switching units when discussing the evidence in relation to the standard level.
1-27
-------
1 valid (73 FR 16496). He further recognized that several additional lines of evidence had
2 progressed sufficiently since the 1997 review to provide a more complete and coherent picture of
3 the scope of Os-related vegetation risks (i.e., visible foliar injury, tree biomass loss, crop yield
4 loss, and others), especially those faced by sensitive seedling, sapling and mature growth stage
5 tree species growing in field settings, and their associated forested ecosystems. This new
6 research reflected an increased emphasis on field-based exposure methods (e.g., free-air, ambient
7 gradient and biomonitoring surveys) (73 FR 16490) in addition to the more traditional controlled
8 open-top chamber (OTC) studies (73 FR 16485), and began to address one of the key data gaps
9 cited by the Administrator in the 1997 review (73 FR 16486). Specifically, by providing
10 additional evidence that (Vinduced crop yield loss and tree seedling biomass loss effects
11 observed in chambers also occurs in the field, this new research qualitatively increased support
12 for, and confidence in, the continued use of OTC-derived crop and tree seedling concentration-
13 response (C-R) functions developed in the National Crop Loss Assessment Network (NCLAN)
14 and National Health and Environmental Effects Research Laboratory - Western Ecology
15 Division (NHEERL-WED) studies, respectively, to predict (Vinduced impacts on crops and tree
16 seedlings in the field (72 FR 37886). All of these areas were considered together, along with
17 associated uncertainties, in an integrated weight-of-evidence approach (73 FR 16490).
18 Beyond the available vegetation effects evidence, the Administrator also considered
19 estimates of Os exposures and risks when air quality was adjusted to simulate just meeting the
20 existing and potential alternative standards. On the basis of these assessments, the Administrator
21 concluded that Os exposures that would be expected to remain after meeting the existing
22 standard would be sufficient to cause visible foliar injury and seedling and mature tree biomass
23 loss in (Vsensitive vegetation (73 FR 16496) and would still allow (Vrelated yield loss to occur
24 in some commodity crop species and fruit and vegetable species grown in the U.S. (73 FR
25 16489). Other Os-induced effects described in the literature, including an impaired ability of
26 many sensitive species and genotypes within species to adapt to or withstand other
27 environmental stresses, such as freezing temperatures, pest infestations and/or disease, and to
28 compete for available resources, would also be anticipated to occur. In the long run, the result of
29 these impairments (e.g., loss in vigor) could lead to premature plant death in Os sensitive species.
30 Though effects on other ecosystem components had only been examined in isolated cases, the
31 Administrator noted effects such as those described above could have significant implications for
32 plant community and associated species biodiversity and the structure and function of whole
33 ecosystems (73 FR 16496).
34 Although the Administrator concluded that the then-current standard was not sufficient to
35 protect against the known and anticipated effects described above, he also recognized that the
36 secondary standard is not meant to protect against all known observed or anticipated Os-related
1-28
-------
1 effects, but only those that can reasonably be judged to be adverse to the public welfare. The
2 Administrator recognized that the degree to which such effects should be considered to be
3 adverse depended on the intended use of the vegetation and its significance to the public welfare
4 (73 FR 16496). In this regard, he took note of a number of actions taken by Congress to establish
5 public lands that are set aside for specific uses that are intended to provide benefits to the public
6 welfare, including lands that are to be protected so as to conserve the scenic value and the natural
7 vegetation and wildlife within such areas, and to leave them unimpaired for the enjoyment of
8 future generations. Based on these considerations, and taking into consideration the advice and
9 recommendations of CASAC, the Administrator concluded that the protection afforded by the
10 existing standard was not sufficient, and that the standard needed to be revised to provide
11 additional protection from known and anticipated adverse effects on sensitive natural vegetation
12 and ecosystems (73 FR 16497).
13 Given his judgment on the need to revise, the Administrator then considered what
14 revisions to the standard were requisite to protect public welfare. Regarding the form of the
15 standard, the Administrator took note that at the conclusion of the 1997 review, the biological
9S
16 basis for a cumulative, seasonal form was not in dispute and that the 2006 Criteria Document
17 also concluded that 63 exposure indices that cumulate differentially-weighted hourly
18 concentrations are the best candidates for relating exposure to plant growth responses (EPA,
19 2006) (61 FR 65716; 73 FR 16486). The CASAC, in its letter to the Administrator following its
20 review of the second draft Staff Paper, stated that "there is a clear need for a secondary standard
21 which is distinctly different from the primary standard in averaging time, level and form" and
22 that "the CASAC unanimously agrees that it is not appropriate to try to protect vegetation from
23 the substantial, known or anticipated, direct and/or indirect, adverse effects of ambient ozone by
24 continuing to promulgate identical primary and secondary standards for ozone" (Henderson,
25 October 24, 2006, pp. 5-7). Although many possible cumulative, seasonal concentration-
26 weighted exposure metrics exist, the Staff Paper and the CASAC Panel concluded that the
r\r
27 W126 form is the most biologically-relevant cumulative, seasonal form appropriate to consider
28 in the context of the secondary standard review (73 FR 16486-87).2?
25 In the 1997 review, a different cumulative metric (SUM06) was proposed. Metric selection in both 1997 and 2008
was based on both science and policy considerations.
26 W126 is a cumulative exposure index that is biologically based. The W126 index focuses on the higher hourly
average concentrations, while retaining the mid-and lower-level values. It is defined as the sum of sigmoidally
weighted hourly O3 concentrations over a specified period, where the daily sigmoidal weighting function is defined
as: l-exp[-(W126/ti)p]
27 In a subsequent letter offering unsolicited advice to the Administrator and Agency staff on development of the
proposed rulemaking, the CASAC reiterated that Panel members ' 'were unanimous in supporting the
recommendation in the Final Ozone Staff Paper that protection of managed agricultural crops and natural terrestrial
ecosystems requires a secondary Ozone NAAQS that is substantially different from the primary ozone standard in
1-29
-------
1 Although agreeing with the Criteria Document, Staff Paper and CAS AC conclusions that
2 a cumulative exposure index that differentially weights 63 concentrations could represent a
3 reasonable policy choice for a seasonal secondary standard to protect against the effects of O^ on
4 vegetation and that the most appropriate cumulative, concentration-weighted form to consider
5 was the sigmoidally weighted W126 form (73 FR 16498), the Administrator also took note of the
6 1997 decision to make the revised secondary standard identical to a revised primary standard
7 after similar considerations (73 FR 16498). In considering the rationale for the 1997 decision, the
8 Administrator observed that it was based in part on an analysis that compared the degree of
9 overlap in county-level air quality measured in terms of alternative standard forms (62 FR
10 38876). Recognizing that significant uncertainty remained in 1997 regarding conclusions drawn
11 from such analyses, the Administrator also considered the results of a similar analysis of recent
12 monitoring data undertaken in the 2007 Staff Paper to assess the degree of overlap expected
13 between the existing standard (4* high, daily maximum 8-hour concentration averaged over
14 three years) and potential alternative standards based on W126 cumulative seasonal forms.
15 The Administrator noted that this analysis showed significant overlap between the 8-hour
16 secondary standard and selected levels of W126 standard forms, with the degree of overlap
17 between these potential alternative standards depending greatly on the W126 level selected and
18 the distribution of hourly 63 concentrations within the annual and/or 3-year average period.
19 From this analysis, the Administrator recognized that a secondary standard set identical to a
20 revised primary standard would provide a significant degree of additional protection for
21 vegetation as compared to that provided by the existing secondary standard. In further
22 considering the significant uncertainties in the available body of evidence and in the exposure
23 and risk analyses, and the difficulty in determining at what point various types of vegetation
24 effects become adverse for sensitive vegetation and ecosystems, the Administrator focused his
25 consideration on a level for an alternative W126 standard (with an annual form) at the upper end
26 of the proposed range (i.e., 21 ppm-hours). The Staff Paper analysis showed that at a W126 level
27 of 21 ppm-hours, there would be essentially no counties with air quality expected both to exceed
28 such an alternative W126 standard and to meet the revised 8-hour primary standard—that is,
29 based on this analysis of counties with ambient Os monitors, a W126-based level of 21 ppm-
30 hours would be unlikely to provide additional protection in any areas beyond that likely to be
31 provided by the revised 2008 primary standard (73 FR 16499/500).
averaging time, level and form"...and "[t]he recommended metric for the secondary ozone standard is the
(sigmoidally-weighted) W126 index, accumulated over at least the 12 'daylight' hours and over at least the three
maximum ozone months of the summer 'growing season" (Henderson, March 26, 2007, p.3).
1-30
-------
1 The Administrator also considered the Staff Paper finding that the degree of overlap
2 between counties (with areas of concern for vegetation) expected to meet an 8-hour level for the
3 form of the existing standard and potential alternative levels of a W126-based standard was
4 inconsistent across years analyzed. This variation depended greatly on levels selected for a
5 W126-based standard and a 3-year average 4 high daily maximum 8-hour standard,
6 respectively, and the distribution of hourly Os concentrations within the annual and/or 3-year
7 average period. From this, the Staff Paper recognized the need for caution in evaluating the
8 likely vegetation impacts associated with a given level of air quality expressed in terms of the
9 existing 8-hour average standard in the absence of parallel W126 information. In considering
10 these findings, the Administrator "recognize[d] that the general lack of rural monitoring data
11 made uncertain the degree to which the revised 8-hour standard or an alternative W126 standard
12 would be protective, and that there was the potential for not providing the appropriate degree of
13 protection for vegetation in areas with air quality distributions that resulted in a high cumulative,
14 seasonal exposure but did not result in high 8-hour average exposures" (73 FR 16500). With
15 regard to the 8-hour standard, he also noted that "[w]hile this potential for under-protection was
16 clear, the number and size of areas [then] at issue and the degree of risk [was] hard to determine.
17 However, such a standard would also tend to avoid the potential for providing more protection
18 than is necessary, a risk that would have arisen from moving to a new form for the secondary
19 standard despite the significant uncertainty in determining the degree of risk for any exposure
20 level and the appropriate level of protection, as well as uncertainty in predicting exposure and
21 risk patterns" (73 FR 16500).
22 Thus, although the Administrator agreed with the views and recommendations of
23 CASAC that a cumulative, seasonal standard was the most biologically relevant way to relate
24 exposure to plant growth response, he also recognized that there remained significant
25 uncertainties in determining or quantifying the degree of risk attributable to varying levels of 63
26 exposure, the degree of protection that any specific cumulative, seasonal standard would
27 produce, and the associated potential for error in determining the secondary standard that would
28 provide a requisite degree of protection—i.e., sufficient but not more than what is necessary.
29 Given these significant uncertainties, the Administrator concluded that establishing a new
30 secondary standard with a cumulative, seasonal form, at that time, would have resulted in
31 uncertain benefits beyond those afforded by the revised primary standard, and therefore, might
32 have been more than necessary to provide the requisite degree of protection (73 FR 16500).
33 Based on his consideration of these issues (73 FR 16497), the Administrator judged that the
34 appropriate balance to be drawn was to set a secondary standard identical in every way to the
35 revised 8-hour primary standard of 0.075 ppm. The Administrator believed that such a standard
36 would be sufficient to protect public welfare from known or anticipated adverse effects, and did
1-31
-------
1 not believe that an alternative cumulative, seasonal standard was needed to provide this degree of
2 protection (73 FR 16500).
3 As noted above, on July 23, 2013 the D.C. Circuit found this approach to be contrary to
4 law because EPA had failed to identify a level of air quality requisite to protect public welfare
5 and, therefore, EPA's comparison between the primary and secondary standards for determining
6 if requisite protection for public welfare was afforded by the primary standard was inherently
7 arbitrary. The court remanded the secondary standard to EPA for further consideration. 723 F. 3d
8 at 270-74.
9 1.3.2.2 Approach for the Current Review
10 To identify the range of options appropriate for the Administrator to consider in the
11 current review, we apply an approach that builds upon the general approach used in the 2008
12 review(and in the 2010 reconsideration proposal), and that reflects the broader body of scientific
13 evidence, updated exposure/risk information, and advances in Oj, air quality modeling now
14 available. As summarized above, the Administrator's decisions in the prior review were based on
15 an integration of information on welfare effects associated with exposure to Os, judgments on the
16 adversity and public welfare significance of key effects, and, expert and policy judgments as to
17 when the standard is requisite to protect public welfare. These considerations were informed by
18 air quality and related analyses, quantitative exposure and risk assessments, and qualitative
19 assessment of impacts that could not be quantified. In performing the evaluation in this
20 document, we are additionally mindful of the recent remand of the secondary standard by the
21 D.C. Circuit.
22 Our approach in this review of the secondary Os standard reflects our consideration of the
23 available scientific evidence, information on biologically-relevant exposure indices,
24 exposure/risk information, and air quality modeling information, within the context of
25 overarching questions related to: (1) the adequacy of the current secondary Os standard to protect
26 against effects associated with cumulative, seasonal exposures and (2) potential alternative
27 standards, if any, that are appropriate to consider in this review. In addressing these broad
28 questions, we have organized the discussions in chapters 5 and 6 of this document around a
29 series of more specific questions reflecting different aspects of each overarching question. When
30 evaluating the welfare protection afforded by the current or potential alternative standards, we
31 take into account the four basic elements of the NAAQS: the indicator, averaging time, form,
32 and level.
33 Figure 1-2 below provides an overview of our approach in this review. We believe that
34 the general approach summarized in this section, and outlined in Figure 1-2, provides a
35 comprehensive basis to help inform the judgments required of the Administrator in reaching
1-32
-------
1 decisions about the current and potential alternative secondary O^ standards. In the subsections
2 below, we summarize our general approaches to considering the scientific evidence (evidence-
3 based considerations) and to considering the exposure and risk information (exposure- and risk-
4 based considerations). We also recognize considerations related to ambient 63 attributable to
5 background sources.
1-33
-------
Adequacy of Current 8-Hour Secondary 0, Standard
Evidence-Based Considerations
HSA weight-of-evidence conclusions
/-Evidence for vegetation effects from cumulative
exposures allowed by the current standard (OTC)?
/- Evidence for vegetation effects in locations that
would likely have met the current standard (field-
based)?
Exposure-/Risk-Based Considerations
/-Nature, magnitude, and importance of
estimated exposures and risks associated with
current 03 standard
/'Risks attributable to U.S. anthropogenic,
international, and natural sources
/-Uncertainties in the exposure and risk
estimates?
Does
information
call into question the
adequacy and
appropriateness of
current 8-hour
secondary 03
^standard?
IYES
NO
Consider retaining
current 8-hour 03
standard
Consider Potential Alternative Standards
\
Indicator
/-Support for retaining
03?
Form
/-Support for cumulative, seasonal form?
/- Support for concentration-weighted
form?
/'Support for single vs. 3-year period?
Averaging Time
/'Support for 12 hour diurnal
period?
/-Support for 3-month seasonal
period?
Level
Evidence-based considerations
•03 exposure concentrations from OTC and field-based studies
•Uncertainties in the evidence
-Exposure and risk considerations
•What is the nature, magnitude, and importance of estimated exposures and risks associated
with potential alternative 03 standards?
•Uncertainties in the exposure/ risk assessment?
I
1
2
Identify range of potential alternative standards for consideration
Figure 1-2. Overview of approach to reviewing the secondary standard.
1-34
-------
1 1.3.2.2.1 Consideration of the Scientific Evidence
2 Our approach in this review draws upon an integrative synthesis of the entire body of
3 available scientific evidence for Os-related welfare effects, including the evidence newly
4 available in the current review and the evidence from previous review, as presented in the ISA
5 (U.S. EPA, 2013). Our approach to considering the scientific evidence for effects on vegetation
6 is based fundamentally on using information from controlled chamber studies and field-based
7 studies. Such evidence informs our consideration of welfare endpoints and at-risk species and
8 ecosystems on which to focus the current review, and our consideration of the ambient Os
9 conditions under which various welfare effects can occur.
10 As in each NAAQS review, we consider the entire body of evidence for the subject
11 criteria pollutant. With regard to identification of the welfare effects that could be caused by a
12 pollutant, we look to controlled exposure studies using chamber or free air methodologies and
13 field-based observational, survey and gradient studies. Evaluating all of the evidence together,
14 the ISA makes a determination with regard to the strength of the evidence for a causal
15 relationship between the air pollutant and specific welfare effects. These determinations inform
16 our identification of welfare effects for which the NAAQS may provide protection.
17 Since the 2008 review of the O^ NAAQS, the Agency has developed a formal framework
18 for characterizing the strength of the scientific evidence with regard to a causal relationship
19 between ambient Os and welfare effects (U.S. EPA, 2013, Preamble; Chapter 9). This framework
20 provides the basis for a robust, consistent, and transparent process for evaluating the scientific
21 evidence, including uncertainties in the evidence, and for drawing weight-of-evidence
22 conclusions regarding air pollution-related welfare effects. In so doing, the ISA uses a five-level
23 hierarchy, classifying the overall weight of evidence into one of the following categories: causal
24 relationship, likely to be a causal relationship, suggestive of a causal relationship, inadequate to
25 infer a causal relationship, and not likely to be a causal relationship (U.S. EPA, 2013, Preamble
26 Table II). In our approach here, we place the greatest weight on the evidence for welfare effects
27 that have been judged in the ISA to be caused by, or likely caused by, Os exposures. Our
28 consideration of the available evidence for such effects is presented below in Chapter 5
29 (consideration of the adequacy of the current standard) and in Chapter 6 (consideration of
30 potential alternative standards).
31 We further consider the evidence base, as assessed in the ISA, with regard to the types
32 and levels of exposure at which welfare effects are indicated. This further consideration of the
33 evidence base, which directly informs EPA's conclusions regarding the adequacy of current or
34 potential alternative standards in providing requisite public welfare protection, differs from
35 consideration of the evidenc in the ISA with regard to overarching determinations of causality.
1-35
-------
1 Studies that have informed determinations of causality may or may not be concluded to be
2 informative with regard to the adequacy of the current or potential alternative standards.
3 Our approach in this review, as in past reviews, included recognition that the available
4 evidence has not provided identification of a threshold in exposure or ambient 63 concentrations
5 below which it can be concluded with confidence that Cb-attributable vegetation effects do not
6 occur across the broad range of Os-sensitive plant species growing within the U.S. This is due in
7 part to the fact that research shows that there is variability in sensitivity between and within
8 species and that numerous factors, i.e. chemical, physical, biological, and genetic, can influence
9 the direction and magnitude of the studied effect (U.S. EPA, 2013, section 9.4.8). In the absence
10 of a discernible threshold, our general approach to considering the available 63 welfare evidence
11 involves characterizing our confidence in conclusions regarding Os-attributable vegetation
12 effects over the ranges of cumulative seasonal 63 exposure values evaluated in chamber studies
13 and in field studies in areas where Cb-sensitive vegetation are known to occur, as well as
14 characterizing the extent to which these effects can be considered adverse. In addition, because
15 63 can indirectly affect other ecosystem components (such as soils, water, and wildlife, and their
16 associated goods and services, through its effects on vegetation) our approach also considers
17 those indirect effects for which the ISA concludes, based on multiple lines of evidence, including
18 mechanistic and physiological processes, to have a causal or likely to be a causal relationship.
19 With respect to ecosystem services for which we may have only limited or qualitative
20 information regarding an association with 63 exposures, our approach is to consider their policy-
21 relevance in the context of section 109(b)(2) of the CAA which specifies that secondary
22 standards provide requisite protection of "public welfare from any ... known or anticipated
23 adverse effects associated with the presence of [the] pollutant in the ambient air". As noted
24 above, our approach recognizes that the effects evidence reflects a general continuum from
25 higher 63 concentrations, at which scientists generally agree that adverse vegetation and
26 ecosystem effects are likely to occur, through lower concentrations, at which the likelihood and
27 magnitude of a response becomes increasingly more uncertain.
28 In this review, the evidence base includes quantitative information across a broad array of
29 vegetation effects (e.g., growth impairment during seedlings, saplings and mature tree growth
30 stages, visible foliar injury, and yield loss in annual crops) and across a diverse set of exposure
31 methods from laboratory and field studies. These methods include the more traditional OTC
32 studies, as well as field-based exposure studies. While we consider the full breadth of
33 information available, we place greater weight on U.S. studies due to the often species-, site-,
34 and climate-specific nature of (Vrelated vegetation responses. We especially weight those
35 studies that include Os exposures that fall within the range of those likely to occur in the ambient
36 air. Further, our approach in the context of the quantitative exposure and risk assessments
1-36
-------
1 (discussed below), places greatest emphasis on studies that have evaluated plant response over
2 multiple exposure levels and developed exposure-response relationships that allow the prediction
3 (estimation) of plant responses over the range of potential alternative standards being assessed.
4 In considering the evidence, we recognize differences across different study types in what
5 information they provide. For example, because conditions can be controlled in laboratory
6 studies, responses in such studies may be less variable and smaller differences may be easier to
7 detect. However, the control conditions may limit the range of responses or incompletely reflect
8 pollutant bioavailability, so they may not reflect responses that would occur in the natural
9 environment. Alternatively, field data can provide important information for assessments of
10 multiple stressors or where site-specific factors significantly influence exposure. They are also
11 often useful for analyses of larger geographic scales and higher levels of biological organization.
12 However, because most field study conditions can not be controlled, variability is expected to be
13 higher and differences harder to detect. The presence of confounding factors can also make it
14 difficult to attribute observed effects to specific stressors.
15 In considering information from across multiple lines of evidence, our approach is to first
16 integrate the evidence from both controlled and field-based studies and assess the coherence and
17 consistency across the available evidence for each effect. We then consider the extent to which
18 these identified effects should be considered adverse to the public welfare, relying largely on the
19 paradigm used in the 2008 review and 2010 proposed reconsideration (e.g., 75 FR 3006). This
20 paradigm recognizes that the significance to the public welfare of Cb-induced effects on sensitive
21 vegetation growing within the U.S. can vary depending on the nature of the effect, the intended
22 use of the sensitive plants or ecosystems, and the types of environments in which the sensitive
23 vegetation and ecosystems are located. Accordingly, any given (Vrelated effect on vegetation
24 and ecosystems (e.g., biomass loss, crop yield loss, foliar injury) may be judged to have a
25 different degree of impact on the public welfare depending, for example, on whether that effect
26 occurs in a Class I area, a city park, or commercial cropland. Our approach takes this variation in
27 the significance of Os-related vegetation effects into account in evaluating the currently available
28 evidence with regard to the extent to which it calls into question the adequacy of the current
29 standard and, as appropriate, indicates potential alternative standards that would be appropriate
30 for the Administrator to consider. In the 2010 proposed reconsideration, the Administrator
31 proposed to place the highest priority and significance on vegetation and ecosystem effects to
32 sensitive species that are known to or are likely to occur in federally protected areas such as
33 national parks and other Class I areas, or on lands set aside by States, Tribes and public interest
34 groups to provide similar benefits to the public welfare (75 FR 3023/24). Our approach in this
35 review considers whether newly available information would suggest any evolution to this
36 paradigm, in particular in the context of considering associated ecosystem services.
1-37
-------
1 Finally, our approach continues to give great weight to the scientific evidence available in
2 this and previous reviews indicating the relevance of cumulative, seasonal, concentration-
3 weighted exposures in inducing vegetation effects. Therefore, we continue to express exposures
4 in terms of the W126 index, and continue to consider the important policy implications regarding
5 selection of an appropriate exposure index for vegetation. Our approach also places primary
6 emphasis on studies that evaluated plant response to exposures that were or can be described
7 using such an index. The policy-relevant discussions in chapters 5 and 6 focus on vegetation
8 effects evidence and exposure/risk information that can be associated with cumulative, seasonal
9 peak-weighted exposures, where possible. Discussions pertaining to the adequacy of the current
10 secondary standard will consider what cumulative seasonal exposures would be allowed under
11 air quality that would just meet the current standard.
12 1.3.2.2.2 Consideration of Exposure and Risk Estimates and Air Quality Analyses
13 To put judgments about Os-related vegetation and ecosystem effects and services into a
14 broader public welfare context, we consider national scale exposure and risk assessments
15 described in the second draft WREA (U.S. EPA, 2014b). We particularly focused on the WREA
16 quantitative risks related to three types of vegetation effects: foliar injury, biomass loss, and crop
17 yield loss. These risks were assessed in a range of WREA analyses variously involving recent Oi
18 monitoring data and/or national-scale model-adjusted air quality scenarios for the current
19 secondary standard and, in some analyses, for a cumulative, seasonal W126 form at one or more
20 levels (15, 11 and 7 ppm-hours). Our consideration of these WREA results provide insight into
21 the extent to which the current or potential alternative standards would be expected to maintain
22 distributions of cumulative, seasonal O3 exposures below those associated with adverse
23 vegetation effects.
24 With regard to quantitative Os risks related to welfare effects and ecosystem services for
25 foliar injury, we consider two main analyses in the WREA: a screening-level assessment of 214
26 National Parks and a case study focused on three National Parks. In the screening-level
27 assessment, Os concentrations in national parks are assessed using criteria developed from a U.S.
28 Forest Service nationwide dataset on foliar injury, ambient Os concentrations (in terms of W126
29 index) and soil moisture (which can influence susceptibility of vegetation to foliar injury).
30 Additionally, we consider a case study for Class I areas (Great Smoky Mountain National Park,
31 Rocky Mountain National Park, and Sequoia/Kings Canyon National Park). We consider results
32 from this case study for three metrics: 1) percent of vegetation cover affected by foliar injury; 2)
33 percent of trails affected by foliar injury; 3) estimates of species specific biomass loss within the
34 case study area. We also consider qualitative analyses on ecosystem services effects for this
35 endpoint. For example, the second draft WREA uses GIS mapping to illustrate where effects
1-38
-------
1 may be occurring and relates those areas to national scale statistics for recreational use and data
2 on hiking trails, campgrounds and other park amenities that intersect with potentially affected
3 areas. These are used to identify impacts on ecosystem services associated with recreation in
4 national parks. We additionally consider analyses relating elevated 63 concentrations to
5 increased vulnerability to fire risk regimes, insect attacks and impacts on hydrological cycles.
6 With regard to risks related to biomass and crop yield loss, we consider WREA results
7 based on exposure-response functions for tree and crop species that predict the growth or yield
8 response of each species, based on the exposure patterns estimated within its growing region. To
9 compare exposure-response across species, genotypes or experiments for which absolute
10 response values may vary greatly, the second draft WREA instead uses estimates of relative
11 biomass loss for trees or yield loss for crops. The WREA develops such estimates nationally and
12 separately for more than 100 federally designated Class I areas. Additionally, we consider
13 WREA-developed estimates of associated impacts on the agriculture and forestry sectors
14 quantifying how O?, exposure to vegetation is estimated to affect the provision of timber and
15 crops and carbon sequestration. We consider estimates for impacts related to tree biomass loss on
16 ecosystem services such as pollution removal, carbon storage and sequestration in five urban
17 case study areas. We consider biomass and crop yield loss estimates in light of advice from
18 CASAC, as discussed in sections 5.3 and 5.4 below.
19 In considering the amount of weight to place on the estimates of exposures and risks at or
20 above specific W126 values described in the second draft WREA, our approach: 1) evaluates the
21 weight of the scientific evidence concerning vegetation effects associated with those Os
22 exposures; 2) considers the importance, from a public welfare perspective, of the Os-induced
23 effects on sensitive vegetation and associated ecosystem services that are known or anticipated to
24 occur as a result of exposures at selected W126 values; and, 3) recognizes that predictions of
25 effects associated with any given Oj, exposure may be mitigated or exacerbated by actual
26 conditions in the field (i.e., co-occurring modifying environmental and genetic factors). When
27 considering analyses in the second draft WREA that involve discrete exposure levels or varying
28 levels of severity of effects, our approach recognizes that welfare-relevant exposures are more
29 appropriately viewed as a continuum with greater confidence and less uncertainty about the
30 existence of welfare effects at higher O?, exposure concentrations and less confidence and greater
31 uncertainty as one considers lower exposure concentrations. We recognize that there is no sharp
32 breakpoint within the continuum ranging from concentrations at and above the level of the
33 current secondary standard down to the lowest cumulative, seasonal W126 value assessed. In
34 considering these results in this second draft PA, we consider both concerns about the potential
35 for welfare effects and their severity with the increasing uncertainty associated and our
36 understanding of the likelihood of such effects following exposures to lower Os concentrations.
1-39
-------
1 1.3.2.2.3 Considerations Regarding Ambient Os Concentration Estimates
2 Attributable to Background Sources
3 As noted above, our approach in this review utilizes recent advances in modeling
4 techniques to estimate the contributions of U.S. anthropogenic, international anthropogenic, and
5 natural sources to ambient O3 (discussed in detail in Chapter 2 of this document). Such model
6 estimates can provide insights into the extent to which different types of emissions sources
7 contribute to total ambient O3 concentrations. Our consideration of this issue in the current
8 review is informed by the approaches taken in previous reviews, and by court decisions on
9 subsequent litigation, as discussed in section 1.3.1.2.3 above.
10 Further, in the 1996 proposal, O3 background concentrations were one of the factors the
11 Administrator considered in selecting the SUM06 index as a form for an alternative secondary
12 standard. This and other cumulative exposure indices under consideration were judged to be
13 equally capable at estimating exposures relevant to vegetation, given the lack of evidence for a
14 discernible threshold for vegetation effects in general (U.S. EPA 1996, p. 225), which might
15 have provided a scientific basis for selecting among different cumulative exposure indices. At
16 that time, the SUM06 metric was selected over the W126 metric because it focused on the
17 policy-relevant (above background) portion of the total cumulative seasonal exposures reaching
18 plants (62 FR 38856). At the conclusion of that review, the Administrator ultimately chose to set
19 the secondary standard identical to the primary standard, including using the 8-hour average
20 instead of a cumulative seasonal form (62 FR 38868). In the 2008 review, staff analyses
21 concluded that the W126 index was more biologically-relevant based on the available science;
22 staff additionally noted, based on then-available estimates of background, that this form was also
23 not likely to be significantly impacted by background concentrations given the very low weight
24 assigned to lower O3 concentrations by the W126 index (2007 SP, 7-22; 72 FR 37893).
25 1.3.3 Organization of this Document
26 Chapter 2 of this second draft PA provides an overview of the O3 ambient monitoring
27 network and O3 air quality, including estimates of O3 concentrations attributable to background
28 sources. The remaining chapters are organized into two main parts. Chapters 3 and 4 focus on the
29 review of the primary O3 NAAQS while chapters 5 and 6 focus on the review of the secondary
30 O3 NAAQS. Staffs considerations and conclusions related to the current primary and secondary
31 standards are discussed in chapters 3 and 5, respectively. Staffs considerations and conclusions
32 related to potential alternative primary and secondary standards are discussed in chapters 4 and
33 6, respectively. Key uncertainties in the review and areas for future research and data collection
34 are additionally identified in chapters 4 and 6 for the two types of standards.
1-40
-------
1 1.4 REFERENCES
2 Frey, H.C. and Samet, J.M. (2012a) CASAC Review of the EPA's Policy Assessment for the Review of the Ozone
3 National Ambient Air Quality Standards (First External Review Draft - August 2012). EP A-C AS AC-13 -
4 003. November 26, 2012.
5 Frey, H.C. and Samet, J.M. (2012b) CASAC Review of the EPA's Health Risk and Exposure Assessment for Ozone
6 (First External Review Draft - Updated August 2012) and Welfare Risk and Exposure Assessment for
7 Ozone (First External Review Draft - Updated August 2012). EPA-CASAC-13-002. November 19, 2012
8 Henderson, R. (2006) Letter from CASAC Chairman Rogene Henderson to EPA Administrator Stephen Johnson.
9 October 24, 2006, EPA-CASAC-07-001.
10 Henderson, R. (2007) Letter from CASAC Chairman Rogene Henderson to EPA Administrator Stephen Johnson.
11 March 26, 2007, EPA-CASAC-07-002.
12 Henderson, R. (2008) Letter from CASAC Chairman Rogene Henderson to EPA Administrator Stephen Johnson.
13 April 7, 2008, EPA-CASAC-08-001.
14 Samet, J.M. (2011) Clean Air Scientific Advisory Committee (CASAC) Response to Charge Questions on the
15 Reconsideration of the 2008 Ozone National Ambient Air Quality Standards. EPA-CASAC-11-004.
16 March 30, 2011. Available online at:
17 http://yosemite.epa.gOv/sab/sabproduct.nsf/0/F08BEB48Cl 139E2A8525785E006909AC/$File/EP A-
18 CASAC-ll-004-unsigned+.pdf
19 U.S. Department of Health, Education, and Welfare (1970). Air Quality Criteria for Photochemical Oxidants.
20 Washington, D.C.: National Air Pollution Control Administration; publication no. AP-63. Available from:
21 NTIS, Springfield, VA; PB-190262/BA.
22 U.S. EPA (U.S. Environmental Protection Agency). (1978). Air quality criteria for ozone and other photochemical
23 oxidants [EPA Report]. (EPA/600/8-78/004). Washington, D.C..
24 U.S. EPA (U.S. Environmental Protection Agency). (1982). Air quality criteria document for ozone and other
25 photochemical oxidants. Fed Reg 47: 11561.
26 U.S. EPA (U.S. Environmental Protection Agency). (1986). Air quality criteria for ozone and other photochemical
27 oxidants [EPA Report]. (EPA-600/8-84-020aF - EPA-600/8-84-020eF). Research Triangle Park, NC.
28 http://www.ntis.gov/search/product.aspx? ABBR=PB87142949
29 U.S. Environmental Protection Agency. (1989) Review of the National Ambient Air Quality Standards for Ozone:
30 Policy Assessment of Scientific and Technical Information. OAQPS Staff Paper. Office of Air Quality
31 Planning and Standards, Research Triangle Park, NC.
32 U.S. EPA. (1996) Air Quality Criteria for Ozone and Related Photochemical Oxidants Volume I of III (Final, 1996).
33 U.S. Environmental Protection Agency, Washington, D.C., EPA/600/AP-93/004aF (NTIS PB94173127).
34 U.S. EPA. (2004) Air Quality Criteria for Paniculate Matter (Final Report). U.S. Environmental Protection Agency,
35 Washington, D.C., EPA 600/P-99/002aF-bF, 2004.
36 U.S. Environmental Protection Agency. (2006). Air quality criteria for ozone and related photochemical oxidants
37 [EPA Report]. (EPA/600/R-05/004AF). Research Triangle Park, NC.
3 8 http://cfpub.epa.gov/ncea/cfm/recordisplay.cfm?deid=149923
1-41
-------
1 U.S. Environmental Protection Agency. (2007) Review of the National Ambient Air Quality Standards for Ozone:
2 Policy Assessment of Scientific and Technical Information. OAQPS Staff Paper. Office of Air Quality
3 Planning and Standards, Research Triangle Park, NC. EPA-452/R-07-007.
4 U.S. Environmental Protection Agency. (201 la). Integrated Review Plan for the O3 National Ambient Air Quality
5 Standards (IRP) U. S. Environmental Protection Agency, National Center for Environmental Assessment
6 Office of Research and Development and Office of Air Quality Planning and Standards Office of Air and
7 Radiation, Research Triangle Park, North Carolina EPA. 452/R-l 1-006.
8 U.S. Environmental Protection Agency. (201 Ib) Integrated Science Assessment for Ozone and Related
9 Photochemical Oxidants: First External Review Draft, U.S. Environmental Protection Agency,
10 Washington, D.C., EPA/600/R-10/076A.
11 U.S. Environmental Protection Agency. (20 lie) Integrated Science Assessment of Ozone and Related
12 Photochemical Oxidants (Second External Review Draft). U.S. Environmental Protection Agency,
13 Washington, D.C., EPA/600/R-10/076B.
14 U.S. Environmental Protection Agency. (201 Id) Policy Assessment for the Review of the Paniculate Matter
15 National Ambient Air Quality Standards. Office of Air Quality Planning and Standards, Research Triangle
16 Park, NC. EPA 452/R-l 1-003.
17 U.S. Environmental Protection Agency. (2012a). Integrated Science Assessment for Ozone and Related
18 Photochemical Oxidants: Third External Review Draft, U.S. Environmental Protection Agency, Research
19 Triangle Park, NC. EPA/600/R-10/076C
20 U.S. Environmental Protection Agency. (2012b). Health Risk and Exposure Assessment for Ozone, First External
21 Review Draft, U.S. Environmental Protection Agency, Research Triangle Park, NC. EPA 452/P-12-001.
22 U.S. Environmental Protection Agency. (2013). Integrated Science Assessment for Ozone and Related
23 Photochemical Oxidants (Final Report). U.S. Environmental Protection Agency, Washington, D.C.,
24 EPA/600/R-10/076F, 2013
25 U.S. Environmental Protection Agency. (2014a). Health Risk and Exposure Assessment for Ozone, Second External
26 Review Draft. Office of Air Quality Planning and Standards, Research Triangle Park, NC. EPA-452/P-14-
27 004a.
28 U.S. Environmental Protection Agency. (2014b). Welfare Risk and Exposure Assessment for Ozone, Second
29 External Review Draft. Office of Air Quality Planning and Standards, Research Triangle Park, NC. EPA-
30 452/P-14-003a.
1-42
-------
1 2 O3 MONITORING AND AIR QUALITY
2 This section provides overviews of ambient Os monitoring in the U.S. (section 2.1); Os
3 precursor emissions and atmospheric chemistry (section 2.2); ambient Os concentrations (section
4 2.3); and available evidence and information related to background Os (section 2.4). These
5 issues are also discussed in detail in chapter 3 of the ISA (US EPA, 2013).
6 2.1 O3 MONITORING
7 2.1.1 Os Monitoring Network
8 To monitor compliance with the NAAQS, state and local environmental agencies operate
9 Os monitoring sites at various locations, depending on the population of the area and typical peak
10 Os concentrations1. In 2010, there were over 1,300 state, local, and tribal Os monitors reporting
11 concentrations to EPA. In areas for which Os monitors are required, at least one site must be
12 designed to record the maximum concentration for that particular metropolitan area. Since Os
13 concentrations are usually significantly lower in the colder months of the year, 63 is required to
r\
14 be monitored only during the Os monitoring season, which varies by state.
15 Figure 2-1 shows the locations of the U.S. ambient Os monitoring sites reporting data to
16 EPA at any time during the 2006-2010 period. The gray dots which make up over 80% of the Os
17 monitoring network are "State and Local Monitoring Stations" (SLAMS) monitors, which are
18 operated by state and local governments to meet regulatory requirements and provide air quality
19 information to public health agencies. Thus, the SLAMS monitoring sites are largely focused on
20 urban areas. The blue dots highlight two important subsets of monitoring sites within the
21 SLAMS network: the "National Core" (NCore) multi-pollutant monitoring network and the
22 "Photochemical Assessment Monitoring Stations" (PAMS) network.
23 While the existing U.S. Os monitoring network has a largely urban focus, to address
24 ecosystem impacts of Os such as biomass loss and foliar injury, it is equally important to focus
25 on Os monitoring in rural areas. The green dots in Figure 2-1 represent the Clean Air Status and
26 Trends Network (CASTNET) monitors which are located in rural areas. There were about 80
27 CASTNET sites operating in 2010, with sites in the eastern U.S. being operated by EPA and
28 sites in the western U.S. being operated by the National Park Service (NPS). Finally, the black
29 dots represent "Special Purpose Monitoring Stations" (SPMS), which include about 20 rural
30 monitors as part of the "Portable Os Monitoring System" (POMS) network operated by the NPS.
1 The minimum O3 monitoring network requirements for urban areas are listed in Table D-2 of Appendix D to 40
CFR Part 58.
2 The required O3 monitoring seasons for each state are listed in Table D-3 of Appendix D to 40 CFR Part 58.
2-1
-------
1
2
3
Between the CASTNET, NCore, and POMS networks, there were about 120 rural
sites operating in the U.S. in 2010.
monitoring
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
SLAMS
CASTNET
NCORE/PAMS
SPMS/OTHER
Figure 2-1. Map of U.S. ambient O3 monitoring sites reporting data to EPA during the
2006-2010 period.
2.1.2 Recent Os Monitoring Data and Trends
To determine whether or not the Os NAAQS has been met at an ambient monitoring site,
a statistic commonly referred to as a "design value" must be calculated based on 3 consecutive
years of data collected from that site. The form of the existing O^ NAAQS design value statistic
is the 3-year average of the annual 4th highest daily maximum 8-hour 63 concentration in parts
per billion (ppb), with decimal digits truncated. The existing primary and secondary Os NAAQS
are met at an ambient monitoring site when the design value is less than or equal to 75 ppb.3 In
counties or other geographic areas with multiple monitors, the area-wide design value is defined
as the design value at the highest individual monitoring site, and the area is said to have met the
NAAQS if all monitors in the area are meeting the NAAQS.
Figure 2-2 shows the trend in the annual 4th highest daily maximum 8-hour Oj
concentrations in ppb based on 933 "trends" sites with complete data records over the 2000 to
3For more details on the data handling procedures used to calculate design values for the existing O3 NAAQS, see
40 CFR Part 50, Appendix P.
2-2
-------
1 2012 period. The center line in this figure represents the median value across the trends sites,
2 while the dashed lines represent the 25th and 75th percentiles, and the bottom and top lines
3 represent the 10* and 90* percentiles. Figures 2-3 and 2-4 show maps of the Oj, design values
4 (ppb) at all U.S. monitoring sites for the 2009-2011 and 2010-2012 periods, respectively. The
5 trend shows that the annual 4* highest daily maximum values decreased for the vast majority of
6 monitoring sites in the U.S. between 2000 and 2009. The decreasing trend is especially sharp
7 from 2002 to 2004, when EPA implemented the "NOx SIP Call", a program designed to reduce
8 summertime emissions of NOx in the eastern U.S.
9 The trends also show a modest increase in the 4th highest daily maximum values from
10 2009 to 2012. This is reflected in the design value maps, which show an increase in the number
11 of monitors violating the existing O?, standard in 2010-2012 relative to 2009-2011. Meteorology
12 played an important role in these trends. 63 concentrations tend to be higher on days with hot
13 and stagnant conditions and lower on days with cool or wet conditions. According to the
14 National Oceanic and Atmospheric Administration's National Climactic Data Center (NOAA-
15 NCDC), the summer of 2009 was cooler and wetter than average over most of the eastern U.S.,
16 while conversely, the summers, of 2010, 2011, and 2012 were all much warmer than average. In
17 particular, the central and eastern U.S. experienced a 2-week period of record-breaking heat in
18 late June and early July of 2012, which contributed to hundreds of violations of the existing 63
19 standard. In contrast, the most recent climatological information available from NOAA-NCDC
20 (http://www.ncdc.noaa.gov/sotc/) shows that the summer of 2013 was cooler and wetter than
21 average for much of the U.S. Thus, EPA does not expect the recent increasing trend in the 4*
22 highest daily maximum 03 concentrations to continue in 2013.
2-3
-------
2
3
o
CD -
CD
CD
Q-
Q.
.0 CO
—
0)
or-
O
Trend in Annual 4th Highest Daily Maximum 8-hour O3 Concentrations
National Trend Based on 933 Monitoring Sites
— i — — r~ — i — ~~i — ~~i — — i — ~~r~ — i — — T—
2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012
Year
Figure 2-2.
th
Trend in U.S. annual 4 highest daily maximum 8-hour Os concentrations in
ppb, 2000 to 2012.
8-hour Ozone Design Values
• 41-60 ppb (92 sites)
• 61 - 65 ppb (179 sites)
O 66-70 ppb (324 sites)
71 - 75 ppb (302 sites)
• 76- 107 ppb (205 sites)
5 Figure 2-3. Map of 8-hour Os design values in ppb for the 2009-2011 period.
2-4
-------
8-hour Ozone Design Values
• 43-60 ppb (87 sites)
©61-65 ppb (130 sites)
O 66- 70 ppb (237 sites)
71 - 75 ppb (309 sites)
76- 106 ppb (363 sites)
2 Figure 2-4. Map of 8-hour Os design values in ppb for the 2010-2012 period.
3 In addition, EPA focused our analyses of welfare and ecosystem effects on a W126 63
4 exposure metric in this review. The W126 metric4 is a seasonal aggregate of daytime (8:00 AM
5 to 8:00 PM) hourly 63 concentrations, designed to measure the cumulative effects of 63
6 exposure on vulnerable plant and tree species, with units in parts per million-hours (ppm-hrs).
7 The W126 metric uses a logistic weighting function to place less emphasis on exposure to low
8 hourly 63 concentrations and more emphasis on exposure to high hourly 63 concentrations
9 (Lefohnetal, 1988).
10 Figure 2-5 shows the trend in annual W126 concentrations in ppm-hrs based on 933
11 "trends" sites with complete data records over the 2000 to 2012 period. Figures 2-6 and 2-7
12 show maps of the 3-year average annual W126 concentrations in ppm-hrs at all U.S. monitoring
13 sites for the 2009-2011 and 2010-2012 periods, respectively. The general patterns seen in these
14 figures are similar to those seen in the design value metric for the existing standard.
15
4 Details on the procedure used to calculate the W126 metric are provided in Chapter 4 of the welfare Risk and
Exposure Assessment.
2-5
-------
Trend in Annual W126 Concentrations
National Trend Based on 933 Monitoring Sites
2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012
I Year
2 Figure 2-5. Trend in U.S. annual W126 concentrations in ppm-hrs, 2000 to 2012.
Average Annual W126 Values
• 0-3 pprn-hr (79 sites)
©4-7 ppm-hr (328 sites)
O 8- 11 ppm-hr (367 sites)
12- 15 ppm-hr (194 sites)
• 16-58 ppm-hr (128 sites)
.•cs»
4 Figure 2-6. Map of 2009-2011 average annual W126 values in ppm-hrs.
2-6
-------
Average Annual W126 Values
• 0-3 ppm-hr (68 sites)
• 4-7 ppm-hr (217 sites)
O 8 - 11 ppm-hr (314 sites)
• 12- 15 ppm-hr (260 sites)
• 16-59 pprn-hr (257 sites)
3 Figure 2-7. Map of 2010-2012 average annual W126 values in ppm-hrs.
4 2.2 EMISSIONS AND ATMOSPHERIC CHEMISTRY
5 63 is formed by photochemical reactions of precursor gases and is not directly emitted
6 from specific sources. In the stratosphere, 63 occurs naturally and provides protection against
7 harmful solar ultraviolet radiation. In the troposphere, near ground level, O^ forms through
8 atmospheric reactions involving two main classes of precursor pollutants: non-methane volatile
9 organic compounds (NMVOCs) and nitrogen oxides (NOx). Carbon monoxide (CO) and
10 methane (CH/i) are also important for O^ formation over longer time periods (US EPA, 2013,
11 section 3.2.2).
12 Emissions of Os precursor compounds can be divided into anthropogenic and natural
13 source categories, with natural sources further divided into biogenic emissions (from vegetation,
14 microbes, and animals) and abiotic emissions (from biomass burning, lightning, and geogenic
15 sources). Anthropogenic sources, including mobile sources and power plants, account for the
16 majority of NOx and CO emissions. Anthropogenic sources are also important for NMVOC
17 emissions, though in some locations and at certain times of the year (e.g., southern states during
18 summer) the majority of NMVOC emissions come from vegetation (US EPA, 2013, section
19 3.2.1). In practice, the distinction between natural and anthropogenic sources is often unclear, as
20 human activities directly or indirectly affect emissions from what would have been considered
2-7
-------
1 natural sources during the preindustrial era. Thus, emissions from plants, animals, and wildfires
2 could be considered either natural or anthropogenic, depending on whether emissions result from
3 agricultural practices, forest management practices, lightning strikes, or other types of events
4 (US EPA, 2013, sections 3.2 and 3.7.1).
5 Rather than varying directly with emissions of its precursors, Os changes in a nonlinear
6 fashion with the concentrations of its precursors. NOX emissions lead to both the formation and
7 destruction of 63, depending on the local quantities of NOX, NMVOC, radicals, and sunlight. In
8 areas dominated by fresh emissions of NOX, these radicals are removed, which lowers the Os
9 formation rate. In addition, the scavenging of 63 by reaction with NO is called "titration" and is
10 often found in downtown metropolitan areas, especially near busy streets and roads, and in
11 power plant plumes. This short-lived titration results in localized areas in which O?,
12 concentrations are suppressed compared to surrounding areas, but which contain NC>2 that
13 contributes to Os formation later and further downwind. Consequently, Os response to
14 reductions in NOX emissions is complex and may include O^ decreases at some times and
15 locations and increases of 63 at other times and locations. In areas with low NOX to VOC ratios,
16 such as those found in remote continental areas and rural and suburban areas downwind of urban
17 centers, 63 production typically varies directly with NOX concentrations (e.g. increases with
18 increasing NOX emissions).
19 At some times and in some locations, reductions in 03 precursors may also yield
20 reductions in ambient air pollutants other than 63. For example, given that NOX emissions
21 contribute to ambient NC>2 (i.e., both because NO2 is a component of NOX emissions and because
22 NO can convert rapidly to NC^), reductions in directly emitted NOX will also result in reductions
23 in ambient NC>2. In addition, NOX and VOCs can contribute to secondary formation of PM2.5
24 constituents. NOX can act as both a direct precursor to NFLJSTCb, and can affect the formation of
25 other PlV^.s constituents because it adds to the oxidative capacity of the atmosphere5. The effects
26 of reducing NOX emissions on ambient PM2.5 concentrations can vary in time and space, with the
27 largest reductions in ambient PM2.5 likely occurring at times when and in locations where
28 concentrations of NFLjNCb are highest. This is usually during the cooler times of the year (e.g.
29 April-November) and in some areas of California, Salt Lake City, The Great Lakes States, and
30 the Northeast corridor between Baltimore and New York City (Carlton et al, 2010)6.
31 The formation of 63 from precursor emissions is also affected by the intensity and
32 spectral distribution of sunlight and atmospheric mixing. Major episodes of high ground-level
33 63 concentrations in the eastern United States are associated with slow-moving high pressure
5 Across North America, approximately 7% of summertime PM2 5 mass is estimated to result from anthropogenic
NOX emissions and up to 0.5 ug/m3 of secondary organic aerosol is estimated to form from NOx emissions (Carlton
etal,2010).
6 In these locations, NH4NO3 contributes more than 30% to average PM2 5 concentrations.
2-8
-------
1 systems. High pressure systems during the warmer seasons are associated with the sinking of
2 air, resulting in warm, generally cloudless skies, with light winds. The sinking of air results in
3 the development of stable conditions near the surface which inhibit or reduce the vertical mixing
4 of Os precursors. The combination of inhibited vertical mixing and light winds minimizes the
5 dispersal of pollutants emitted in urban areas, allowing their concentrations to build up. In
6 addition, in some parts of the United States (e.g., in Los Angeles), mountain barriers limit mixing
7 and result in a higher frequency and duration of days with high Os concentrations.
8 Photochemical activity involving precursors is enhanced during warmer seasons because of the
9 availability of sunlight and higher temperatures (US EPA, 2013, section 3.2).
10 Os concentrations in a region are affected both by local formation and by transport of 63
11 and its precursors from upwind areas. Os transport occurs on many spatial scales including local
12 transport between cities, regional transport over large regions of the U.S. and internal! onal/long-
13 range transport. In addition, Os can be transferred into the troposphere from the stratosphere,
14 which is rich in 03, through stratosphere-troposphere exchange (STE). These inversions or
15 "folds" usually occur behind cold fronts, bringing stratospheric air with them and typically affect
16 Os concentrations in high elevation areas (e.g. > 1500 m) more than areas at low elevations (U.S.
17 EPA, 2012, section 3.4.1.1). The role of long-range transport of ozone and other elements of
18 ozone background is discussed in more detail in Section 2.4,
19 2.3 AIR QUALITY CONCENTRATIONS
20 Because 63 is a secondary pollutant formed in the atmosphere from precursor emissions,
21 concentrations are generally more regionally homogeneous than concentrations of primary
22 pollutants emitted directly from stationary and mobile sources (US EPA, 2013, section 3.6.2.1).
23 However, variation in local emissions characteristics, meteorological conditions, and topography
24 can result in daily and seasonal temporal variability in ambient Os concentrations, as well as
25 local and national-scale spatial variability.
26 Temporal variation in ambient Os concentrations results largely from daily and seasonal
27 patterns in sunlight, precursor emissions, atmospheric stability, wind direction, and temperature
28 (US EPA, 2013, section 3.7.5). On average, ambient Os concentrations follow well-recognized
29 daily and seasonal patterns, particularly in urban areas. Specifically, daily maximum Os
30 concentrations in urban areas tend to occur in mid-afternoon, with more pronounced peaks in the
31 warm months of the Os season than in the colder months (US EPA, 2013, Figures 3-54, 3-156 to
32 3-157). Rural sites also followed this general pattern, though it is less pronounced in colder
33 months (US EPA, 2013, Figure 3-55). With regard to day-to-day variability, median maximum
34 daily average 8-hour (MDA8) Os concentrations in U.S. cities in 2007-2009 were approximately
2-9
-------
1 47 ppb, with typical ranges between 35 to 60 ppb and the highest MDA8 concentrations above
2 100 ppb in several U.S. cities (as noted further below).
3 In addition to temporal variability, there is considerable spatial variability in ambient O^
4 concentrations within cities and across different cities in the United States. With regard to
5 spatial variability within a city, local emissions characteristics, geography, and topography can
6 have important impacts. For example, as noted above, fresh NO emissions from motor vehicles
7 titrate 63 present in the urban background air, resulting in an 63 gradient around roadways with
8 Os concentrations increasing as distance from the road increases (US EPA, 2013, section
9 3.6.2.1). In comparing urban areas, the ISA notes that measured 63 concentrations are relatively
10 uniform and well-correlated across some cities (e.g., Atlanta) while they are more variable in
11 others (e.g., Los Angeles) (US EPA, 2013, section 3.6.2.1 and Figures 3-28 to 3-36).
12 With regard to spatial variability across cities, when the ISA evaluated the distributions
13 of 8-hour Os concentrations for the years 2007 to 2009 in 20 cities, the highest concentrations
14 were reported in Los Angeles, with high concentrations also reported in several eastern and
15 southern cities. The maximum recorded MDA8 was 137 ppb in Los Angeles, and was near or
16 above 120 ppb in Atlanta, Baltimore, Dallas, New York City, Philadelphia, and St. Louis (US
17 EPA, 2013, Table 3-10). The pattern was similar for the 98th percentile of the distribution of
18 MDA8 concentrations7, with Los Angeles recording the highest 98th percentile concentration (91
19 ppb) and many eastern and southern cities reporting 98* percentile concentrations near or above
20 75 ppb. In contrast, somewhat lower 98th percentile 63 concentrations were recorded in cities in
21 the western United States outside of California (US EPA, 2013, Table 3-10).
22 Although rural monitoring sites tend to be less directly affected by anthropogenic
23 pollution sources than urban sites, rural sites can be affected by transport of 63 or 63 precursors
24 from upwind urban areas and by local anthropogenic sources such as motor vehicles, power
25 generation, biomass combustion, or oil and gas operations (US EPA, 2013, section 3.6.2.2). In
26 addition, 63 tends to persist longer in rural than in urban areas due to lower rates of chemical
27 scavenging in non-urban environments. At higher elevations, increased O^ concentrations can
28 also result from stratospheric intrusions (US EPA, 2013, sections 3.4, 3.6.2.2). As a result, 63
29 concentrations measured in some rural sites can be higher than those measured in nearby urban
30 areas (US EPA, 2013, section 3.6.2.2), and the ISA concludes that cumulative exposures for
31 humans and vegetation in rural areas can be substantial, often higher than cumulative exposures
32 in urban areas (US EPA, 2013, section 3.7.5).
7 Table 3-10 in the ISA analyzes the warm season. Therefore, the 98th percentile values would be an approximation
of the 4th highest value.
2-10
-------
1 2.4 BACKGROUND O3
2 Generically, background Os can originate from natural sources of 63 and 63 precursors,
3 as well as from manmade international emissions of Os precursors. Natural sources of Os
4 precursor emissions such as wildfires, lightning, and vegetation can lead to 63 formation by
5 chemical reactions with other natural sources. Another important natural component of
6 background is Os that is naturally formed in the stratosphere through interactions of ultraviolet
7 light with molecular oxygen. Stratospheric 63 can mix down to the surface at high
8 concentrations in discrete events called intrusions, especially at higher-altitude locations. The
9 manmade portion of the background includes any 03 formed due to anthropogenic sources of Os
10 precursors emitted far away from the local area (e.g., international emissions). Finally, both
11 biogenic and international anthropogenic emissions of methane, which can be chemically
12 converted to 63 over relatively long time scales, can also contribute to global background 63
13 levels.
14 As indicated in the first draft policy assessment (US EPA, 2012, sections 1.3.4 and 3),
15 EPA has updated several aspects of our methodology for estimating the change in health risk and
16 exposure that would result from a revision to the Os NAAQS. First, risk estimates are now based
17 on total Os concentrations, as opposed to previous reviews which only considered risk above
18 background levels. Second, EPA is now using air quality models to estimate the spatial patterns
19 of Os that would result from attaining various levels of the NAAQS, as opposed to a quadratic
20 rollback approach that required the estimation of a background "floor" beyond which the
21 rollback would not take place. Both of these revisions have had the indirect effect of reducing
22 the need for estimates of background 03 levels as part of the Os risk and exposure assessment
23 (REA). Regardless, EPA expects that a well-founded understanding of the fractional
24 contribution of background sources and processes to surface Os levels will be valuable towards
25 informing policy decisions about the Os NAAQS. Accordingly, in this section, we briefly
26 summarize existing results on background Os from the ISA (US EPA, 2013, section 3.4) as
27 supplemented by additional EPA modeling recently conducted for a 2007 base year. The
28 summary will focus on national estimates of: 1) seasonal mean background Os concentrations for
29 three specific definitions of background Os, 2) the relative proportion of background Os to total
30 Os for the same three definitions from a seasonal mean perspective, 3) the distributions of
31 background Os within a seasonal mean, 4) the fractional background Os in the 12 REA urban
32 case study areas, 5) the relative proportion of background Os concentrations to total ozone from a
33 W126 perspective, and 6) the relative roles of different components of background Os.
34 The definition of background Os can vary depending upon context, but it generally refers
35 to Os that is formed by sources or processes that cannot be influenced by actions within the
36 jurisdiction of concern. In the first draft policy assessment document (US EPA, 2012), EPA
2-11
-------
1 presented three specific definitions of background 03: natural background, North American
2 background, and United States background. Natural background (NB) was the narrowest
3 definition of background, and it was defined as the O?, that would exist in the absence of any
4 manmade Os precursor emissions. The other two previously established definitions of
5 background presume that the U.S. has little influence over anthropogenic emissions outside our
6 continental or domestic borders. North American background (NAB) is defined as that O^ that
7 would exist in the absence of any manmade Oi precursor emissions from North America. U.S.
8 background (USB) is defined as that Os that would exist in the absence of any manmade
9 emissions inside the United States. Each of these three definitions of background 63 requires
10 photochemical modeling simulations to estimate what the residual 63 concentrations would be
11 were the various anthropogenic emissions to be removed. Previous modeling studies have
12 estimated what background levels would be in the absence of certain sets of emissions by simply
13 assessing the remaining Os in a simulation in which certain emissions were removed (Zhang et
14 al. (2011), Emery et al. (2012)). This basic approach is often referred to as "zero-out" modeling
15 or "emissions perturbation" modeling.
16 While the zero-out approach has traditionally been used to estimate background Os
17 levels, the methodology has some acknowledged limitations. First, from a policy perspective,
18 the hypothetical and unrealizable zero manmade emissions scenarios have limited application.
19 Secondly, the assumption that background Os is what is left after specific emissions have been
20 removed within the model simulation can be misleading in locations where Os chemistry is
21 highly non-linear. Depending upon the local composition of Os precursors, NOx emissions
22 reductions can either increase or decrease O?, levels in the immediate vicinity of those reductions.
23 For those specific urban areas in which NOx titration of 63 can be significant, zero-out modeling
24 can result in inflated estimates of background Os when these NOx emissions are completely and
25 unrealistically removed. Paradoxically, in certain times and locations in a zero-out scenario
26 there can be more background 63 than actual 63 within the model. A separate modeling
27 technique circumvents these limitations by apportioning the total Os within the model to its
28 contributing source terms. This basic approach, referred to as "source apportionment" modeling,
29 has been described and evaluated in the peer-reviewed literature (Dunker et al., 2002; Kemball-
30 Cook et al., 2009). While source apportionment modeling has not been previously used in the
31 context of estimating background ozone levels as part of an ozone NAAQS review, it has
32 frequently been used in other regulatory settings to estimate the "contribution" to ozone of
33 certain sets of emissions (EPA 2005, EPA 2011). The source apportionment technique provides
34 a means of estimating the contributions of user-identified source categories to ozone formation in
35 a single model simulation. This is achieved by using multiple tracer species to track the fate of
36 ozone precursor emissions (VOC and NOx) and the ozone formation caused by these emissions.
2-12
-------
1 The methodology is designed so that all ozone and precursor concentrations are attributed to the
2 selected source categories at all times without perturbing the inherent chemistry. EPA recently
3 completed updated zero-out and source apportionment modeling for a 2007 base year to
4 supplement our characterization of background 63 over the U.S. Prior to using model
5 simulations to estimate background Os levels over the U.S., EPA confirmed that the modeling
6 was able to reproduce historical O?, levels and that there was limited correlation between model
7 errors and the background estimates. The key findings from the updated modeling are described
8 below; a more detailed description of the modeling is provided in Appendix A.
9 2.4.1 Seasonal Mean Background O3 in the U.S.
10 The ISA (US EPA 2013, section 3.4) previously established that background
11 concentrations vary spatially and temporally and that simulated mean background concentrations
12 are highest at high-elevation sites within the western U.S. Background levels typically are
13 greatest over the U.S. in the spring and early summer. Figure 2-8 displays the spatial patterns of
o
14 seasonal mean natural background Os as estimated by a 2007 zero-out scenario. This figure
15 shows the average daily maximum 8-hour O^ concentration (MDA8) that would exist in the
16 absence of any anthropogenic 63 precursor emissions at monitor locations. Seasonal mean NB
17 levels range from approximately 15-35 ppb with the highest values at higher-elevation sites in
18 the western U.S. The median value over these locations is 24.2 ppb, and more than 50 percent of
19 the locations have natural background levels of 20-25 ppb. The highest modeled estimate of
20 seasonal average, natural background, MDA8 03 is 34.3 ppb at the high-elevation CASTNET
21 site (Gothic) in Gunnison County, CO. Natural background levels are higher at these high-
22 elevation locations primarily because natural stratospheric OT, impacts and international transport
23 impacts increase with altitude (where 03 lifetimes are longer).
8 The recent EPA modeling focused on the period from April through October. Seasonal means are computed over
those seven months.
2-13
-------
Ozone(ppb)
< 20 (104)
20-25 (740]
25-30(331!
30-35(119)
35 - 40 (0)
O 40-45 (0)
O "5-50 (0)
O 50-55 (0)
55-60 (0)
>60(0)
UG'iC ESRI.'TftWWND. £
2 Figure 2-8. Map of 2007 CMAQ-estimated seasonal mean natural background Os levels
3 from zero-out modeling.
4 Figures 2-9 and 2-10 show the same information for the NAB and USB scenarios. In
5 these model runs, all anthropogenic Os precursor emissions were removed from the U.S.,
6 Canada, and Mexico portions of the modeling domain (NAB scenario) and then only from the
7 U.S. (USB scenario). The figures show that there is not a large difference between the NAB and
8 USB scenarios. Seasonal mean NAB and USB Os levels range from 25-50 ppb, with the most
9 frequent values estimated in the 30-35 ppb bin. The median seasonal mean background levels
10 are 31.5 and 32.7 ppb (NAB and USB, respectively). Again, the highest levels of seasonal mean
11 background are predicted over the intermountain western U.S. Locations with NAB and USB
12 concentrations greater than 40 ppb are confined to Colorado, Nevada, Utah, Wyoming, northern
13 Arizona, eastern California, and parts of New Mexico. The 2007 EPA modeling suggests that
14 seasonal mean USB concentrations are on average 1-3 ppb higher than NAB background. These
15 results were similar to those reported by Wang et al. (2009). From a seasonal mean perspective,
16 background levels are typically well-below the NAAQS thresholds.
2-14
-------
Ozone (ppb)
• < 20 (0)
20-25 (0)
25-30 (396)
30-35 (628)
35-40 (147)
O 40-45(121)
O 45 - 5° (2)
50-55 (0)
55-60 (0)
> 60 (0)
Figure 2-9. Map of 2007 CMAQ-estimated seasonal mean North American background
Os levels from zero-out modeling.
Ozone(ppb)
< 20 (0)
20 - 25 (0)
25-30 (127)
30 - 35 (842)
35-40 (188)
O 40-45(132)
O 45-50(5)
50 - 55 (0)
55 - 60 (0)
> 60 (0)
5 Figure 2-10. Map of 2007 CMAQ-estimated seasonal mean United States background Os
6 levels from zero-out modeling.
7 2.4.2 Seasonal Mean Background O3 in the U.S. as a Proportion of Total O3
8 Another informative way to assess the importance of background as part of seasonal
9 mean O^ levels across the U.S. is to consider the fractional contribution of NB, NAB, and USB
10 to total modeled 63 at each monitoring location. Considering the proportional role of
2-15
-------
1 background allows for an initial assessment of the relative importance of background and non-
2 background sources. Figures 2-11 and 2-12 show the percent contribution of U.S. anthropogenic
3 sources to total O^ using the metric of the seasonal mean MDA8 O?, concentrations as estimated
4 by both the zero-out and source apportionment modeling methodologies. Recall that the terms
5 NB, NAB, and USB are explicitly linked to the zero-out modeling approach. For comparison
6 sake, in Figure 2-12 we are extending the definition of USB to also include the source
7 apportionment model estimates of the Oi that is attributable to sources other than U.S.
8 anthropogenic emissions. To preserve the original definition of USB, this second term will be
9 hereafter referred to as "apportionment-based USB". As noted earlier, the advantage of the
10 source apportionment modeling is that all of the modeled 63 is attributed to various source terms
11 and thus this approach is not affected by the confounding occurrences of background Os values
12 exceeding the base 63 values as can happen in the zero-out modeling (i.e., background
13 proportions > 100%). Consequently, one would expect the fractional background levels to be
14 lower in the source apportionment methodology as a result of removing this artifact.
15 When averaged over all sites, 63 from sources other than U.S. anthropogenic emissions
16 is estimated to comprise 66 (zero-out) and 59 (source apportionment) percent of the total
17 seasonal 63 mean. The spatial patterns of apportionment-based USB are similar across the two
18 modeling exercises. Background 63 is a relatively larger percentage (e.g., 70-80%) of the total
19 seasonal mean 03 in locations within the intermountain western U.S. and along the U.S. border.
20 In locations where 63 levels are generally higher, like California and the eastern U.S. the
21 seasonal mean background fractions are relatively smaller (e.g., 40-60%). The additional 2007
22 modeling confirms that background ozone, while generally not approaching levels of the ozone
23 standard, can comprise a considerable fraction of total seasonal mean ozone across the U.S.
24 2.4.3 Daily Distributions of Background Os within the Seasonal Mean
25 As a first-order understanding, it is valuable to be able to characterize seasonal mean
26 levels of background 63. However, it is well established that background levels can vary
27 substantially from day-to-day. From an implementation perspective, the values of background
28 03 on possible exceedance days are a more meaningful consideration. The first draft policy
29 assessment (US EPA, 2012) considered this issue in detail, via summaries of the existing 2006
30 zero-out modeling (Henderson et al., 2012), and concluded that "results suggest that background
31 concentrations on the days with the highest total O?, concentrations are not dramatically higher
32 than typical seasonal average background concentrations." Based on this finding, EPA
33 determined that "anthropogenic sources within the U.S. are largely responsible for 4th highest 8-
34 hour daily maximum O^ concentrations." The recent EPA modeling using a 2007 base year and
35 the two distinct modeling methodologies supports this finding from the previous 2006-based
2-16
-------
1 modeling analyses. That is, the highest modeled O?, site-days tend to have background Os levels
2 similar to mid-range Os days. Figure 2-13 and 2-14 show the distribution of daily MDA8
3 apportionment-based USB levels (absolute magnitudes and relative fractions, respectively) from
4 the CAMx source apportionment simulation9. Again, the 2007 modeling shows that the days
5 with highest Os levels have similar distributions (i.e., means, inter-quartile ranges) of
6 background levels as days with lower values down to approximately 40 ppb. As a result, the
7 proportion of total 63 that has background origins is smaller on high 63 days (e.g., days > 70
8 ppb) than the more common lower OT, days that drive seasonal means. This helps put the results
9 from section 2.4.2 into better context. For example, for site-days in which base 63 is between
10 70-75 ppb, the source apportionment modeling estimates that approximately 37 percent of those
11 Os levels originate from sources other than U.S. anthropogenic emissions (i.e., apportionment-
12 based USB). Figure 2-14 also indicates that there are cases in which the model predicts much
13 larger background proportions, as shown by the upper outliers in the figure. These infrequent
14 episodes usually occur in relation to a specific event, and occur more often in specific
15 geographical locations, such as at high elevations or wildfire prone areas during the local dry
16 season.
17 It should be noted here that EPA has policies for treatment of air quality monitoring data
18 affected by these types of events. EPA's exceptional events policy allows exclusion of certain
19 air quality monitoring data from regulatory determinations if a State adequately demonstrates
20 that an exceptional event has caused the exceedance or violation of a NAAQS. In addition,
21 Section 179B of the CAA also provides for treatment of air quality data from international
22 transport when an exceedance or violation of a NAAQS would not have occurred but for
23 emissions emanating from outside of the United States. From an overarching perspective, the
24 Clean Air Act requires the NAAQS to be set at a level requisite to protect public health and
25 welfare. Case law makes it clear that attainability and technical feasibility are not relevant
26 considerations in the setting of a NAAQS. In previous reviews, EPA has assessed the proximity
27 of ozone concentrations to peak background levels only as a secondary consideration between
28 potential threshold levels where health and welfare was determined to have been protected.
29
9 Similar plots from the zero-out modeling for natural background, North American background, and U.S.
background are provided in Appendix A.
2-17
-------
2 Figure 2-11. Map of U.S. background percent contribution to seasonal mean Os based on
3 2007 CMAQ zero-out modeling.
< 40% (0)
40 - 50% (343)
50 - 60% (483)
60 - 70% (237)
70-80% (178)
> 80% (52)
6 Figure 2-12. Map of apportionment-based U.S. background percent contribution to
7 seasonal mean Os based on 2007 CAMx source apportionment modeling.
2-18
-------
<25 25-30 30-35 35-40 40-45 45-50 50-55 55-60 SD-8E- SE-70 70-75 7E-SO BO-B5 85-90 90-95 95-100 > 100
Bins of Base Model MDA8 Ozone (ppb)
2 Figure 2-13. Distributions of absolute estimates of apportionment-based U.S. Background
3 (all site-days), binned by modeled MDA8 from the 2007 source
4 apportionment simulation.
•o
re
<25 25-30 30-35 35-40 40-45 45-50 50-55 55-60 60-65 65-70 70-75 75-80 80-85 85-90 90-95 95-100 >100
Bins of Base Model MDA8 Ozone (ppb)
7 Figure 2-14. Distributions of the relative proportion of apportionment-based U.S.
8 Background to total Os (all site-days), binned by modeled MDA8 from the
9 2007 source apportionment simulation.
2-19
-------
1
2 2.4.4 Proportion of Background Os in 12 Urban Case Study Areas
3 As noted in the introduction, the current approach to estimating changes in risk across
4 various possible levels of the 03 NAAQS no longer requires a quantitative estimate of
5 background 63 levels. However, EPA expects to be asked the question: "how much of the total
6 risk is due to background sources." This section of the policy assessment presents estimates of
7 the overall fraction of 03 that is estimated to result from background sources or processes in each
8 of the 12 urban case study areas considered in the epidemiological-based risk assessment of the
9 REA (US EPA 2014, Chapter 7). The results are based on the recent EPA 2007 source
10 apportionment modeling. Table 2-1 summarizes the estimated fractional contributions of sources
11 other than U.S. anthropogenic emissions (i.e., apportionment-based USB) to total seasonal mean
12 MDA8 03 in each of the 12 urban case study areas. The table shows that the fractional
13 contributions from sources other than anthropogenic emissions within the U.S. can range from
14 43 to 66 percent across these 12 urban areas. These fractions are consistent with the national
15 ratios summarized in section 2.4.2, although the fractions of background are generally smaller at
16 urban sites than at rural sites.
17 As shown in section 2.4.3, the fractional contributions from background are smaller on
18 days with high modeled 63 (i.e., days that may exceed the level of the NAAQS). Table 2-2
19 provides the fractional contributions from these apportionment-based USB sources, only
20 considering days in which base model MDA8 03 was greater than 60 ppb. As expected, the
21 fractional background contributions are smaller, ranging from 31 to 55 percent.
22 Rather than taking the fractions of the seasonal means (as in Table 2-1), Table 2-3
23 displays the mean and median daily MDA8 background fractions. These metrics may be more
24 appropriate for application to health studies. The fractional contributions to backgrounds
25 calculated via this approach are very similar to the Table 2-1 calculations. Although EPA
26 expects the source apportionment results to provide a more realistic estimate of fractional
27 background values, for completeness, we also provide USB fractions based on the zero-out
28 modeling for the 12 cities (see Table 2-4). The results are similar to the source apportionment
29 findings (Table 2-1), though the zero-out technique provides slightly higher background
30 proportions, as expected. It should be noted that all fractional contributions are based on recent
31 conditions from 2007. These fractional contributions would be expected to change as
32 anthropogenic emissions and 63 levels are lowered. Based on the source apportionment
33 modeling for these 12 areas, there is evidence that background levels comprise a non-negligible
34 fraction of the total ozone observed within these locations. However, for site-days in which
35 model MDA8 ozone exceeds 60 ppb, ozone formed from U.S. anthropogenic emissions comprise
2-20
-------
1
2
3
4
5
6
7
an even larger fraction of the total ozone in 11 of the 12 areas. The major metropolitan areas in
the eastern U.S. (e.g., Atlanta, New York City, Philadelphia) are less influenced by background
sources than a higher-elevation, western U.S., location like Denver. Even in Denver, though,
U.S. anthropogenic emissions have a large influence on total ozone (45 percent).
Table 2-1. Seasonal mean MDA8 O3 (ppb), seasonal mean apportionment-based USB
contribution (ppb), and fractional apportionment-based USB contribution to
total O3 (all site-days) in the 12 REA urban case study areas (%).
All days, CAMx
Model MDA8 seasonal mean
Model MDA8 seasonal mean
from emissions other than
U.S. anthropogenic sources
Fractional contribution from
background
ATL
59.3
25.3
0.43
BAL
54.4
25.9
0.48
BOS
43.0
26.2
0.61
CLE
48.9
25.7
0.52
DEN
47.3
31.3
0.66
DET
39.1
23.3
0.60
HOU
48.5
27.0
0.56
LA
51.1
29.1
0.57
NYC
45.4
24.5
0.54
PHI
48.7
24.2
0.50
SAC
46.4
29.7
0.64
STL
49.8
24.3
0.49
9
10
11
Table 2-2. Seasonal mean MDA8 Os (ppb), seasonal mean apportionment-based USB
contribution (ppb), and fractional apportionment-based USB contribution to
total Os (site-days > 60 ppb) in the 12 REA urban study areas (%).
Only days w/ base
MDA8 > 60 ppb
Model MDA8 seasonal mean
Model MDA8 seasonal mean
from emissions other than
U.S. anthropogenic sources
Fractional contribution from
background
ATL
74.0
25.4
0.34
BAL
75.3
23.7
0.31
BOS
70.7
24.4
0.35
CLE
72.0
25.4
0.35
DEN
67.5
37.3
0.55
DET
68.9
24.4
0.35
HOU
70.3
28.0
0.40
LA
74.4
31.9
0.43
NYC
74.1
23.5
0.32
PHI
74.0
22.9
0.31
SAC
68.3
32.1
0.47
STL
70.0
25.4
0.36
12
13
14
15
Table 2-3. Fractional contribution of apportionment-based USB in the 12 REA urban
study areas (%), using the means and medians of daily MDA8 fractions
(instead of fractions of seasonal means).
Mean of daily MDA8
background fractions
Median of daily MDA8
background fractions
ATL
0.46
0.43
BAL
0.53
0.51
BOS
0.68
0.73
CLE
0.58
0.54
DEN
0.69
0.69
DET
0.64
0.66
HOU
0.59
0.59
LA
0.61
0.60
NYC
0.61
0.63
PHI
0.56
0.54
SAC
0.67
0.66
STL
0.52
0.49
16
2-21
-------
1
2
Table 2-4. Seasonal mean MDA8 Os (ppb), seasonal mean USB contribution (ppb), and
fractional USB contribution to total Os (all site-days) in the 12 REA urban case
study areas (%).
All days, CMAQ
Model MDA8 seasonal mean
Model MDA8 seasonal mean
from USB emissions
Fractional contribution from
background
ATL
58.6
30.0
0.51
BAL
55.6
29.9
0.54
BOS
45.2
28.5
0.63
CLE
51.8
31.6
0.61
DEN
57.1
42.2
0.74
DET
43.5
31.7
0.73
HOU
49.4
33.0
0.67
LA
54.8
33.3
0.61
NYC
47.7
29.1
0.61
PHI
50.5
29 .4
0.58
SAC
51.9
34.4
0.66
STL
52.6
32.0
0.61
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
2.4.5 Influence of Background Os on W126 levels
EPA also conducted a limited assessment of the impacts of background 63 sources on the
W126 metric. The W126 metric (LeFohn et al., 1988) is a cumulative peak-weighted index
designed to estimate longer-term effects of daytime ozone levels on sensitive vegetation and
ecosystems. EPA used the 2007 zero-out modeling to assess NB, NAB, and USB influences at
four sample locations: Atlanta GA, Denver CO, Farmington NM, and Riverside CA. As shown
in Figure 2-7, each of the four analyses locations had relatively high observed values of W126 in
2010-2012.
As discussed above, in previous EPA reviews of the Os NAAQS, the influence of
background ozone was estimated according to a counterfactual assumption. Background 63 was
defined as the ozone that would exist in the absence of a particular set of emissions (e.g., NAB is
the ozone that would exist if there were no anthropogenic emissions in North America). In the
current review, EPA is supplementing the counterfactual assessment with analyses that estimate
the fraction of the existing ozone that is due to background sources. This has important
ramifications for assessing the influence of background on W126 concentrations, because of the
non-linear weighting function used in the metric, which emphasizes high ozone hours (e.g.,
periods in which ozone is greater than -60 ppb). As an example, consider a sample site in the
intermountain western U.S. region with very high modeled estimates of U.S background (e.g.,
seasonal mean of 45 ppb with some days as high as 65 ppb). Even at this high background
location, when the W126 calculation is made for the USB simulation, the resultant annual W126
(USB) values are quite low, on the order of 3 ppm-hrs. Sites in the domain with lower U.S.
background levels have even smaller USB W126 values, on the order of the 1 ppm-hrs, which is
consistent with values mentioned in past reviews (USEPA, 2007). Using the counterfactual
scenarios, background ozone has a relatively small impact on W126 levels across the U.S.
However, because of the non-linear weighting function used in the W126 calculation, the
sum of the W126 from the USB scenario and the W126 resulting from US anthropogenic sources
will not equal the total W126. In most cases, the sum of those two components will be
substantially less than total W126. As a result, EPA believes it is more informative to estimate
2-22
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
the fractional influence of background ozone to W126 levels. Using a methodology that is
described in more detail in Appendix A, EPA considered the fractional influence of background
ozone on annual W126 levels in four locations. The fractional influence methodology utilizes
the 2007 zero-out modeling but places higher weights on background fractions on days that are
going to contribute most substantially to the yearly W126 value. Figure 2-15 shows the results.
Based on the fractional influence methodology, natural background sources are estimated to
contribute 29-50% of the total modeled W126 with the highest relative influence in the
intermountain western U.S. (e.g., Farmington NM) and the lowest relative influence in the
eastern U.S. (e.g., Atlanta). U.S. background is estimated to contribute 37-65% of the total
modeled W126. The proportional impacts of background are slightly less for the W126 metric
than for seasonal mean MDA8 (discussed in section 2.4.2), because of the sigmoidal weighting
function that places more emphasis on higher ozone days when background fractions are
generally lower.
The key conclusion from this cursory analysis is that background ozone may comprise a
non-negligible portion of current W126 levels across the U.S. These fractional influences are
greatest in the intermountain western U.S. and are slightly smaller than the seasonal mean
MDA8 metric. In the counterfactual cases, when non-background sources are completely
removed, the remaining W126 levels are low (< 3 ppm-hrs).
INB
I NAB
USB
Farmington
Denver
Riverside
Atlanta
Figure 2-15. Fractional influence of background sources to W126 levels in four sample
locations. Model estimates based on 2007 CMAQ zero-out modeling.
2-23
-------
1 2.4.6 Estimated Magnitude of Individual Components of Background Os
2 While local and regional controls of manmade 63 precursor emissions are still expected
3 to be the most effective mechanism for reducing local Os levels, an understanding of the relative
4 contributions of various background elements can be instructive in determining ways to mitigate
5 the impact of background. This section will utilize the supplemental 2007 air quality modeling
6 estimates to consider the relative importance of specific elements of background 03. Several
7 background elements were isolated in either the zero-out or source apportionment modeling.
8 Appendix A provides more detail on these analyses. In conjunction with the previous analyses
9 summarized in the ISA, some broad characterizations of the individual components of
10 background 63 can be developed.
11 The recent 2007 EPA modeling confirms the importance of methane emissions in
12 contributing to background 63. Methane has an atmospheric lifetime of about a decade and can
13 be an important contributor to ozone on longer time scales. Anthropogenic methane emission
14 sources include agriculture, coal mines, landfills, and natural gas and oil systems. The difference
15 between the NAB and NB zero-out scenarios provides an estimate of contributions from
16 international anthropogenic emissions and anthropogenic methane, which is modeled by
17 reducing model concentrations from present-day values to pre-industrial levels. The ISA (US
18 EPA, 2013, section 3.4) estimated that roughly half of the difference between the NB and NAB
19 scenarios resulted from the removal of anthropogenic methane emissions and that the other half
20 resulted from international anthropogenic emissions of shorter-lived 63 precursors (e.g., NOx
21 and nmVOC). Figure 2-16 shows the difference in seasonal mean MDA8 O^ levels between the
22 NB and NAB scenarios. North American seasonal mean background is 6-15 ppb higher than
23 comparable natural background levels. The most frequent increment is an 8-10 ppb increase
24 when the methane is increased and the non-North American emissions are re-added. It is not
25 possible via the EPA 2007 modeling to parse out what fraction of this change is due to emissions
26 outside of North America, as opposed to the fraction due to anthropogenic methane emissions,
27 but the modeling suggests that both of these terms have the potential to contribute in an
28 important way to average background levels in the U.S.
29 The difference between the NAB and USB scenarios is easier to interpret as it only
30 involves one change, the inclusion of anthropogenic emissions from the in-domain portion of
31 Canada and Mexico. These emissions (not shown here, but depicted in Appendix A) contribute
32 less than 2 ppb to the seasonal mean MDA8 Os levels over most of the U.S. There are 70 sites,
33 near an international border, where the modeling estimates Canadian/Mexican seasonal average
34 impacts of 2-4 ppb. Peak single-day MDA8 impacts from these specific international emissions
35 sources can approach 25 ppb (e.g., San Diego, Buffalo NY).
36
2-24
-------
2 Figure 2-16. Differences in seasonal mean Os between the NAB and NB scenarios.
3 Eleven separate source categories were tracked in the source apportionment modeling,
4 including five boundary condition terms (East, South, West, North, and top) and six emissions
5 sectors within the domain. The contributions of each of these terms is provided in the Appendix
6 and summarized below. At most locations, the five model boundary terms contributed an
7 aggregate 40-60 percent of the total seasonal mean MDA8 O^ across the U.S. The highest
8 proportional impacts from the boundary conditions are along the coastlines and the
9 intermountain West. The Os entering the model domain via the boundary conditions can have a
10 variety of origins including: a) natural sources of 63 and precursors emanating from outside the
11 domain, b) anthropogenic sources of 63 precursors emanating from outside the domain, and c)
12 some fraction of U.S. emissions (natural and anthropogenic) which exit the model domain but
13 get re-imported into the domain via synoptic-scale recirculation. Accordingly, it is not possible
14 to relate the boundary condition contribution to any specific background element. The single
15 largest sector that was tracked in the source apportionment modeling was U.S. anthropogenic
16 emissions. Figure 2-17 shows the contributions from this sector to seasonal mean MDA8 Oi
17 levels. Over most of the U.S. this term contributes 40-60 percent to the total seasonal mean 03.
18 As discussed in section 2.4.3, these contributions are even higher when only high 63 days are
19 considered. International shipping emissions, as well as fires and other biogenic emissions also
20 contribute in a non-negligible way to background O^ over the U.S. The key finding from this
21 analysis is that air quality planning efforts to reduce anthropogenic methane emissions and
22 international NOx/nmVOC emissions (e.g., migrating from Asia, Canada, and Mexico; and from
2-25
-------
1 commercial shipping) have the potential to lower background levels and ease eventual attainment
2 oftheNAAQS.
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
O 10-20% (591
20-40% (441)
40-60% (791)
> 60% 101
Figure 2-17. Percent contribution of U.S. anthropogenic emissions to total seasonal mean
MDA8 Os levels, based on 2007 source apportionment modeling.
2.4.7 Summary
For a variety of reasons, it is challenging to present a comprehensive summary of all the
components and implications of background 63. In many forums the term "background" is used
generically and the lack of specificity can lead to confusion as to what sources are being
considered. Additionally, it is well established that the impacts of background sources can vary
greatly over space and time which makes it difficult to present a simple summary of background
Os levels. Further, background Os can be generated by a variety of processes, each of which can
lead to differential patterns in space and time, and which often have different regulatory
ramifications. Finally, background 63 is difficult to measure and thus, typically requires air
quality modeling which has inherent uncertainties and potential errors and biases.
That said, EPA believes the following concise and four-stage summary of the
implications of background 63 on the NAAQS review is appropriate, as based on previous
modeling exercises and the more recent EPA analyses summarized herein. First, background O^
exists and can comprise a considerable fraction of total seasonal mean 63 and W126 across the
2-26
-------
1 U.S. Air quality models can estimate the fractional contribution of backgroud sources to total O?,
2 in an individual area. The largest absolute values of background (NB, NAB, USB, or
3 apportionment-based USB) are modeled to occur at locations in the intermountain western U.S.
4 and are maximized in the spring and early summer seasons. Second, the modeling indicates that
5 U.S. anthropogenic emission sources are the dominant contributor to the majority of modeled Os
6 exceedances of the NAAQS. Higher O^ days generally have smaller fractional contributions
7 from background. This finding indicates that the relative importance of background 63 would
8 increase were Os concentrations to decrease with a lower level of the Os NAAQS. Third, while
9 the majority of modeled 63 exceedances have local and regional emissions as their primary
10 cause, there can be events where 63 levels approach or exceed 60-75 ppb due to background
11 sources. These events are relatively infrequent and EPA has policies that could allow for the
12 exclusion of air quality monitoring data affected by these types of events from design value
13 calculations. Fourth and finally, the Clean Air Act requires the NAAQS to be set at a level
14 requisite to protect public health and welfare. Proximity to background levels could be an
15 additional consideration, but only where it would support a decision based on the health evidence
16 and analyses.
17 2.5 REFERENCES
18 Carlton, A.G.; Finder, R.W.; Bhave, P.V.; Pouliot, G.A. (2010). To What Extent Can biogenic SOA be Controlled?
19 Environmental Science & Technology, 44, 3376-3380.
20 Dunker, A.M.; Yarwood, G; Ortmann, J.P.; Wilson, G.M. (2002). Comparison of source apportionment and source
21 sensitivity of ozone in a three-dimensional air quality model. Environmental Science & Technology 36:
22 2953-2964.
23 Emery, C; Jung, J; Downey, N; Johnson, J; Jimenez, M; Yarwood, G; Morris, R. (2012). Regional and global
24 modeling estimates of policy relevant background ozone over the United States. Atmos Environ 47: 206-
25 217. http://dx.doi.0rg/10.1016/j.atmosenv.2011.ll.012.
26 Lefohn, A. S.; Laurence, J. A.; Kohut, R. J. (1988). A comparison of indices that describe the relationship between
27 exposure to ozone and reduction in the yield of agricultural crops. Atmos. Environ. 22: 1229-1240.
28 Henderson, B.H., Possiel, N., Akhtar, F., Simon, H.A. (2012). Regional and Seasonal Analysis of North American
29 Background Ozone Estimates from Two Studies. Available on the Internet at:
3 0 http://www.epa. gov/ttn/naaqs/standards/ozone/s_o3_td.html
31 Kemball-Cook, S.; Parrish, D.; Ryerson, T; Nopmongcol, U.; Johnson, J.; Tai, E.; Yarwood, G. (2009).
32 Contributions of regional transport and local sources to ozone exceedances in Houston and Dallas:
3 3 Comparison of results from a photochemical grid model to aircraft and surface measurements. Journal of
34 Geophysical Research-Atmospheres, 114: DOOF02. DOI: 10.1029/2008JD010248.
35 U.S. Environmental Protection Agency (2005). Technical Support Document for the Final Clean Air Interstate Rule
3 6 Air Quality Modeling. Office of Air Quality Planning and Standards, Research Triangle Park, NC, 285pp.
3 7 http://www.epa.gov/cair/technical.html.
2-27
-------
1 U.S. Environmental Protection Agency. (2007). Review of the National Ambient Air Quality Standards for Ozone:
2 Policy Assessment of Scientific and Technical Information - OAQPS Staff Paper, U.S. Environmental
3 Protection Agency, Research Triangle Park, NC. EPA 452/R-07-007.
4 U.S. Environmental Protection Agency (2011). Air Quality Modeling Final Rule Technical Support Document.
5 Office of Air Quality Planning and Standards, Research Triangle Park, NC, 363pp.
6 http://www.epa.gov/airtransport/CSAPR/techinfo.html.
7 U.S. Environmental Protection Agency. (2012). Policy Assessment for the Review of the Ozone National Ambient
8 Air Quality Standards, First External Review Draft, U.S. Environmental Protection Agency, Research
9 Triangle Park, NC. EPA 452/P-12-002.
10 U.S. Environmental Protection Agency. (2013). Integrated Science Assessment for Ozone and Related
11 Photochemical Oxidants, U.S. Environmental Protection Agency, Research Triangle Park, NC.
12 EPA/600/R-10/076.
13 U.S. Environmental Protection Agency. (2014). Health Risk and Exposure Assessment for Ozone, Second External
14 Review Draft, U.S. Environmental Protection Agency, Research Triangle Park, NC. EPA-452/P-14-004a.
15 Wang, HQ; Jacob, DJ; Le Sager, P; Streets, DG; Park, RJ; Gilliland, AB; van Donkelaar, A. (2009). Surface ozone
16 background in the United States: Canadian and Mexican pollution influences. Atmos Environ 43: 1310-
17 1319. http://dx.doi.0rg/10.1016/j.atmosenv.2008.ll.036.
18 Zhang, L; Jacob, DJ; Downey, NV; Wood, DA; Blewitt, D; Carouge, CC; Van donkelaar, A; Jones, DBA; Murray,
19 LT; Wang, Y. (2011). Improved estimate of the policy-relevant background ozone in the United States
20 using the GEOS-Chem global model with 1/2° x 2/3° horizontal resolution over North America. Atmos
21 Environ45: 6769-6776. http://dx.doi.0rg/10.1016/j.atmosenv.2011.07.054.
2-28
-------
1 3 Adequacy of the Current Primary Standard
2 This chapter presents staffs considerations and conclusions regarding the adequacy of
3 the current primary 63 NAAQS. In doing so, we pose the following overarching question:
Does the currently available scientific evidence and exposure/risk information, as
5 reflected in the ISA and HREA, support or call into question the adequacy of the
current Os standard?
6 I
7 In addressing this overarching question, we pose a series of more specific questions, as discussed
8 in sections 3.1 through 3.4 below. Section 3.1 presents our consideration of the available
9 scientific evidence (i.e., evidence-based considerations) about the health effects associated with
10 short- and long-term 63 exposures. Section 3.2 presents our consideration of available estimates
11 of 63 exposures and health risks (exposure- and risk-based considerations). Section 3.3 discusses
12 the advice and recommendations that we have received from the CAS AC on the first draft 03
13 PA, and on documents from previous reviews of the OT, NAAQS. Section 3.4 revisits the
14 overarching question of this section, and presents staffs conclusions regarding the adequacy of
15 the current primary 63 NAAQS.
16 3.1 EVIDENCE-BASED CONSIDERATIONS
17 This section presents our consideration of the available scientific evidence with regard to
18 the adequacy of the current 63 standard. Our approach, as summarized in section 1.3.1 above, is
19 based on the full body of evidence in this review. We use information from the full evidence
20 base to characterize our confidence in the extent to which Os-attributable effects occur, and the
21 extent to which such effects are adverse, over the ranges of 03 exposure concentrations evaluated
22 in controlled human exposure studies and over the distributions of ambient O^ concentrations in
23 locations where epidemiologic studies have been conducted. In doing so, we recognize that the
24 available health effects evidence reflects a continuum from relatively high Oj, concentrations, at
25 which scientists generally agree that adverse health effects are likely to occur, through lower
26 concentrations, at which the likelihood and magnitude of a response become increasingly
27 uncertain.
3-1
-------
1 Section 3.1.1 summarizes a mode of action framework for understanding the effects of
2 both short- and long-term 63 exposures, based on Chapter 5 of the ISA (U.S. EPA, 2013).
3 Section 3.1.2 presents our consideration of the evidence for health effects attributable to short-
4 term and long-term 63 exposures. Section 3.1.3 discusses the adversity of the effects. Section
5 3.1.4 presents our consideration of evidence with regard to concentrations associated with health
6 effects and section 3.1.5 presents our consideration of the public health implications of exposures
7 to 03, including the adversity of effects and evidence for at-risk populations and lifestages.1
8 3.1.1 Modes of Action
9 Our consideration of the evidence of effects attributable to short-and long-term exposures
10 and the factors that increase risk in populations and lifestages builds upon evidence about the
11 modes of action by which 63 exerts effects (U.S. EPA, 2013; section 5.3). Mode of action refers
12 to a sequence of key events and processes that result in a given toxic effect; elucidation of
13 mechanisms provides a more detailed understanding of these key events and processes. The
14 purpose of this section is to describe the key events and pathways that contribute to health effects
15 resulting from both short-term and long-term exposures to 03. The extensive research carried out
16 over several decades in humans and animals has yielded numerous studies on mechanisms by
17 which 63 exerts its effects. It is well-understood that secondary oxidation products, which form
18 as a result of 63 exposure, initiate numerous responses at the cellular, tissue and whole organ
19 level of the respiratory system. These responses include the activation of neural reflexes,
20 initiation of inflammation, alteration of barrier epithelial function, sensitization of bronchial
21 smooth muscle, modification of lung host defenses, and airways remodeling, as discussed below.
22 These key events have the potential to affect other organ systems such as the cardiovascular
23 system. It has been proposed that secondary oxidation products, which are bioactive and
24 cytotoxic in the respiratory system, are also responsible for systemic effects. Recent studies in
25 animal models show that inhalation of 63 results in systematic oxidative stress.
26 Figure 3.1 below, adapted from Figure 5-8 of the ISA (ISA, Section 5.3.10, U.S. EPA,
27 2013), shows key events in the toxicity pathway of O^ that are described in more detail in
28 Appendix 3-A. The initial key event in the toxicity pathway of Os is the formation of secondary
lrThe term "at-risk populations" includes the factor lifestages, specifically childhood and older adulthood, that are
experienced by most people over the course of a lifetime, unlike other factors associated with at-risk populations.
3-2
-------
1
2
3
4
5
6
7
10
oxidation products in the respiratory tract (ISA, section 5-3, U.S. EPA, 2013). This mainly
involves direct reactions with components of the extracellular lining fluid (ELF). Although the
ELF has inherent capacity to quench (based on individual antioxidant capacity), this attenuative
capacity can be overwhelmed, especially with exposure to elevated concentrations of Oj2.
The resulting secondary oxidation products transmit signals to the epithelium, pain receptive
nerve fibers and, if present, immune cells (i.e., eosinophils, dendritic cells and mast cells)
involved in allergic responses. Thus, the effects of Os are mediated by components of ELF and
by the multiple cell types found in the respiratory tract. Further, oxidative stress3 is an implicit
part of this initial key event.
Mode of Action/Possible Pathways
Ozone + Respiratory Tract
i
Formation of secondary oxidation products
I
Activation
of neural
reflexes
Initiation of
inflammation
Sensitization
of bronchial
smooth muscle
Systemic inflammation and
oxidative/nitrosative stress
Extrapulmonary Effects
Decrements in pulmonary function
Pulmonary inflammation/oxidative stress
Increases in airways permeability
Airways hyperresponsiveness
Exacerbation/induction of asthma
Decreased host defenses
Epithelial metaplasia andfibrotic changes
Altered lung development
The ELF is a complex mixture of lipids (fats), proteins, and antioxidants that serve as the first barrier and target for
inhaled O3. The quenching ability of antioxidant substances present in the ELF appear in most cases to limit
interaction of O3 with underlying tissues and to prevent penetration of O3 deeper into the lung. However, as the ELF
thickness decreases and becomes ultra thin in the alveolar region, it may be possible for direct interaction of O3 with
the underlying epithelial cells to occur. The formation of secondary oxidation products is likely related to the
concentration of antioxidants present and the quenching ability of the lining fluid.
Oxidative stress reflects an imbalance between the systemic manifestation of reactive oxygen species, such as
superoxides, and a biological system's ability to readily detoxify the reactive intermediates or to repair the resulting
damage.
3-3
-------
1 Figure 3-1. Modes of action/possible pathways underlying the health effects resulting from
2 inhalation exposure to Os. (Adapted from U.S. EPA, 2013, Figure 5-8)
3 Figure 3-1 illustrates pathways identified in the ISA by which the array of key events
4 identified may lead to health effects associated with inhalation exposure to 03. For example, the
5 activation of neural reflexes, which may be triggered by secondary oxidative products, lead to
6 lung function decrements and may play a role in extrapulmonary effects such as slow resting
7 heart rate or bradycardia. Secondary oxidation products have also been implicated in the
8 initiation of inflammation and inflammation further contributes to (Vmediated oxidative stress.
9 Alteration of epithelial barrier function may play a role in allergic sensitization and in enhanced
10 sensitization of bronchial smooth muscle, resulting in airways hyperresponsiveness; genetic
11 susceptibility has been associated with this pathway. In addition to genetic factors, pre-existing
12 conditions and diseases, nutritional status, lifestage and co-exposures may affect multiple key
13 events in Figure 3-1 and contribute to altered risk of Os-induced effects (U.S. EPA, 2013, section
14 5.4). Evidence also indicates that exposure to Os modifies innate and adaptive immunity; such
15 effects can result in both short- and longer-term consequences related to the exacerbation and/or
16 induction of asthma and to alterations in host defense. Another event, airways remodeling, has
17 been demonstrated following chronic and/or intermittent exposure to O^ in animal models.
18 Additionally, there is evidence that 63 exposure results in systemic inflammation and vascular
19 oxidative/nitrosative stress, which may lead to downstream systemic responses (U.S. EPA, 2013,
20 section 2.4).
21 Experimental evidence for such Os-induced changes contributes to our understanding of
22 the biological plausibility of adverse Ch-related health effects, including a range of respiratory
23 effects as well as effects outside the respiratory system (e.g., cardiovascular effects) (U.S. EPA,
24 2013, Chapters 6 and 7). The range of respiratory effects that could be mediated by the
25 secondary oxidation products formed following reactions with 63 include decrements in
26 pulmonary function; pulmonary inflammation and injury; increased airway permeability; airway
27 hyperresponsiveness; decreased lung host defense, exacerbation and/or induction of asthma; and
28 alterations in pulmonary structure and/or development (Figure 3-1, above). These effects are
29 logically linked to the types of adverse Os-attributable effects evaluated and observed in
30 epidemiologic studies, including respiratory symptoms, respiratory hospital admissions and
31 emergency department visits, and premature mortality (U.S. EPA, 2013, Chapters 6 and 7).
32 Moreover, it has been proposed that some of these key events, including Os-mediated systemic
33 oxidative stress and activation of neural reflexes, are linked to the extrapulmonary effects of 63
34 that have been noted for decades (U.S. EPA, 2013, section 5.3.2 and 5.3.8). Further,
35 interindividual variability in the various key events (e.g., due to genetic variants or diet affecting
3-4
-------
1 antioxidant defenses) illustrated in Figure 3-1 may contribute to differences in susceptibility to
2 O3 health effects (ISA, section 5.4).
3 3.1.2 Nature of Effects
4 • To what extent does the currently available scientific evidence alter or strengthen
5 our conclusions from the last review regarding health effects attributable to Os
6 exposure in ambient air? Are previously identified uncertainties reduced or do
7 important uncertainties remain?
8 The health effects of ozone are described in detail in the assessment of the evidence
9 available in this review which is largely consistent with conclusions of past CDs. In some
10 categories of health effects, there is newly available evidence regarding some aspects of the
11 effects described in the last review or that strengthens our conclusions regarding aspects of 63
12 toxicity on a particular physiological system (U.S. EPA, 2013, Table 1-1). A sizeable number of
13 studies on Os health effects are newly available in this review and are critically assessed in the
14 ISA as part of the full body of evidence. Based on this assessment, the ISA determined that a
15 causal relationship4 exists between short-term exposure to Oj, in ambient air5 and effects on the
16 respiratory system and that a likely to be causal relationship6 exists between long-term exposure
17 to O3 in ambient air and respiratory effects (U.S. EPA 2013, pp. 1-6 to 1-7). As stated in the
18 ISA, "[c]ollectively, a very large amount of evidence spanning several decades supports a
19 relationship between exposure to 63 and a broad range of respiratory effects" (ISA, p. 1-6).
20 Additionally, the ISA determined likely to be causal relationships exist between short-term
21 exposures to 63 in ambient air and both total mortality and cardiovascular effects, based on
22 expanded evidence bases in the current review (U.S. EPA, 2013, pp. 1-7 to 1-8). In the ISA, EPA
23 additionally determined that the currently available evidence for additional endpoints is
24 suggestive of causal relationships between short-term (central nervous system effects) and long-
25 term exposure (cardiovascular effects, central nervous system effects and total mortality) to
4 Since the last O3 NAAQS review, the IS As which have replaced CDs in documenting each review of the scientific
evidence (or air quality criteria) employ a systematic framework for weighing the evidence and describing
associated conclusions with regard to causality, using established descriptors, as summarized in section 1.3.1 above
(U.S. EPA, 2013, Preamble).
5 In determining that a causal relationship exists for O3 with specific health effects, EPA has concluded that
"[e]vidence is sufficient to conclude that there is a causal relationship with relevant pollutant exposures" (ISA, p.
Ixiv).
6 In determining a likely to be a causal relationship exists for O3 with specific health effects, EPA has concluded that
"[e]vidence is sufficient to conclude that a causal relationship is likely to exist with relevant pollutant exposures, but
important uncertainties remain" (ISA, p. Ixiv).
3-5
-------
1 ambient Os. Consistent with emphasis in past reviews on O?, health effects for which the evidence
2 is strongest, we place the greatest emphasis on studies of health effects that have been judged in
3 the ISA to be caused by, or likely to be caused by, 63 exposures (U.S. EPA, 2013, section 2.5.2).
4 This section presents our consideration of the evidence for health effects attributable to
5 63 exposures, including respiratory morbidity and mortality effects attributable to short- and
6 long-term exposures, and cardiovascular system effects (including mortality) and total mortality
7 attributable to short-term exposures. We focus particularly on considering the extent to which the
8 scientific evidence available in the current review has been strengthened since the last review,
9 and the extent to which important uncertainties and limitations in the evidence from the last
10 review have been addressed. In section 3.1.2.2, we then consider the extent to which the
11 available evidence indicates health effects may be attributable to ambient 03 concentrations
12 likely to be allowed by the current 03 NAAQS. In this section, we address the following specific
13 question for each category of health effects considering the evidence available in the 2008
14 review of the standard as well as evidence that has become available since then. The ISA
15 summarizes the longstanding body of evidence for 63 respiratory effects as follows (U.S. EPA,
16 2013, p. 1-5).
17 The clearest evidence for health effects associated with exposure to 0$ is provided
18 by studies of respiratory effects. Collectively, a very large amount of evidence
19 spanning several decades supports a relationship between exposure to O3 and a
20 broad range of respiratory effects (see Section 6.2.9 and Section 7.2.8). The
21 majority of this evidence is derived from studies investigating short-term
22 exposures (i.e., hours to weeks) to Os, although animal toxicologicalstudies and
23 recent epidemiologic evidence demonstrate that long-term exposure (i.e., months
24 to years) may also harm the respiratory system.
25 The extensive body of evidence supporting a causal relationship between short-term Os
26 exposures and respiratory effects is discussed in detail in Chapter 6 of the ISA (U.S. EPA, 2013),
27 while evidence for respiratory effects associated with long-term or repeated 03 exposures are
28 discussed in chapter 7 of that document (U.S., EPA, 2013).
29 3.1.2.1 Respiratory Effects - Short-term Exposures
30 • To what extent does the currently available scientific evidence, including related
31 uncertainties, strengthen or alter our understanding from the last review of respiratory
32 effects attributable to short-term Os exposures?
3-6
-------
1 The 2006 Os AQCD concluded that there was clear, consistent evidence of a causal
2 relationship between short-term Oj, exposure and respiratory effects (U.S. EPA, 2006b). This
3 conclusion was substantiated by evidence from controlled human exposure and toxicological
4 studies indicating a range of respiratory effects in response to short-term 63 exposures, including
5 pulmonary function decrements and increases in respiratory symptoms, lung inflammation, lung
6 permeability, and airway hyperresponsiveness. Toxicological studies provided additional
7 evidence for Os-induced impairment of host defenses. Combined, these findings from
8 experimental studies provided support for epidemiologic evidence, in which short-term increases
9 in ambient 63 concentration were consistently associated with decreases in lung function in
10 populations with increased outdoor exposures, especially children with asthma and healthy
11 children; increases in respiratory symptoms and asthma medication use in children with asthma;
12 and increases in respiratory-related hospital admissions and asthma-related ED visits (U.S. EPA,
13 2013, pp. 6-1 to 6-2).
14 As discussed in detail in the ISA (U.S. EPA, 2013, section 6.2.9), studies evaluated since
15 the completion of the 2006 63 AQCD support and expand upon the strong body of evidence that,
16 in the last review, indicated a causal relationship between short-term 63 exposures and
17 respiratory health effects. Recent controlled human exposure studies conducted in young, healthy
18 adults with moderate exertion have reported FEVi decrements and pulmonary inflammation
19 following prolonged exposures to O^ concentrations as low as 60 ppb, and respiratory symptoms
20 following exposures to concentrations as low as 70 ppb. Epidemiologic studies provide evidence
21 that increases in ambient 63 exposures can result in lung function decrements, increases in
22 respiratory symptoms, and pulmonary inflammation in children with asthma; increases in
23 respiratory-related hospital admissions and emergency department visits; and increases in
24 respiratory mortality. Some of these studies report such associations even for 63 concentrations
25 at the low end of the distribution of daily concentrations. Recent epidemiologic studies report
26 that associations with respiratory morbidity and mortality are stronger during the warm/summer
27 months and remain robust after adjustment for copollutants. Recent toxicological studies
28 reporting Os-induced inflammation, airway hyperresponsiveness, and impaired lung host defense
29 continue to support the biological plausibility and modes of action for the (Vinduced respiratory
30 effects observed in the controlled human exposure and epidemiologic studies. Further support is
31 provided by recent studies that found (^-associated increases in indicators of airway
3-7
-------
1 inflammation and oxidative stress in children with asthma (U.S. EPA, 2013, section 6.2.9).
2 Together, epidemiologic and experimental studies support a continuum of respiratory effects
3 associated with 63 exposure that can result in respiratory-related emergency department visits,
4 hospital admissions, and/or mortality (U.S. EPA, 2013, section 6.2.9).
5 Across respiratory endpoints, evidence indicates antioxidant capacity may modify the
6 risk of respiratory morbidity associated with 63 exposure (U.S. EPA, 2013, section 6.2.9, p. 6-
7 161). The potentially elevated risk of populations with diminished antioxidant capacity and the
8 reduced risk of populations with sufficient antioxidant capacity identified in epidemiologic
9 studies are strongly supported by similar findings from controlled human exposure studies and
10 by evidence that characterizes Os-induced decreases in intracellular antioxidant levels as a mode
11 of action for downstream effects.
12 We describe key aspects of this evidence below with regard to lung function decrements;
13 pulmonary inflammation, injury, and oxidative stress; airway hyperresponsiveness; respiratory
14 symptoms and medication use; lung host defense; allergic and asthma-related responses; hospital
15 admissions and emergency department visits; and respiratory mortality.
16 Lung Function Decrements
17 In the 2008 review, a large number of controlled human exposure studies7 reported Os-
18 induced lung function decrements in young, healthy adults engaged in intermittent, moderate
19 exertion following 6.6 hour exposures to O^ concentrations at or above 80 ppb. Although two
20 studies also reported effects following exposures to lower concentrations, an important
21 uncertainty in the last review was the extent to which exposures to Os concentrations below 80
22 ppb result in lung function decrements. In addition, in the last review epidemiologic panel
23 studies had reported (Vassociated lung function decrements in a variety of different populations
24 (e.g., children, outdoor workers) likely to experience increased exposures. In the current review,
25 additional controlled human exposure studies are available that have evaluated exposures to Os
26 concentrations of 60 or 70 ppb. The available evidence from controlled human exposure and
27 panel studies is assessed in detail in the ISA (U.S. EPA, section 6.2.1) and is summarized below.
7 The controlled human exposure studies discussed in this PA utilize only healthy adult subjects. In the absence of
controlled human exposure data for children, HREA estimates of lung function decrements are based on the
assumption that children exhibit the same lung function responses following O3 exposures as healthy 18 year olds
(U.S. EPA, 2014, section 6.2.4 and 6.5). Thus, the conclusions about the occurrence of lung function decrements
that follow generally apply to children as well as adults.
3-8
-------
1 Controlled exposures to O^ concentrations that can be found in the ambient air can result
2 in a number of lung function effects, including decreased inspiratory capacity; mild
3 bronchoconstriction; and rapid, shallow breathing patterns during exercise. Reflex inhibition of
4 inspiration results in a decrease in forced vital capacity (FVC) and total lung capacity (TLC) and,
5 in combination with mild bronchoconstriction, contributes to a decrease in the forced expiratory
6 volume in 1 second (FEVi) (U.S. EPA, 2013, section 6.2.1.1). Accumulating evidence indicates
7 that such effects are mediated by activation of sensory nerves, resulting in the involuntary
8 truncation of inspiration and a mild increase in airway obstruction due to bronchoconstriction
9 (U.S. EPA, 2013, section 5.3.10).
10 Data from controlled human exposure studies indicate that increasing the duration of Os
11 exposures and increasing ventilation rates decreases the O?, exposure concentrations required to
12 impair lung function. Ozone exposure concentrations well above those typically found in
13 ambient air are required to impair lung function in healthy resting adults, while exposure to 63
14 concentrations at or below those in the ambient air have been reported to impair lung function in
15 healthy adults exposed for longer durations while undergoing intermittent, moderate exertion
16 (U.S. EPA, 2012a, section 6.2.1.1). With repeated 63 exposures over several days, FEVi
17 responses become attenuated in both healthy adults and adults with mild asthma, though this
18 attenuation of response is lost after about a week without exposure (U.S. EPA, 2013, section
19 6.2.1.1; page 6-27).
20 When considering controlled human exposures studies of (^-induced lung function
21 decrements we evaluate both group mean changes in lung function and the interindividual
22 variability in the magnitude of responses. To the extent studies report statistically significant
23 decrements in mean lung function following O^ exposures after controlling for other factors, we
24 have more confidence that measured decrements are due to the 03 exposure itself, rather than
25 chance alone. As discussed below, group mean changes in lung function are often small,
26 especially following exposures to relatively low 63 concentrations (e.g., 60 ppb). However, even
27 when group mean decrements in lung function are small, some individuals could experience
28 decrements that are "clinically meaningful" (Pellegrino et al., 2005; ATS, 1991; EPA, 2010)
29 with respect to criteria for spirometric testing, and/or that could be considered "adverse" with
30 respect to public health policy decisions. See section 3.1.3 below.
3-9
-------
1 At the time of the last review, a number of controlled human exposure studies had
2 reported lung function decrements in young, healthy adults following prolonged (6.6-hour)
3 exposures while at moderate exertion to O^ concentrations at and above 80 ppb. In addition,
4 there were two controlled human exposure studies by Adams (2002, 2006) that examined lung
5 function effects following exposures to Os concentrations of 60 ppb. The EPA's analysis of the
6 data from the Adams (2006) study reported a small but statistically significant Os-induced
7 decrement in group mean FEVi following exposures of young, healthy adults, while at moderate
o
8 exertion, to 60 ppb Os, when compared with filtered air controls (Brown, 2007). Further
9 examination of the post-exposure FEVi data, and mean data for other time points and other
10 concentrations, indicated that the temporal pattern of the response to 60 ppb 63 was generally
11 consistent with the temporal patterns of responses to higher O^ concentrations in this and other
12 studies. (75 FR 2950, January 19, 2010). This suggested a pattern of response following
13 exposures to 60 ppb Os that was consistent with a dose-response relationship, rather than random
14 variability. See also State of Mississippi v. EPAJ21 F. 3d at 259 (upholding EPA's
15 interpretation of the Adams studies).
16 Figure 6-1 in the ISA summarizes the currently available evidence from multiple
17 controlled human exposure studies evaluating group mean changes in FEVi following prolonged
18 63 exposures (i.e., 6.6 hours) in young, healthy adults engaged in moderate levels of physical
19 activity (U.S. EPA, 2012, section 6.2.1.1). With regard to the group mean changes reported in
20 these studies, the ISA specifically notes the following (U.S. EPA, 2012a, section 6.2.1.1, Figure
21 6-1):
22 1. Prolonged exposure to 40 ppb 63 results in a small decrease in group mean FEVi that is
23 not statistically different from responses following exposure to filtered air (Adams, 2002;
24 Adams, 2006).
25 2. Prolonged exposure to an average Os concentration of 60 ppb results in group mean FEVi
26 decrements ranging from 1.8% to 3.6% (Adams 2002; Adams, 2006;9 Schelegle et al.,
27 2009; Kim et al., 2011). Based on data from multiple studies, the weighted average group
28 mean decrement was 2.7%. In some analyses, these group mean decrements in lung
Adams (2006a) did not find effects on FEVI at 60 ppb to be statistically significant. In an analysis of the Adams
(2006a) data, even after removal of potential outliers, Brown et al. (2008) found the average effect on FEVI at
60 ppb to be small, but highly statistically significant (p < 0.002) using several common statistical tests.
9 Adams (2006); (2002) both provide data for an additional group of 30 healthy subjects that were exposed via
facemask to 60 ppb (square-wave) O3 for 6.6 hours with moderate exercise ( ) = 23 L/min per m2 BSA). These
subj ects are described on page 13 3 of Adams (2006) and pages 747 and 761 of Adams (2002). The FEVi decrement
may be somewhat increased due to a target ( ) of 23 L/min per m2 BSA relative to other studies with which it is
listed having the target ( ) of 20 L/min per m2 BSA. The facemask exposure is not expected to affect the
responses relative to a chamber exposure.
3-10
-------
1 function were statistically significant (Brown et al., 2008; Kim et al., 2011), while in
2 other analyses they were not (Adams, 2006; Schelegle et al., 2009).10
3 3. Prolonged exposure to an average O^ concentration of 70 ppb results in a statistically
4 significant group mean decrement in FEVi of about 6% (Schelegle et al., 2009).
5 4. Prolonged square-wave exposure to average O?, concentrations of 80 ppb, 100 ppb, or 120
6 ppb Os results in statistically significant group mean decrements in FEVi ranging from 6
7 to 8%, 8 to 14%, and 13 to 16%, respectively (Folinsbee et al., 1988; Horstman et al.,
8 1990; McDonnell et al., 1991; Adams, 2002; Adams, 2003; Adams, 2006).
9 As illustrated in Figure 6-1 of the ISA, there is a smooth dose-response curve without
10 evidence of a threshold for exposures between 40 and 120 ppb 63 (U.S. EPA, 2012a, Figure 6-
11 1). When these data are taken together, the ISA concludes that "mean FEVi is clearly decreased
12 by 6.6-h exposures to 60 ppb 63 and higher concentrations in [healthy, young adult] subjects
13 performing moderate exercise" (U.S. EPA, 2012a, p. 6-9).
14 With respect to interindividual variability in lung function, in an individual with
15 relatively "normal" lung function, with recognition of the technical and biological variability in
16 measurements, within-day changes in FEVi of > 5% are clinically meaningful (Pellegrino et al.,
17 2005; ATS, 1991). The ISA (U.S. EPA, 2013, section 6.1.) focuses on individuals with >10%
18 decrements in FEVi for two reasons. A 10% FEVi decrement is accepted by the American
19 Thoracic Society (ATS) as an abnormal response and a reasonable criterion for assessing
20 exercise-induced bronchoconstriction (Dryden et al., 2010; ATS, 2000a). (U.S. EPA, 2013,
21 section 6.2.1.1). Also, some individuals in the Schelegle et al. (2009) study experienced 5-10%
22 FEVi decrements following exposure to filtered air.
23 In previous NAAQS reviews, the EPA has made judgments regarding the potential
24 implications for individuals experiencing FEVi decrements of varying degrees of severity.11 For
25 people with lung disease, the EPA judged that moderate functional decrements (e.g., FEVi
26 decrements > 10 percent but < 20 percent, lasting up to 24 hours) would likely interfere with
Adams (2006) did not find effects on FEVi at 60 ppb to be statistically significant. In an analysis of the Adams
(2006) data, Brown et al. (2008) addressed the more fundamental question of whether there were statistically
significant differences in responses before and after the 6.6 hour exposure period and found the average effect on
FEVi at 60 ppb to be small, but highly statistically significant using several common statistical tests, even after
removal of potential outliers.
Such judgments have been made for decrements in FEVi as well as for increased airway responsiveness and
symptomatic responses (e.g., cough, chest pain, wheeze). Ranges of pulmonary responses and their associated
potential impacts are presented in Tables 3-2 and 3-3 of the 2007 Staff Paper (U.S. EPA, 2007).
3-11
-------
1 normal activity for many individuals, and would likely result in more frequent use of medication
2 (75 FR 2973, January 19, 2010). In previous reviews CASAC has endorsed these conclusions. In
3 the context of standard setting, in the last review 63 review CASAC indicated that it is
4 appropriate to focus on the lower end of the range of moderate functional responses (e.g., FEVi
5 decrements > 10 percent) when estimating potentially adverse lung function decrements in
6 people with lung disease, especially children with asthma (Henderson, 2006c; transcript of
7 CASAC meeting, day 8/24/06, page 149). More specifically, CASAC stated that "[a] 10%
8 decrement in FEVi can lead to respiratory symptoms, especially in individuals with pre-existing
9 pulmonary or cardiac disease. For example, people with chronic obstructive pulmonary disease
10 have decreased ventilatory reserve (i.e., decreased baseline FEVi) such that a > 10% decrement
11 could lead to moderate to severe respiratory symptoms" (Samet, 2011). Therefore, in considering
12 interindividual variability in Os-induced lung function decrements in the current review, we also
13 focus on the extent to which individuals were reported to experience FEVi decrements of 10% or
14 greater.
15 New studies (Schelegle et al., 2009; Kim et al., 2011) add to the previously available
16 evidence for interindividual variability in the responses of healthy adults following exposures to
17 63. Following prolonged exposures to 80 ppb 63 while at moderate exertion, the proportion of
18 healthy adults experiencing FEVi decrements greater than 10% was 17% by Adams (2006a),
19 26% by McDonnell (1996), and 29% by Schelegle et al. (2009). Following exposures to 60 ppb
20 O3, that proportion was 20% by Adams (2002), 3% by Adams (2006a), 16% by Schelegle et al.
21 (2009), and 5% by Kim et al. (2011). Based on these studies, the weighted average proportion of
22 young, healthy adults with >10% FEVi decrements is 25% following exposure to 80 ppb O^ and
23 10% following exposure to 60 ppb O3 (U.S. EPA, 2013, page 6-19).12 The ISA notes that
24 responses within an individual tend to be reproducible over a period of several months,
25 indicating that interindividual differences reflect differences in intrinsic responsiveness. Given
26 this, the ISA concludes that "a considerable fraction" of healthy individuals experience clinically
12
The ISA also notes that by considering responses uncorrected for filtered air exposures, during which lung
function typically improves (which would increase the size of the change, pre-and post-exposure), 10% is an
underestimate of the proportion of healthy individuals that are likely to experience clinically meaningful changes in
lung function following exposure for 6.6 hours to 60 ppb O3 during intermittent moderate exertion (U.S. EPA, 2012,
section 6.2.1.1).
3-12
-------
1 meaningful decrements in lung function when exposed for 6.6 hours to 60 ppb O^ during quasi
2 continuous, moderate exertion (U.S. EPA, 2013, section 6.2.1.1, p. 6-20).
3 As discussed above (Figure 3-1) and in the ISA (U.S EPA, 2013, Section 5.3.2),
4 secondary oxidation products formed following Oj, exposures can activate neural reflexes leading
5 to decreased lung function. Two new quantitative models, discussed in section 6.2.1.1 of the ISA
6 (U.S. EPA, 2013, p. 6-15), make use of the concept of oxidant stress to estimate the occurrence
7 of lung function decrements following exposures to relatively low Oi concentrations (McDonnell
8 et al., 2012; Schelegle et al., 2012). These models reflect the protective effect of antioxidants in
9 the ELF at lower ambient 63 concentrations, and include a threshold related to an integrated dose
10 rate.
11 McDonnell et al. (2012) and Schelegle et al. (2012) developed models using data on 03
12 exposure concentrations, ventilation rates, duration of exposures, and lung function responses
13 from a number of controlled human exposure studies. The McDonnell et al. (2012) and Schelegle
14 et al. (2012) studies analyzed large datasets to fit compartmental models that included the
15 concept of a dose of onset in lung function response or a response threshold based upon the
16 inhaled Os dose. The first compartment in the McDonnell et al. (2012) model considers the level
17 of oxidant stress in response to 63 exposure to increase over time as a function of dose rate
18 (Cx ) and decrease by clearance or metabolism over time. In the second compartment of the
19 McDonnell model, once oxidant stress reaches a threshold level the decrement in FEVi increases
20 as a sigmoid-shaped function. In the Schelegle et al. (2012) model, a first compartment acts as a
21 reservoir in which oxidant stress builds up until the dose of onset, at which time it spills over into
22 a second compartment. The second compartment is identical to the first compartment in
23 McDonnell et al. (2012) model. The oxidant levels in the second compartment were multiplied
24 by a responsiveness coefficient to predict FEVi responses for the Schelegle et al. (2012) model.
25 The McDonnell et al. (2012) model was fit to a large dataset consisting of the FEVi
26 responses of 741 young, healthy adults (18-35 years of age) from 23 individual controlled
27 exposure studies. Concentrations across individual studies ranged from 40 ppb to 400 ppb,
28 activity level ranged from rest to heavy exercise, duration of exposure was from 2 to 7.6 hours.
29 The extension of the McDonnell et al. (2012) model to children and older adults is discussed in
30 section 6.2.4 of U.S. EPA (2014). Schelegle et al. (2012) also analyzed a large dataset with
31 substantial overlap to that used by McDonnell et al. (2012). The Schelegle et al. (2012) model
32 was fit to the FEVi responses of 220 young healthy adults (taken from a dataset of 704
33 individuals) from 21 individual controlled exposure studies. The resulting empirical models can
34 estimate the frequency distribution of individual responses for any exposure scenario as well as
3-13
-------
1 summary measures of the distribution such as the mean or median response and the proportions
2 of individuals with FEVi decrements > 10%, 15%, and 20%.
3 The predictions of the McDonnell and Schelegle models are consistent with the observed
4 results from the individual studies of Cb-induced FEVi decrements. Specifically, the model
5 developed by McDonnell et al. (2012) predicts that 9% of healthy exercising adults would
6 experience FEVi decrements greater than 10% following 6.6 hour exposure to 60 ppb Os, and
7 that 22% would experience such decrements following exposure to 80 ppb 63 (U.S. EPA, 2013,
8 p. 6-18 and Figure 6-3). The model developed by Schelegle et al. (2012) predicts that, for a
9 prolonged (6.6 hours) 63 exposure with moderate, quasi continuous exercise, the average dose of
10 onset for FEVi decrement would be reached following 4 to 5 hours of exposure to 60 ppb, and
11 following 3 to 4 hours of exposure to 80 ppb. However, 14% of the individuals had a dose of
12 onset that was less than 40% of the average. Those individuals would reach their dose of onset
13 following 1 to 2 hours of exposure to 50 to 80 ppb O3 (U.S. EPA, 2013, p. 6-16), which is
14 consistent with the threshold FEVi responses reported by McDonnell et al. (2012).
,13
15 Epidemiologic studies have consistently linked short-term increases in ambient 63
3
16 concentrations with lung function decrements in diverse populations and lifestages, including
17 children attending summer camps, adults exercising or working outdoors, and groups with pre-
18 existing respiratory diseases such as asthmatic children (U.S. EPA, 2013, section 6.2.1.2). Some
19 of these studies reported ozone-associated lung function decrements accompanied by respiratory
20 symptoms14 in asthmatic children (Just et al., 2002; Mortimer et al., 2002; Ross et al., 2002;
21 Gielen et al., 1997; Romieu et al., 1997; Thurston et al., 1997; Romieu et al., 1996). In contrast,
22 studies of children in the general population have reported similar (^-associated lung function
23 decrements but without accompanying respiratory symptoms (Ward et al., 2002; Gold et al.,
24 1999; Linn et al., 1996) (U.S. EPA, 2013, section 6.2.1.2).
25 Several panel studies reported that associations with lung function decrements persisted
26 at relatively low ambient 63 concentrations. For outdoor recreation or exercise, associations were
27 reported in analyses restricted to 1-hour average 03 concentrations less than 80 ppb (Spektor et
28 al., 1988a; Spektor et al., 1988b), 60 ppb (Brunekreef et al., 1994; Spektor et al., 1988a), and
29 50 ppb (Brunekreef et al., 1994). Among outdoor workers, Brauer et al. (1996) found a robust
30 association using daily 1-hour max O^ concentrations less than 40 ppb. Ulmer et al. (1997) found
31 a robust association in schoolchildren using 30-minute maximum 63 concentrations less than
32 60 ppb. For 8-hour average Os concentrations, associations with lung function decrements in
13
Unless otherwise specified, the epidemiologic studies discussed in this PA evaluate only adults.
Reversible loss of lung function in combination with the presence of symptoms meets the ATS definition of
adversity (ATS, 2000).
3-14
-------
1 children with asthma were found to persist at concentrations less than 80 ppb in a U.S. multicity
2 study (Mortimer et al., 2002) and less than 51 ppb in a study conducted in the Netherlands
3 (Gielenetal., 1997).
4 Studies investigating the effects of short-term exposure to 63 provided information on
5 potential confounding by copollutants such as PM2.5, PMio, NC>2, or SC>2. These studies varied in
6 how they evaluated confounding. Some studies of subjects exercising outdoors indicated that
7 ambient concentrations of copollutants such as NC>2, 862, or acid aerosol were low, and thus not
8 likely to confound associations observed for Os (Hoppe et al., 2003; Brunekreef et al., 1994;
9 Hoek et al., 1993). In other studies of children with increased outdoor exposures, 63 was
10 consistently associated with decreases in lung function, whereas other pollutants such as PM2.s,
11 sulfate, and acid aerosol individually showed variable associations across studies (Thurston et
12 al., 1997; Castillejos et al., 1995; Berry et al., 1991; Avol et al., 1990; Spektor et al., 1988a).
13 Studies that conducted copollutant modeling generally found Cb-associated lung function
14 decrements to be robust (i.e., most copollutant-adjusted effect estimates fell within the 95% CI of
15 the single-pollutant effect estimates) (U.S. EPA, 2013, Figure 6-10 and Table 6-14). Most O3
16 effect estimates for lung function were robust to adjustment for temperature, humidity, and
17 copollutants such as PM2.5, PMio, NO2, or 862. Although examined in only a few epidemiologic
18 studies, 63 also remained associated with decreases in lung function with adjustment for pollen
19 or acid aerosols (U.S. EPA, 2013, section 6.2.1.2).
20 Several epidemiologic studies demonstrated the protective effects of vitamin E and
21 vitamin C supplementation, and increased dietary antioxidant intake, on Os-induced lung
22 function decrements (Romieu et al., 2002) (U.S. EPA, 2013, Figure 6-7 and Table 6-8). These
23 results provide support for the new, quantitative models (McDonnell et al., 2012; Schelegle et
24 al., 2012), discussed above, which make use of the concept of oxidant stress to estimate the
25 occurrence of lung function decrements following exposures to relatively low 63 concentrations.
26 In conclusion, new information from controlled human exposure studies considerably
27 strengthens the evidence and reduces the uncertainties, relative to the evidence that was available
28 at the time of the 2008 review, regarding the presence and magnitude of lung function
29 decrements in healthy adults following prolonged exposures to Os concentrations below 80 ppb.
30 As discussed in Section 6.2.1.1 in the ISA (EPA, 2013, p. 6-12), there is information available
31 from four separate studies that evaluated exposures to 60 ppb 63 (Kim et al., 2011; Schelegle et
32 al., 2009; Adams 2002; 2006). Although not consistently statistically significant, group mean
33 FEVi decrements following exposures to 60 ppb 63 are consistent among studies. Moreover, as
34 is illustrated in Figure 6-1 of the ISA (U.S. EPA, 2013), the group mean FEVi responses at
35 60 ppb fall on a smooth intake dose-response curve for exposures between 40 and 120 ppb 03.
36 These studies also indicate that, on average, 10% of young, healthy adults experience clinically
3-15
-------
1 meaningful decrements in lung function when exposed for 6.6 hours to 60 ppb O^ during
2 intermittent, moderate exertion. One recent study has also reported statistically significant
3 decrements following exposures to 70 ppb Os (Schelegle et al., 2009). Predictions from newly
4 developed quantitative models, based on the concept that Cb-induced oxidation results in lung
5 function decrements, are consistent with these experimental results. Additionally, as discussed in
6 more detail in section 3.1.4 below, epidemiologic studies continue to provide evidence of lung
7 function decrements in people who are active outdoors, including people engaged in outdoor
8 recreation or exercise, children, and outdoor workers, at low ambient Os concentrations. While
9 few new epidemiologic studies of Cb-associated lung function decrements are available in this
10 review, previously available studies have reported associations with decrements, including at
11 relatively low ambient Os concentrations.
12 Pulmonary Inflammation, Injury, and Oxidative Stress
13 Ozone exposures result in increased respiratory tract inflammation and epithelial
14 permeability. Inflammation is a host response to injury, and the induction of inflammation is
15 evidence that injury has occurred. Oxidative stress has been shown to play a key role in initiating
16 and sustaining Os-induced inflammation. Secondary oxidation products formed as a result of
17 reactions between O3 and components of the ELF can increase the expression of molecules (i.e.,
18 cytokines, chemokines, and adhesion molecules) that can enhance airway epithelium
19 permeability (U.S. EPA, 2013, Sections 5.3.3 and 5.3.4). As discussed in detail in the ISA (U.S.
20 EPA, 2013, section 6.2.3), Os exposures can initiate an acute inflammatory response throughout
21 the respiratory tract that has been reported to persist for at least 18-24 hours after exposure.
22 Inflammation induced by exposure of humans to Os can have several potential outcomes:
23 (1) inflammation induced by a single exposure (or several exposures over the course of a
24 summer) can resolve entirely; (2) continued acute inflammation can evolve into a chronic
25 inflammatory state; (3) continued inflammation can alter the structure and function of other
26 pulmonary tissue, leading to diseases such as asthma; (4) inflammation can alter the body's host
27 defense response to inhaled microorganisms, particularly in potentially at-risk populations or
28 lifestages such as the very young and old; and (5) inflammation can alter the lung's response to
29 other agents such as allergens or toxins (U.S. EPA, 2013, Section 6.2.3). Thus, lung injury and
30 the resulting inflammation provide a mechanism by which O?, may cause other more serious
31 morbidity effects (e.g., asthma exacerbations).
3-16
-------
1 In the last review, controlled human exposure studies reported Os-induced airway
2 inflammation following exposures at or above 80 ppb. In the current review, the link between Os
3 exposures and airway inflammation and injury has been evaluated in additional controlled human
4 exposure studies, as well as in recent epidemiologic studies. Controlled human exposure studies
5 have generally been conducted in young, healthy adults or in adults with asthma using lavage
6 (proximal airway and bronchoalveolar), bronchial biopsy, and more recently, induced sputum.
7 These studies have evaluated one or more indicators of inflammation, including neutrophil15
8 (PMN) influx, markers of eosinophilic inflammation, increased permeability of the respiratory
9 epithelium, and/or prevalence of proinflammatory molecules (U.S. EPA, 2013, section 6.2.3.1).
10 Epidemiologic studies have generally evaluated associations between ambient 63 and markers of
11 inflammation and/or oxidative stress, which plays a key role in initiating and sustaining
12 inflammation (U.S. EPA, 2013, section 6.2.3.2).
13 There is an extensive body of evidence from controlled human exposure studies
14 indicating that short-term exposures to Os can cause pulmonary inflammation. Previously
15 available evidence indicated that Os causes an inflammatory response in the lungs (U.S. EPA,
16 1996a). A single acute exposure (1-4 hours) of humans to moderate concentrations of Os (200-
17 600 ppb) while exercising at moderate to heavy intensities resulted in a number of cellular and
18 biochemical changes in the lung, including inflammation characterized by increased numbers of
19 PMNs, increased permeability of the epithelial lining of the respiratory tract, cell damage, and
20 production of proinflammatory molecules (i.e., cytokines and prostaglandins, U.S. EPA, 2006b).
21 A meta-analysis of 21 controlled human exposure studies (Mudway and Kelly, 2004) using
22 varied experimental protocols (80-600 ppb Os exposures; 1-6.6 hours exposure duration; light to
23 heavy exercise; bronchoscopy at 0-24 hours post-Os exposure) reported that PMN influx in
24 healthy subjects is linearly associated with total O^ dose. Animal toxicological studies also
25 provided evidence for increases in inflammation and permeability in rabbits at levels as low as
26 100 ppb O3 (Section 2.5.3.1, ISA, U.S. EPA, 2013).
15 Referred to as either neutrophils or polymorphonuclear neutrophils (or PMNs), these are the most abundant type
of white blood cells in mammals. PMNs are recruited to the site of injury following trauma and are the hallmark of
acute inflammation. The presence of PMNs in the lung has long been accepted as a hallmark of inflammation and is
an important indicator that O3 causes inflammation in the lungs. Neutrophilic inflammation of tissues indicates
activation of the innate immune system and requires a complex series of events, that then are normally followed by
processes that clear the evidence of acute inflammation.
3-17
-------
1 Several studies, including one published since the last review (Alexis et al., 2010), have
2 reported (Vinduced increases in PMN influx and permeability following exposures at or above
3 80 ppb (Alexis et al., 2010; Peden et al., 1997; Devlin et al., 1991), and eosinophilic
4 inflammation following exposures at or above 160 ppb (Scannell et al., 1996; Peden et al., 1997;
5 Hiltermann et al., 1999; Vagaggini et al., 2002). In addition, one recent controlled human
6 exposure study has reported Cb-induced PMN influx following exposures of healthy adults to Os
7 concentrations of 60 ppb (Kim et al., 2011), the lowest concentration at which inflammatory
8 responses have been evaluated in human studies.
9 As with FEVi responses to 63, inflammatory responses to 63 are generally reproducible
10 within individuals, with some individuals experiencing more severe Cb-induced airway
11 inflammation than indicated by group averages (Holz et al., 2005; Holz et al., 1999). Unlike 63-
12 induced decrements in lung function, which are attenuated following repeated exposures over
13 several days (U.S. EPA, 2013, section 6.2.1.1), some markers of Cb-induced inflammation and
14 tissue damage remain elevated during repeated exposures, indicating ongoing damage to the
15 respiratory system (U.S. EPA, 2013, section 6.2.3.1).
16 Most controlled human exposure studies have reported that asthmatics experience larger
17 Os-induced inflammatory responses than non-asthmatics. Specifically, asthmatics exposed to
18 200 ppb Os for 4-6 hours with exercise show significantly more neutrophils in bronchoalveolar
19 lavage fluid (BALF) than similarly exposed healthy individuals (Scannell et al., 1996; Basha et
20 al., 1994). Bosson et al. (2003) reported significantly greater expression of a variety of pro-
21 inflammatory cytokines in asthmatics, compared to healthy subjects, following exposure to
22 200 ppb 63 for 2 hours. In addition, research available in the last review, combined with a recent
23 study newly available in this review, indicates that pretreatment of asthmatics with
24 corticosteroids can prevent the (Vinduced inflammatory response in induced sputum, though
25 pretreatment did not prevent FEVi decrements (Vagaggini et al., 2001; 2007). In contrast,
26 Stenfors et al. (2002) did not detect a difference in the Cb-induced increases in neutrophil
27 numbers between 15 subjects with mild asthma and 15 healthy subjects by bronchial wash at the
28 6 hours postexposure time point, although the neutrophil increase in the asthmatic group was on
29 top of an elevated baseline.
30 In people with allergic airway disease, including people with rhinitis and asthma,
31 evidence available in the last review indicated that proinflammatory mediators also cause
3-18
-------
1 accumulation of eosinophils in the airways (Bascom et al., 1990; Torres et al., 1996; Peden et
2 al.,1995 and 1997; Frampton et al., 1997; Michelson et al., 1999; Hiltermann et al., 1999; Holz et
3 al., 2002; Vagaggini et al., 2002). The eosinophil, which increases inflammation and allergic
4 responses, is the cell most frequently associated with exacerbations of asthma (75 CFR 2969,
5 January 19, 2010).
6 Studies reporting inflammatory responses and markers of lung injury have clearly
7 demonstrated that there is important variation in the responses of exposed subjects (75 FR 2953,
8 January 19, 2010). Some individuals also appear to be intrinsically more susceptible to increased
9 inflammatory responses (Holz et al., 2005). In healthy adults exposed to each 80 and 100 ppb Os,
10 Devlin et al. (1991) observed group average increases in neutrophilic inflammation of 2.1- and
11 3.8-fold, respectively. However, there was a 20-fold range in inflammatory responses between
12 individuals at both concentrations. Relative to an earlier, similar study conducted at 400 ppb
13 (Koren et al., 1989), Devlin et al. (1991) noted that although some of the study population
14 showed little or no increase in inflammatory and cellular injury indicators analyzed after
15 exposures to lower levels of 63 (i.e., 80 and 100 ppb), others had changes that were as large as
16 those seen when subjects were exposed to 400 ppb 63. The data suggest that as a whole the
17 healthy population, on average, may have small inflammatory responses to near-ambient levels
18 of Os, though there may be a significant subpopulation that is very sensitive to low levels of Os.
19 Devlin et al. (1991) expressed the view that "susceptible subpopulations such as the very young,
20 elderly, and people with pulmonary impairment or disease may be even more affected."
21 A number of studies report that OT, exposures increase epithelial cell permeability.
22 Increased BALF protein, suggesting Os-induced changes in epithelial permeability, has been
23 reported at 1 hour and 18 hours postexposure (Devlin et al., 1997; Balmes et al., 1996). A meta-
24 analysis of results from 21 publications (Mudway and Kelly, 2004a) for varied experimental
25 protocols (80-600 ppb 63; 1-6.6 hours duration; light to heavy exercise; bronchoscopy at 0-24
26 hours post-Cb exposure; healthy subjects), showed that increased BALF protein is associated
27 with total inhaled 63 dose (i.e., the product of 63 concentration, exposure duration, and ). As
28 noted in the 2009 PM ISA (U.S. EPA, 2009), it has been postulated that changes in permeability
29 associated with acute inflammation may provide increased access of inhaled antigens, particles,
30 and other inhaled substances deposited on lung surfaces to the smooth muscle, interstitial cells,
31 immune cells underlying the epithelium, and the blood (U.S. EPA, 2013, sections 5.3.4, 5.3.5).
3-19
-------
1 Animal toxicology studies have provided some support for this hypothesis (Adamson and
2 Prieditis, 1995; Chen et al 2006), though these studies did not specifically evaluate 63 exposures
3 (U.S. EPA, 2009). Because of this potentially increased access, it has been postulated that
4 increases in epithelial permeability following 63 exposure might lead to increases in airway
5 responsiveness to specific and nonspecific agents. In a recent study, Que et al. (2011)
6 investigated this hypothesis in healthy young adults (83M, 55 F) exposed to 220 ppb Os for 2.25
7 hours (alternating 15 min periods of rest and brisk treadmill walking). As has been observed for
8 FEVi responses, within-individual changes in permeability were correlated between sequential
9 63 exposures, indicating intrinsic differences among individuals in susceptibility to epithelial
10 damage following 63 exposures. However, increases in epithelial permeability at 1 day post-Cb
11 exposure were not correlated with with changes in airway responsiveness assessed 1 day post-Os
12 exposure. The authors concluded that changes in epithelial permeability is relatively constant
13 over time in young healthy adults, although changes in permeability and AHR appear to be
14 mediated by different physiologic pathways.
15 The limited epidemiologic evidence reviewed in the 2006 O3 AQCD (U.S. EPA, 2006)
16 demonstrated an association between short-term increases in ambient 63 concentrations and
17 airways inflammation in children (1-hour max OT, of approximately 100 ppb). In the 2006 Os
18 AQCD (U.S. EPA, 2006), there was limited evidence for increases in nasal lavage levels of
19 inflammatory cell counts and molecules released by inflammatory cells (i.e., eosinophilic
20 cationic protein, and myeloperoxidases). Since 2006, as a result of the development of less
21 invasive methods, there has been a large increase in the number of studies assessing ambient 63-
22 associated changes in airway inflammation and oxidative stress, the types of biological samples
23 collected (e.g., lower airway), and the types of indicators. Most of these recent studies have
24 evaluated biomarkers of inflammation or oxidative stress in exhaled breath, nasal lavage fluid, or
25 induced sputum (U.S. EPA, 2013, section 6.2.3.2). These recent studies form a larger database to
26 establish coherence with findings from controlled human exposure and animal studies that have
27 measured the same or related biological markers. Additionally, results from these studies provide
28 further biological plausibility for the associations observed between ambient 03 concentrations
29 and respiratory symptoms and asthma exacerbations.
30 A number of epidemiologic studies provide evidence that short-term increases in ambient
31 03 exposure increase pulmonary inflammation and oxidative stress in children, including those
3-20
-------
1 with asthma (Sienra-Monge et al., 2004; Barraza-Villarreal et al., 2008; Romieu et al., 2008;
2 Berhane et al., 2011). Multiple studies examined and found increases in exhaled NO (eNO)16
3 (Berhane et al., 2011; Khatri et al., 2009; Barraza-Villarreal et al., 2008). In some studies of
4 subjects with asthma, increases in ambient 63 concentration at the same lag were associated with
5 both increases in pulmonary inflammation and respiratory symptoms (Khatri et al., 2009;
6 Barraza-Villarreal et al., 2008). Although more limited in number, epidemiologic studies also
7 found associations with cytokines such as IL-6 or IL-8 (Barraza-Villarreal et al., 2008; Sienra-
8 Monge et al., 2004), eosinophils (Khatri et al., 2009), antioxidants (Sienra-Monge et al., 2004),
9 and indicators of oxidative stress (Romieu et al., 2008) (ISA, Section 6.2.3.2, U.S. EPA, 2013).
10 Because associations with inflammation were attenuated with higher antioxidant intake the study
11 by Sienra-Monge et al. (2004) provides additional evidence that inhaled 63 is likely to be an
12 important source of reactive oxygen species in airways and/or may increase pulmonary
13 inflammation via oxidative stress-mediated mechanisms among all age groups. Limitations in
14 some recent studies have contributed to inconsistent results in adults (U.S. EPA, 2013, section
15 6.2.3.2).
16 Exposure to ambient 63 on multiple days can result in larger increases in pulmonary
17 inflammation and oxidative stress, as discussed in section 6.2.3.2 of the ISA (U.S. EPA, 2013).
18 In studies that examined multiple Os lags, multiday averages of 8-hour maximum or
19 8-hour average concentrations were associated with larger increases in pulmonary inflammation
20 and oxidative stress (Berhane et al., 2011; Delfino et al., 2010a; Sienra-Monge et al., 2004),
21 consistent with controlled human exposure (U.S. EPA, 2013, section 6.2.3.1) and animal studies
22 (U.S. EPA, 2013, section 6.2.3.3) reporting that some markers of pulmonary inflammation
23 remain elevated with 63 exposures repeated over multiple days. Evidence from animal
24 toxicological studies also clearly indicates that 63 exposures result in damage and inflammation
25 in the lung (ISA, Section 5.3, U.S. EPA, 2013). In the few studies that evaluated the potential for
26 confounding, Os effect estimates were not confounded by temperature or humidity, and were
27 robust to adjustment for PM2.5 or PMio (Barraza-Villarreal et al., 2008; Romieu et al., 2008;
28 Sienra-Monge et al., 2004).
16 Exhaled NO has been shown to be a useful biomarker for airway inflammation in large population-based studies
(Linnetal., 2009) (ISA, U.S. EPA, 2013, Section 7.2.4).
3-21
-------
1 In conclusion, a relatively small number of controlled human exposure studies evaluating
2 Os-induced airway inflammation have become available since the last review. For purposes of
3 reviewing the current 63 NAAQS, the most important of these recent studies reported a
4 statistically significant increase in airway inflammation in healthy adults at moderate exertion
5 following exposures to 60 ppb Os, the lowest concentration that has been evaluated for
6 inflammation. In addition, a number of recent epidemiologic studies report Os-associated
7 increases in markers of pulmonary inflammation, particularly in children. Thus, recent studies
8 continue to support the evidence for airway inflammation and injury that was available in
9 previous reviews, with new evidence for such effects following exposures to lower
10 concentrations than had been evaluated previously.
11 Airway Hyperresponsiveness
12 Airway hyperresponsiveness (AHR) refers to a condition in which the conducting
13 airways undergo enhanced bronchoconstriction in response to a variety of stimuli. Airway
14 hyperresponsiveness is an important consequence of exposure to ambient 63 because its presence
15 reflects a change in airway smooth muscle reactivity, and indicates that the airways are
16 predisposed to narrowing upon inhalation of a variety of ambient stimuli including specific
17 triggers (i.e., allergens) and nonspecific triggers (e.g., SC>2, and cold air). People with asthma are
18 generally more sensitive to bronchoconstricting agents than those without asthma, and the use of
19 an airway challenge to inhaled bronchoconstricting agents is a diagnostic test in asthma.
20 Standards for airway responsiveness testing have been developed for the clinical laboratory
21 (ATS, 2000a), although variation in the methodology for administering the bronchoconstricting
22 agent may affect the results (Cockcroft et al., 2005). There is a wide range of airway
23 responsiveness in people without asthma, and responsiveness is influenced by a number of
24 factors, including cigarette smoke, pollutant exposures, respiratory infections, occupational
25 exposures, and respiratory irritants. Dietary antioxidants have been reported to attenuate Os-
26 induced bronchial hyperresponsiveness in people with asthma (Trenga et al., 2001).
27 Evidence for airway hyperresponsiveness following Os exposures is derived primarily
28 from controlled human exposure and toxicological studies (U.S. EPA, 2013, section 6.2.2).
29 Airway responsiveness is often quantified by measuring changes in pulmonary function
30 following the inhalation of an aerosolized allergen or a nonspecific bronchoconstricting agent
3-22
-------
1 (e.g., methacholine), or following exposure to a bronchoconstricting stimulus such as cold air. In
2 the last review, controlled human exposure studies of mostly adults (> 18 years of age) had
3 shown that exposures to 63 concentrations at or above 80 ppb increase airway responsiveness, as
4 indicated by a reduction in the concentration of specific (e.g., ragweed) and non-specific (e.g.,
5 methacholine) agents required to produce a given reduction in lung function (e.g., as measured
6 by FEVi or specific airway resistance) (U.S. EPA, 2013, section 6.2.2.1). This Os-induced AHR
7 has been reported to be dose-dependent (Horstman et al., 1990). Animal toxicology studies have
8 reported Os-induced airway hyperresponsiveness in a number of species, with some rat strains
9 exhibiting hyperresponsiveness following 4-hour exposures to Os concentrations as low as 50
10 ppb (Depuydt et al., 1999). Since the last review, there have been relatively few new controlled
11 human exposure and animal toxicology studies of 63 and airway hyperresponsiveness, and no
12 new studies have evaluated exposures to 63 concentrations at or below 80 ppb (U.S. EPA, 2013,
13 section 6.2.2.1)
14 Airway hyperresponsiveness is linked with the accumulation and/or activation of
15 eosinophils in the airways of asthmatics, which is followed by production of mucus and a late-
16 phase asthmatic response (75 FR 2970, January 19, 2010). In a study of 16 intermittent
17 asthmatics, Hiltermann et al. (1999) found that there was a significant inverse correlation
18 between the Os-induced change in the percentage of eosinophils in induced sputum and the
19 concentration of methacholine causing a 20% decrease in FEVi. Hiltermann et al. (1999)
20 concluded that the results point to the role of eosinophils in Os-induced airway
21 hyperresponsiveness. Increases in Os-induced nonspecific airway responsiveness incidence and
22 duration could have important clinical implications for children and adults with asthma, such as
23 exacerbations of their disease.
24 Airway hyperresponsiveness after O^ exposure appears to resolve more slowly than
25 changes in FEVi or respiratory symptoms (Folinsbee and Hazucha, 2000). Studies suggest that
26 Os-induced AHR usually resolves 18 to 24 hours after exposure, but may persist in some
27 individuals for longer periods (Folinsbee and Hazucha, 1989). Furthermore, in studies of
28 repeated exposure to 63, changes in AHR tend to be somewhat less susceptible to attenuation
29 with consecutive exposures than changes in FEVi (Gong et al., 1997a; Folinsbee et al., 1994;
30 Kulle et al., 1982; Dimeo et al., 1981) (U.S. EPA, 2013, section 6.2.2). In animal studies a 3-day
31 continuous exposure resulted in attenuation of Os-induced airway hyperresponsiveness (Johnston
3-23
-------
1 et al., 2005) while repeated exposures for 2 hours per day over 10 days did not (Chhabra et al.,
2 2010), suggesting that attenuation could be lost when repeated exposures are interspersed with
3 periods of rest (U.S. EPA, 2013, section 6.2.2.2).
4 Increases in airway responsiveness do not appear to be strongly associated with
5 decrements in lung function or increases in symptoms (Aris et al., 1995). Recently, Que et al.
6 (2011) assessed methacholine responsiveness in healthy young adults (83M, 55 F) one day after
7 exposure to 220 ppb 63 and filtered air for 2.25 hours (alternating 15 minute periods of rest and
8 brisk treadmill walking). Increases in airways responsiveness at 1 day post-Os exposure were not
9 correlated with FEVi responses immediately following the Os exposure or with changes in
10 epithelial permeability assessed 1-day post-Cb exposure. This indicates that airway hyper-
11 responsiveness also appears to be mediated by a differing physiologic pathway.
12 As mentioned above, in addition to human subjects a number of species, including
13 nonhuman primates, dogs, cats, rabbits, and rodents, have been used to examine the effect of Os
14 exposure on airway hyperresponsiveness (see Table 6-14, (U.S. EPA, 1996n) of the 1996 Os
15 AQCD and Annex Table AX5-12 on page AX5-36 (U.S. EPA, 2006h) of the 2006 O3 AQCD). A
16 body of animal toxicology studies, including some recent studies conducted since the last review,
17 provides support for the (Vinduced AHR reported in humans (U.S. EPA, 2013, section 6.2.2.2).
18 Although most of these studies evaluated 63 concentrations above those typically found in
19 ambient air in cities in the United States (i.e., most studies evaluated 63 concentrations of 100
20 ppb or greater), one study reported that a very low exposure concentration (50 ppb for 4 hours)
21 induced AHR in some rat strains (Depuydt et al., 1999). Additional recent rodent studies
22 reported Cb-induced AHR following exposures to Os concentrations from 100 to 500 ppb
23 (Johnston et al., 2005; Chhabra et al., 2010; Larsen et al., 2010). In characterizing the relevance
24 of these exposure concentrations, the ISA noted that a study using radiolabeled Os suggests that
25 even very high 63 exposure concentrations in rodents could be equivalent to much lower
26 exposure concentrations in humans. Specifically, a 2000 ppb (2 ppm) 63 exposure concentration
27 in resting rats was reported to be roughly equivalent to a 400 ppb exposure concentration in
28 exercising humans (Hatch et al., 1994). Given this relationship, the ISA noted that animal data
29 obtained in resting conditions could underestimate the risk of effects for humans (U.S. EPA,
30 2013, section 2.4, p. 2-14).
3-24
-------
1 The 2006 AQCD (U.S. EPA, 2006, p. 6-34) concluded that spirometric responses to O3
2 are independent of inflammatory responses and markers of epithelial injury (Balmes et al., 1996;
3 Blomberg et al., 1999; Hazucha et al., 1996; Torres et al., 1997). Significant inflammatory
4 responses to 63 exposures that did not elicit significant spirometric responses have been reported
5 (Holz et al., 2005; McBride et al., 1994). A recent study (Que et al., 2011) indicates that airway
6 hyper-responsiveness also appears to be mediated by a differing physiologic pathway. These
7 results from controlled human exposure studies indicate that sub-populations of healthy study
8 subjects consistently experience larger than average lung function decrements, greater than
9 average inflammatory responses and pulmonary injury as expressed by increased epithelial
10 permeability, and greater than average airway responsiveness, and that these effects are mediated
11 by apparently different physiologic pathways. Except for lung function decrements, we do not
12 have the concentration- or exposure-response function information about the other, potentially
13 more sensitive,17 clinical endpoints (i.e., inflammation, increased epithelial permeability, airway
14 hyperresponsiveness) that would allow us to quantitatively estimate the size of the population
15 affected and the magnitude of their responses. Moreover, some uncertainties about the exact
16 physiological pathways underlying these endpoints prevents us from knowing whether the
17 exaggerated responses are distributed in sub-populations evenly across the population, or may be
18 clustered with more than one type of exaggerated response in particular sub-populations, or both.
19 In summary, a strong body of controlled human exposure and animal toxicological
20 studies, most of which were available in the last review of the O^ NAAQS, report Os-induced
21 airway hyperresponsiveness after either acute or repeated exposures (U.S. EPA, 2013, section
22 6.2.2.2). People with asthma often exhibit increased airway responsiveness at baseline relative to
23 healthy controls, and they can experience further increases in responsiveness following
24 exposures to 63. Studies reporting increased airway responsiveness after 63 exposure contribute
25 to a plausible link between ambient Os exposures and increased respiratory symptoms in
26 asthmatics, and increased hospital admissions and emergency department visits for asthma (U.S.
27 EPA, 2013, section 6.2.2.2).
17 CASAC noted that "...[W]hile measures of FEVi are quantitative and readily obtainable in humans, they are not
the only measures — and perhaps not the most sensitive measures — of the adverse health effects induced by ozone
exposure." (Henderson, 2006).
3-25
-------
1 Respiratory Symptoms and Medication Use
2 Because respiratory symptoms are associated with adverse outcomes such as limitations
3 in activity, and are the primary reason for people with asthma to use quick relief medication and
4 seek medical care, studies evaluating the link between 63 exposures and such symptoms allow a
5 more direct characterization of the clinical and public health significance of ambient 63 exposure
6 than measures of lung function decrements and pulmonary inflammation. Controlled human
7 exposure and toxicological studies have described modes of action through which short-term OT,
8 exposures may increase respiratory symptoms by demonstrating Os-induced airway
9 hyperresponsiveness (U.S. EPA, 2013, section 6.2.2) and pulmonary inflammation (U.S. EPA,
10 2013, section 6.2.3).
11 The link between subjective respiratory symptoms and Os exposures has been evaluated
12 in both controlled human exposure and epidemiologic studies, and the link with medication use
13 has been evaluated in epidemiologic studies. In the last review, several controlled human
14 exposure studies reported respiratory symptoms following exposures to O?, concentrations at or
15 above 80 ppb. In addition, one study reported such symptoms following exposures to 60 ppb 63,
16 though the increase was not statistically different from filtered air controls. Epidemiologic
17 studies reported associations between ambient 63 and respiratory symptoms and medication use
18 in a variety of locations and populations, including asthmatic children living in U.S. cities. In the
19 current review, additional controlled human exposure studies have evaluated respiratory
20 symptoms following exposures to Os concentrations below 80 ppb and recent epidemiologic
21 studies have evaluated associations with respiratory symptoms and medication use (U.S. EPA,
22 2013, sections 6.2.1,6.2.4).
23 In controlled human exposure studies available in the last review as well as newly
24 available studies, statistically significant increases in respiratory symptoms have been
25 consistently reported in healthy volunteers engaged in intermittent, moderate exertion following
26 6.6 hour exposures to average 03 concentrations at or above 80 ppb (Adams, 2003; Adams,
27 2006; Schelegle et al., 2009). Such symptoms have been reported to increase with increasing 63
28 exposure concentrations, duration of exposure, and activity level (McDonnell et al., 1999b). For
29 example, in a study available during the last review, Adams (2006) reported an increase in
30 respiratory symptoms in healthy adults during a 6.6 hour exposure protocol with an average 63
3-26
-------
1 exposure concentration of 60 ppb. This increase was significantly different from initial
2 respiratory symptoms, but not from filtered air controls. Two recent controlled human exposure
3 studies that have become available since the last review did not report statistically significant
4 increases in respiratory symptoms following exposures of healthy adults to 60 ppb 63 (Schelegle
5 et al., 2009; Kim et al., 2011). A recent study by Schelegle et al. (2009) did report a statistically
6 significant increase in respiratory symptoms in healthy adults following 6.6 hour exposures to an
7 average O?, concentration of 70 ppb. The findings for Os-induced respiratory symptoms in
8 controlled human exposure studies, and the evidence integrated across disciplines describing
9 underlying modes of action, provide biological plausibility for epidemiologic associations
10 observed between short-term increases in ambient 63 concentration and increases in respiratory
11 symptoms (U.S. EPA, 2013, section 6.2.4).
12 In epidemiologic studies, respiratory symptom data typically are collected by having
13 subjects (or their parents) record symptoms and medication use in a diary without direct
14 supervision by study staff. Several limitations of symptom reports are well recognized, as
15 described in the ISA (U.S. EPA, 2013, section 6.2.4). Nonetheless, symptom diaries remain a
16 convenient tool to collect individual-level data from a large number of subjects and allow
17 modeling of associations between daily changes in Os concentration and daily changes in
18 respiratory morbidity over multiple weeks or months. Importantly, many of the limitations in
19 these studies are sources of random measurement error that can bias effect estimates to the null
20 or increase the uncertainty around effect estimates (U.S. EPA, 2013, Section 6.2.4). Because
21 respiratory symptoms are associated with limitations in activity and daily function and are the
22 primary reason for using medication and seeking medical care, the evidence is directly coherent
23 with the associations consistently observed between increases in ambient 63 concentration and
24 increases in asthma emergency department visits, discussed below (U.S. EPA, 2013, Section
25 6.2.4).
26 Most epidemiologic studies of 63 and respiratory symptoms and medication use have
27 been conducted in children and/or adults with asthma, with fewer studies, and less consistent
28 results, in non-asthmatic populations (U.S. EPA, 2013, section 6.2.4). The 2006 AQCD (U.S.
29 EPA, 2006b, U.S. EPA, 2013, section 6.2.4) concluded that the collective body of epidemiologic
30 evidence indicated that short-term increases in ambient Os concentrations are associated with
31 increases in respiratory symptoms in children with asthma. A large body of single-city and
3-27
-------
1 single-region studies of asthmatic children provides consistent evidence for associations between
2 short-term increases in ambient Os concentrations and increased respiratory symptoms and
3 asthma medication use in children with asthma (U.S. EPA, 2013, Figure 6-12, Table 6-20).
4 Methodological differences among studies make comparisons across recent multicity
5 studies of respiratory symptoms difficult. Because of fewer person-days of data (Schildcrout et
6 al., 2006) or examination of 19-day averages of ambient Os concentrations (O'Connor et al.,
7 2008), the ISA did not give greater weight to results from recent multicity studies than results
8 from single-city studies (U.S. EPA, 2013, section 6.2.4.5). While evidence from the few
9 available U.S. multicity studies is less consistent (O'Connor et al., 2008; Schildcrout et al., 2006;
10 Mortimer et al., 2002), the overall body of epidemiologic evidence with respect to the
11 association betweeen exposure to Os and respiratory symptoms in asthmatic children remains
12 compelling (U.S. EPA, 2013, section 6.2.4.1). Findings from a small body of studies indicate that
13 Os is also associated with increased respiratory symptoms in adults with asthma (Khatri et al.,
14 2009; Feo Brito et al., 2007; Ross et al., 2002) (U.S. EPA, 2013, section 6.2.4.2).
15 Available evidence indicates that Os-associated increases in respiratory symptoms are not
16 confounded by temperature, pollen, or copollutants (primarily PM) (U.S. EPA, 2013, section
17 6.2.4.5; Table 6-25; Romieu et al., 1996; Romieu et al., 1997; Thurston et al., 1997; Gent et al.,
18 2003). However, identifying the independent effects of Os in some studies was complicated due
19 to the high correlations observed between Os and PM or different lags and averaging times
20 examined for copollutants. Nonetheless, the ISA noted that the robustness of associations in
21 some studies of individuals with asthma, combined with findings from controlled human
22 exposure studies for the direct effects of Os exposure, provide substantial evidence supporting
23 the independent effects of short-term ambient Os exposure on respiratory symptoms (U.S. EPA,
24 2013, section 6.2.4.5).
25 Epidemiologic studies of medication use have reported associations with
26 1-hour maximum Os concentrations and with multiday average Os concentrations (Romieu et al.,
27 2006; Just et al., 2002). Some studies reported Os associations for both respiratory symptoms and
28 asthma medication use (Escamilla-Nufiez et al., 2008; Romieu et al., 2006; Schildcrout et al.,
29 2006; Jalaludin et al., 2004; Romieu et al., 1997; Thurston et al., 1997) while others reported
30 associations for either respiratory symptoms or medication use (Romieu et al., 1996; Rabinovitch
31 et al., 2004; Just et al., 2002; Ostro et al., 2001).
32 In summary, both controlled human exposure and epidemiologic studies have reported
33 respiratory symptoms attributable to short-term Os exposures. In the last review, the majority of
34 the evidence from controlled human exposure studies in young, healthy adults was for symptoms
3-28
-------
1 following exposures to Os concentrations at or above 80 ppb. Although studies that have become
2 available since the last review have not reported respiratory symptoms in young, healthy adults
3 following exposures with moderate exertion to 60 ppb, one recent study has reported increased
4 symptoms in young, healthy adults while at moderate exertion following exposures to 63
5 concentrations as low as 70 ppb. As was concluded in the 2006 63 AQCD (U.S. EPA, 2006b,
6 1996a), the collective body of epidemiologic evidence indicates that short-term increases in
7 ambient 63 concentration are associated with increases in respiratory symptoms in children with
8 asthma (U.S. EPA, 2013, section 6.2.4). Recent studies of respiratory symptoms and medication
9 use, primarily in asthmatic children, add to this evidence. In a smaller body of studies, increases
10 in ambient 63 concentration were associated with increases in respiratory symptoms in adults
11 with asthma.
12 Lung Host Defense
13 The mammalian respiratory tract has a number of closely integrated defense mechanisms
14 that, when functioning normally, provide protection from the potential health effects of
15 exposures to a wide variety of inhaled particles and microbes. These defense mechanisms
16 include mucociliary clearance, alveolobronchiolar transport mechanism, alveolar macrophages18,
17 and adaptive immunity19 (U.S. EPA, 2013, section 6.2.5). The previous O3 AQCD (U.S. EPA,
18 2006) concluded that animal toxicological studies provided evidence that acute exposure to 63
19 concentrations as low as 100 to 500 ppb can increase susceptibility to infectious diseases due to
20 modulation of these lung host defenses. This conclusion was based in large part on animal
21 studies of alveolar macrophage functioning and mucociliary clearance (U.S. EPA, 2013, section
22 6.2.5).
23 With regard to alveolar macrophage functioning, the previous Os AQCD (U.S. EPA,
24 2006) concluded that short periods of Os exposure can cause a reduction in the number of free
25 alveolar macrophages available for pulmonary defense, and that these alveolar macrophages are
90
26 more fragile, less able to engulf particles (i.e., phagocytic), and exhibit decreased lysosomal
27 enzyme activities (U.S. EPA, 2013, section 6.2.5). These conclusions were based largely on
18 Phagocytic white blood cells within the alveoli of the lungs that ingest inhaled particles.
19 The adaptive immune system, is also known as the acquired immune system. Acquired immunity creates
immunological memory after an initial response to a specific pathogen, leading to an enhanced response to
subsequent encounters with that same pathogen.
20 Lysosomes are cellular organelles that contain acid hydrolase enzymes that break down waste materials and
cellular debris.
3-29
-------
1 studies conducted in animals exposed for several hours up to several weeks to Os concentrations
2 from 100 to 250 ppb (Hurst et al., 1970; Driscoll et al., 1987; Cohen et al., 2002). Consistent
3 with the animal evidence, a controlled human exposure study available in the last review had
4 reported decrements in the ability of alveolar macrophages to phagocytize yeast following
5 exposures of healthy volunteers to Os concentrations of 80 and 100 ppb for 6.6-hour during
6 moderate exercise (Devlin et al., 1991). Integrating the animal study results with human
7 exposure evidence available in the 1996 Criteria Document, the 2006 Criteria Document
8 concluded that available evidence indicates that short-term 03 exposures have the potential to
9 impair host defenses in humans, primarily by interfering with alveolar macrophage function. Any
10 impairment in alveolar macrophage function may lead to decreased clearance of microorganisms
11 or nonviable particles. Compromised alveolar macrophage functions in asthmatics may increase
12 their susceptibility to other 63 effects, the effects of particles, and respiratory infections (EPA,
13 2006a, p. 8-26).
14 With regard to mucociliary clearance, in the last review a number of studies conducted in
15 different animal species had reported morphological damage to the cells of the tracheobronchial
16 tree from acute and sub-chronic exposure to 63 concentrations at or above 200 ppb. The cilia
17 were either completely absent or had become noticeably shorter or blunt. In general, functional
18 studies of mucociliary transport had observed a delay in particle clearance soon after acute
19 exposure, with decreased clearance more evident at higher doses (1 ppm) (U.S. EPA, 1986; U.S.
20 EPA, 2013, section 6.2.5.1).
21 Alveolobronchiolar transport mechanisms refers to the transport of particles deposited in
22 the deep lung (alveoli) which may be removed either up through the respiratory tract (bronchi)
23 by alveolobronchiolar transport or through the lymphatic system. The pivotal mechanism of
24 alveolobronchiolar transport involves the movement of alveolar macrophages with ingested
25 particles to the bottom of the conducting airways. These airways are lined with ciliated epithelial
26 cells and cells that produce mucous, which surrounds the macrophages. The ciliated epithelial
27 cells move the mucous packets up the resiratory tract, hence the term "mucociliary escalator."
28 Although some studies show reduced tracheobronchial clearance after 63 exposure, alveolar
29 clearance of deposited material is accelerated, presumably due to macrophage influx, which in
30 itself can be damaging.
3-30
-------
1 With regard to adaptive immunity, a limited number of epidemiologic studies have
2 examined associations between 63 exposure and hospital admissions or ED visits for respiratory
3 infection, pneumonia, or influenza. Results have been mixed, and in some cases conflicting (U.S.
4 EPA, 2013, Sections 6.2.7.2 and 6.2.7.3). With the exception of influenza, it is difficult to
5 ascertain whether cases of respiratory infection or pneumonia are of viral or bacterial etiology.
6 A recent study that examined the association between Os exposure and respiratory hospital
7 admissions in response to an increase in influenza intensity did observe an increase in respiratory
8 hospital admissions (Wong et al., 2009), but information from toxicological studies of Os and
9 viral infections is ambiguous.
10 In summary, relatively few studies conducted since the last review have evaluated the
11 effects of Os exposures on lung host defense. When the available evidence is taken as a whole,
12 the ISA concludes that acute Os exposures impair the host defense capability of animals,
13 primarily by depressing alveolar macrophage function and perhaps also by decreasing
14 mucociliary clearance of inhaled particles and microorganisms. Coupled with limited evidence
15 from controlled human exposure studies, this suggests that humans exposed to 63 could be
16 predisposed to bacterial infections in the lower respiratory tract (EPA, 2013, section 6.2.5.5).
17 The seriousness of such infections may depend on how quickly bacteria develop virulence
18 factors and how rapidly PMNs are mobilized to compensate for the deficit in alveolar
19 macrophage function.
20 Allergic and Asthma-Related Responses
21 Effects resulting from combined exposures to Os and allergens have been studied in a
22 variety of animal species, generally as models of experimental asthma. Pulmonary function and
23 AHR in animal models of asthma are discussed in detail in Section 6.2.1.3 and Section 6.2.2.2,
24 respectively, in the ISA (U.S. EPA, 2013). Studies of allergic and asthma-related responses are
25 discussed in detail in sections 5.3.6 and 6.2.6 of the ISA (U.S. EPA, 2013).
26 Evidence available in the last review indicates that Os exposure skews immune responses
27 toward an allergic phenotype. For example, Gershwin et al. (1981) reported that Os (800 and 500
28 ppb for 4 days) exposure caused a 34-fold increase in the number of IgE (allergic antibody)-
29 containing cells in the lungs of mice. In general, the number of IgE-containing cells correlated
30 positively with levels of anaphylactic sensitivity. In humans, allergic rhinoconjunctivitis
3-31
-------
1 symptoms are associated with increases in ambient O^ concentrations (Riediker et al., 2001).
2 Controlled human exposure studies have observed Cb-induced changes indicating allergic
3 skewing. Airway eosinophils, which are white blood cells that participate in allergic disease and
4 inflammation, were observed to increase in volunteers with atopy21 and mild asthma (Peden et
5 al., 1997). In a more recent study, expression of IL-5, a cytokine involved in eosinophil
6 recruitment and activation, was increased in subjects with atopy but not in healthy subjects
7 (Hernandez et al., 2010). Epidemiologic studies describe associations between eosinophils in
8 both short- (U.S. EPA, 2013, Section 6.2.3.2) and long-term (U.S. EPA, 2013, Section 7.2.5) O3
9 exposure, as do chronic exposure studies in non-human primates. Collectively, findings from
10 these studies suggest that 63 can induce or enhance certain components of allergic inflammation
11 in individuals with allergy or allergic asthma.
12 Evidence available in the last review indicates that ozone may also increase AHR to
13 specific allergen triggers (75 FR 2970, January 19, 2010). Two studies (Torres et al., 1996; Holz
14 et al., 2002) observed increased airway responsiveness to 63 exposure with bronchial allergen
15 challenge in subjects with preexisting allergic airway disease. Ozone-induced exacerbation of
16 airway responsiveness persists longer and attenuates more slowly than (Vinduced lung function
17 decrements and respiratory symptom responses and can have important clinical implications for
18 asthmatics.
19 Animal toxicology studies indicate that 63 enhances inflammatory and allergic responses
20 to allergen challenge in sensitized animals. In addition to exacerbating existing allergic
21 responses, toxicology studies indicate that Os can also act as an adjuvant to produce sensitization
22 in the respiratory tract. Along with its pro-allergic effects (inducing or enhancing certain
23 components of allergic inflammation in individuals with allergy or allergic asthma), Os could
24 also make airborne allergens more allergenic. When combined with NC>2, Os has been shown to
25 enhance nitration of common protein allergens, which may increase their allergenicity Franze et
26 al. (2005).
21 Atopy is a predisposition toward developing certain allergic hypersensitivity reactions. A person with atopy
typically presents with one or more of the following: eczema (atopic dermatitis), allergic rhinitis (hay fever), allergic
conjunctivitis, or allergic asthma.
3-32
-------
1 Hospital Admissions and Emergency Department Visits
2 The 2006 Os AQCD evaluated numerous studies of respiratory-related emergency
3 department visits and hospital admissions. These were primarily time-series studies conducted in
4 the U.S., Canada, Europe, South America, Australia, and Asia. Based on such studies, the 2006
5 Oj, AQCD concluded that "the overall evidence supports a causal relationship between acute
6 ambient 63 exposures and increased respiratory morbidity resulting in increased ED visits and
99 _
7 [hospital admissions] during the warm season " (U.S. EPA, 2006). This conclusion was
8 "strongly supported by the human clinical, animal toxicologic[al], and epidemiologic evidence
9 for [Os-induced] lung function decrements, increased respiratory symptoms, airway
10 inflammation, and airway hyperreactivity" (U.S. EPA, 2006).
11 The results of recent studies largely support the conclusions of the 2006 OsAQCD (U.S.
12 EPA, 2013, section 6.2.7). Since the completion of the 2006 O3 AQCD, relatively fewer studies
13 conducted in the U.S., Canada, and Europe have evaluated associations between short-term Os
14 concentrations and respiratory hospital admissions and emergency department visits, with a
15 growing number of studies conducted in Asia. This epidemiologic evidence is summarized in
16 Appendix 3-B and discussed in detail in the ISA (U.S. EPA, 2013, section 6.2.7).
17 In considering this body of evidence, the ISA focused primarily on multicity studies
18 because they examine associations with respiratory-related hospital admissions and emergency
19 department visits over large geographic areas using consistent statistical methodologies (U.S.
20 EPA, 2013, section 6.2.7.1). The ISA also focused on single-city studies that encompassed a
21 large number of hospital admissions or emergency department visits, included long study-
22 durations, were conducted in locations not represented by the larger studies, or examined
23 population-specific characteristics that were not evaluated in the larger studies (U.S. EPA, 2013,
24 section 6.2.7.1). When examining the association between short-term Os exposure and
25 respiratory health effects that require medical attention, the ISA distinguishes between hospital
26 admissions and emergency department visits because it is likely that a small percentage of
27 respiratory emergency department visits will be admitted to the hospital, therefore respiratory
28 emergency department visits may represent potentially less serious, but more common outcomes
29 (U.S. EPA, 2013, section 6.2.7.1).
30 The results of studies evaluated in this review largely support the conclusions of the 2006
31 AQCD. Several recent multicity studies (e.g., Cakmak et al., 2006; Dales et al., 2006) and a
32 multi-continent study (Katsouyanni et al., 2009) report associations between short-term Os
22Epidemiologic associations for O3 are more robust during the warm season than during cooler months (e.g.,
smaller measurement error, less potential confounding by copollutants). Rationale for focusing on warm season
epidemiologic studies for O3 can be found at 72 FR 37838-37840.
3-33
-------
1 concentrations and increased respiratory-related hospital admissions and emergency department
2 visits. These multeity studies are supported by consistent results from single-city studies using
3 different exposure assignment approaches (i.e., average of multiple monitors, single monitor,
4 population-weighted average) and averaging times (i.e., 1-hour max and 8-hour max) (U.S. EPA,
5 2013, sections 6.2.7.1 to 6.2.7.5). Recent studies also report associations with hospital
6 admissions and emergency department visits for asthma (Strickland et al., 2010; Stieb et al.,
7 2009) and COPD (Stieb et al., 2009; Medina-Ramon et al., 2006), with more limited evidence for
8 pneumonia (Medina-Ramon et al., 2006; Zanobetti and Schwartz, 2006). In seasonal analyses
9 (Figure 3-2 below; U.S. EPA, 2013, Figure 6-19, Table 6-28), stronger associations were
10 reported in the warm season or summer months (red circles) compared to the cold season (blue
11 circles), particularly for asthma (Strickland et al., 2010; Ito et al., 2007b) and COPD (Medina-
12 Ramon et al., 2006). The available evidence indicates that children are at greatest risk for
13 Os-induced respiratory effects (Silverman and Ito, 2010; Strickland et al., 2010; Mar and Koenig,
14 2009; Villeneuve et al., 2007; Dales et al., 2006).
15 Although the collective evidence across studies indicates a mostly consistent positive
16 association between Oj, exposure and respiratory-related hospital admissions and ED visits, the
17 magnitude of these associations may be underestimated due to behavioral modification in
18 response to air quality forecasts (U.S. EPA, 2013, Section 4.6.6).
3-34
-------
Study
Wongetal. (2009)
Cakmaketal. (2006)
Dales etal. (2006)
Orazzoet al.(2009)a
Katsouyanni et al. (2009)
Darrowet al.(2009)
Tolbertetal. (2007)
Biggerietal. (2005)c
Katsouyanni et al.(2009)
Stiebetal. (2009)
Villeneuveetal. (2007)
Strickland etal. (2010)
Silverman and Ito (2010)d
Itp etal. (2007
Villeneuveetal. (2007)
Mar and Koenig 2009
Strickland etal. (2010)
Location
Visit Type
Age
Lag
Silverman and Ito (2010)d
Mar and Koenig (2009)
Itp etal. (2007),
Villeneuveetal. (2007)
Strickland etal. (2010)
Wongetal. (2009)
Stiebetal. (2009)
Yang etal. (2006)
Medina-Ramon etal. (2006)
Stiebetal. (2009)e
Medina-Ramon etal. (2006)
Zanobettiand Schwartz (2006)
Medina-Ramon etal. (2006)
Hong Kong
10 Canadian cities
11 Canadian cities
6 Italian cities
APHENA-Europe
APHENA-U.S.
APHENA-Canada
APHENA-Canada
Atlanta
Atlanta
8 Italian cities
APHENA-Europe
APHENA-U.S\
APHENA-Canada
APHENA-Canada
7 Canadian Cities
Alberta, CAN
Atlanta
New York
New York
Alberta, CAN
Seattle, WA
Atlanta
New York
Seattle, WA
New York
Alberta, CAN
Atlanta
Hong Kong
7 Canadian Cities
Vancouver
36 U.S. cities
7 Canadian Cities
36 U.S. cities
36 U.S. cities
Boston
36 U.S. cities
36 U.S. cities
36 U.S. cities
HA
HA
HA
ED
HA
HA
HA
HA
ED
ED
HA
HA
HA
HA
HA
ED
ED
ED
HA
ED
ED
ED
ED
HA
ED
ED
ED
ED
HA
ED
HA
HA
ED
HA
HA
HA
HA
HA
HA
All
All
0-27 days
0-2
65+
65+
65+
65+
All
All
All
65+
65+
65+
65+
All
>2
Children
All
All
>2
18+
Children
6-18
All
>2
Children
All
All
65+
65+
All
65+
65+
65+
65+
65+
65+
0-1
1.2
2
0-6
0-1
0-1
DLjO-2
DL 0-2
1
0-2
0-3
0-1
0-1
DLjO-2
DL(0-2
2
0-2
0-2
0-1
0-1
0-2
2
0-2
0-1
0
0-1
0-2
0-2
0-1
2
0-3
DL(0-1
NR
DL 0-1
DL 0-1
0-1
DL 0-1
DL 0-1
DL 0-1
Respiratory
Asthma
COPD
Pneumonia
•*----•-
v
I 1 1 1 1 1 1 1 T T 1 1 1
-25 -20 -15 -10 -5 0 5 10 15 20 25 30 35 40
% Increase
Note: Effect estimates are for a 20 ppb increase in 24-h; 30 ppb increase in 8-h max; and 40 ppb increase in 1-h max O3 concentrations. HA=hospital admission; ED=emergency
department. Black=AII-year analysis; Red=Summer only analysis; Blue=Winter only analysis.
a Wheeze used as indicator of lower respiratory disease.
b APHENA-Canada results standardized to approximate IQR of 5.1 ppb for 1-h max O3 concentrations.
0 Study included 8 cities; but of those 8, only 4 had O3 data.
d non-ICU effect estimates.
e The study did not specify the lag day of the summer season estimate.
2 Figure 3-2. Percent increase in respiratory-related hospital admission and emergency department visits in studies that
3 presented all-year and/or seasonal results.
3-35
-------
1 Studies examining the potential confounding effects of copollutants have reported that O?,
2 effect estimates remained relatively robust upon the inclusion of PM and gaseous pollutants in
3 two-pollutant models (U.S. 2013, Figure 6-20, Table 6-29). Additional studies that conducted
4 copollutant analyses, but did not present quantitative results, also support these conclusions
5 (Strickland et al., 2010; Tolbert et al., 2007; Medina-Ramon et al., 2006) (U.S. 2013, section
6 6.2.7.5).
7 In the last review, studies had not evaluated the concentration-response relationship
8 between short-term Oj, exposure and respiratory-related hospital admissions and emergency
9 department visits. A preliminary examination of this relationship in studies that have become
10 available since the last review found no evidence of a deviation from linearity when examining
11 the association between short-term 63 exposure and asthma hospital admissions (U.S. EPA,
12 2013, page 6-157, and Silverman and Ito, 2010). In addition, an examination of the
13 concentration-response relationship for 63 exposure and pediatric asthma emergency department
14 visits found no evidence of a threshold at 63 concentrations as low as 30 ppb (for 8-hour daily
15 maximum concentrations) (Strickland et al., 2010). However, in both studies there is uncertainty
16 in the shape of the concentration-response curve at the lower end of the distribution of 63
17 concentrations due to the low density of data in this range (U.S. 2013, page 6-157).
18 Respiratory Mortality
19 The controlled human exposure, epidemiologic, and toxicological studies discussed in
20 section 6.2 of the ISA (U.S. EPA, 1013a, section 6.2) provide strong evidence for respiratory
21 morbidity effects, including ED visits and hospital admissions, in response to short-term 63
22 exposures. Moreover, evidence from experimental studies indicates multiple potential pathways
23 of respiratory effects from short-term 63 exposures, which support the continuum of respiratory
24 effects that could potentially result in respiratory-related mortality in adults (U.S. EPA, 1013a,
25 section 6.2.8). The 2006 O^ AQCD found inconsistent evidence for associations between short-
26 term 63 concentrations and respiratory mortality (U.S. EPA, 2006). Although some studies
27 reported a strong positive association between Os and respiratory mortality, additional studies
28 reported small associations or no associations. New epidemiologic evidence for respiratory
29 mortality is discussed in detail in section 6.2.8 of the ISA (U.S. EPA, 2013). The majority of
30 recent multicity studies have reported positive associations between short-term Os exposures and
31 respiratory mortality, particularly during the summer months (U.S. EPA, 2013, Figure 6-36).
32 Specifically, recent multicity studies from the U.S. (Zanobetti and Schwartz, 2008b),
33 Europe (Samoli et al., 2009), Italy (Stafoggia et al., 2010), and Asia (Wong et al., 2010), as well
34 as a multi-continent study (Katsouyanni et al., 2009), reported associations between short-term
3-36
-------
1 Os concentrations and respiratory mortality (U.S. EPA, 2013, Figure 6-37, page 6-259). With
2 respect to respiratory mortality, summer-only analyses were consistently positive and most were
3 statistically significant. All-year analyses had more mixed results, but most were positive.
4 Of the studies evaluated, only the studies by Katsouyanni et al. (2009) and by Stafoggia
5 et al. (2010) analyzed the potential for copollutant confounding of the Os-respiratory mortality
6 relationship. Based on the results of these analyses, the ISA concluded that 63 respiratory
7 mortality risk estimates appear to be moderately to substantially sensitive (e.g., increased or
8 attenuated) to inclusion of PMio. However, in the APHENA study, the mostly every-6th-day
9 sampling schedule for PMio in the Canadian and U.S. datasets greatly reduced their sample size
10 and limits the interpretation of these results (U.S. EPA, 2013, section 6.2.8).
11 In summary, recent epidemiologic studies support and reinforce the epidemiologic
12 evidence for (Vassociated respiratory hospital admissions and emergency department visits
13 from the last review. In addition, the evidence for associations with respiratory mortality has
14 been strengthened considerably since the last review, with the addition of several large multicity
15 studies. The plausibility of the associations reported in these studies is supported by the
16 experimental evidence for respiratory effects.
17 3.1.2.2 Respiratory Effects - Long-term Exposures
18 • To what extent does the currently available scientific evidence, including related
19 uncertainties, strengthen or alter our understanding from the last review of
20 respiratory effects attributable to long-term Os exposures?
21 As recognized in section 3.1.2.1, "the clearest evidence for health effects associated with
22 exposure to 63 is provided by studies of respiratory effects" (U.S. EPA, 2013, section 1, p. 1-6).
23 Collectively, there is a vast amount of evidence spanning several decades that supports a causal
24 association between exposure to 63 and a continuum of respiratory effects (U.S. EPA, 2013,
25 section 2.5). While the majority of this evidence is derived from studies investigating short-term
26 exposures, evidence from animal toxicological studies and recent epidemiologic evidence
27 indicate that long-term exposures (i.e., months to years) may also be detrimental to the
28 respiratory system. Across this evidence, particularly the epidemiologic evidence, the exposures
29 of focus vary, often involving repeated short concentrations extending over a long period, rather
30 than a continuous long-term exposure period.
31 In the 2006 Os AQCD, evidence was examined for relationships between long-term Os
32 exposure and effects on respiratory health outcomes including declines in lung function,
33 increases in inflammation, and development of asthma in children and adults. Animal toxicology
34 data provided a clearer picture indicating that long-term O3 exposure may have lasting effects.
3-37
-------
9^
1 Chronic exposure studies in animals have reported biochemical and morphological changes
2 suggestive of irreversible long-term 63 impacts on the lung. In contrast to supportive evidence
3 from chronic animal studies, the epidemiologic studies on longer-term (annual) lung function
4 declines, inflammation, and new asthma development remained inconclusive.
5 Several epidemiologic studies collectively indicated that Os exposure averaged over
6 several summer months was associated with smaller increases in lung function growth in
7 children. For longer averaging periods (annual), the analysis in the Children's Health Study
8 (CHS) reported by Gauderman et al. (2004) provided little evidence that such long-term
9 exposure to ambient Os was associated with significant deficits in the growth rate of lung
10 function in children. Limited epidemiologic research examined the relationship between long-
11 term 63 exposures and inflammation. Cross-sectional studies detected no associations between
12 long-term Os exposures and asthma prevalence, asthma-related symptoms or allergy to common
13 aeroallergens in children. However, longitudinal studies provided evidence that long-term 63
14 exposure influences the risk of asthma development in children and adults.
15 The currently available body of evidence supporting a relationship between long-term 03
16 exposures and adverse respiratory health effects that is likely to be causal is discussed in detail in
17 the ISA (EPA 2013, section 7.2). New evidence reports interactions between genetic variants and
18 long-term Os exposure affect the occurrence of new-onset asthma in multi-community, U.S.
19 cohort studies where protection by specific oxidant gene variants was restricted to children living
20 in low Os communities. A new line of evidence reports a positive concentration-response
21 relationship between first asthma hospitalization and long-term 63 exposure. Related studies
22 report coherent relationships between asthma severity and control, and respiratory symptoms
23 among asthmatics and long-term Os exposure. There is also limited evidence for an association
24 between long-term exposure to ambient 63 concentrations and respiratory mortality. These
25 studies are summarized briefly below for new-onset asthma and asthma prevalence, asthma
26 hospital admissions and other morbidity effects, pulmonary structure and function, and
27 respiratory mortality.
28 Currently available scientific evidence of the adverse health effects attributable to long-
29 term Os exposures, even considering related uncertainties, is much stronger than the body of
30 evidence available at the time of the 2008 review of the O3 standard. The 2006 O3 AQCD (U.S.
31 EPA 2006) concluded that epidemiologic studies provided no evidence of associations between
32 long-term (annual) Os exposures and asthma-related symptoms, asthma prevalence, or allergy to
23 Unless otherwise specified, the term "chronic" generally refers to an annual exposure duration for epidemiologic
studies and a duration of greater than 10% of the lifespan of the animal in lexicological studies.
3-38
-------
1 common allergens after controlling for covariates. It found limited evidence for a relationship
2 between long-term exposures to ambient 63 and deficits in the growth rate of lung-function in
3 children, pulmonary inflammation and other endpoints. Episodic exposures were also known to
4 cause more severe pulmonary morphological changes than continuous exposure.
5 The evidence base available in this review includes additional epidemiologic studies
6 using a variety of designs and analysis methods evaluating the relationship between long-term 63
7 exposures and measures of respiratory morbidity and mortality effects conducted by different
8 research groups in different locations. The ISA (U.S. EPA, 2013, p. 7-33), in Table 7-2 presents
9 selected key new longitudinal and cross-sectional studies of respiratory health effects and
10 associated 03 concentrations. The positive results from various designs and locations support a
11 relationship between long-term exposure to ambient 63 and respiratory health effects and
12 mortality.
13 In this review, the evidence of effects associated with long-term exposures strengthens
14 the relationship between 03 exposure and health effects defined as adverse by the American
15 Thoracic Society (ATS), a definition that has been used in previous reviews of the 63 standard.
16 As discussed in more detail in section 3.1.3 below, the ATS (1985) defined adverse as
17 "medically significant physiologic or pathologic changes generally evidenced by one or more of
18 the following: (1) interference with the normal activity of the affected person or persons, (2)
19 episodic respiratory illness, (3) incapacitating illness, (4) permanent respiratory injury, and/or (5)
20 progressive respiratory dysfunction." As discussed below, in this review there is now credible
21 evidence of respiratory health effects associated with long-term OT, exposures that would fall in
22 to each of these five categories that define adversity.
23 From a policy perspective, the recent epidemiologic studies from the CHS of long-term
24 63 exposures that shed light on the interaction between genetic variability, 63 exposures, and
25 health effects in children are important, not only because they help clarify previous findings, but
26 also because the effects evaluated, such as new-onset asthma, are clearly adverse. The ISA (U.S.
27 EPA, 2013, p. 7-12) notes that the collective evidence from CHS provides an important
28 demonstration of gene-environment interactions. It further notes that in the complex
29 gene-environment setting a modifying effect might not be reflected in an exposure main effect
30 and that the simultaneous occurrence of main effect and interaction effect can occur. Moreover,
31 the study of gene-environment interactions elucidates disease mechanisms in humans by using
32 information on susceptibility genes to focus on the biological pathways that are most relevant to
33 that disease (Hunter, 2005).
34 In the CHS cohort of children in 12 Southern California communities, long-term
35 exposure to 03 concentrations was not associated with increased risk of developing asthma
3-39
-------
1 (McConnell et al., 2010); however, greater outdoor exercise was associated with development of
2 asthma in children living in communities with higher ambient Oi concentrations (McConnell et
3 al., 2002). Recent CHS studies examined interactions among genetic variants, long-term O^
4 exposure, and new onset asthma in children. These prospective cohort studies are
5 methodologically rigorous epidemiologic studies, and evidence indicates gene-Cb interactions.
6 These studies have provided data supporting decreased risk of certain genetic variants on new
7 onset asthma (e.g., HMOX-1, ARG) that is limited to children either in low (Islam et al., 2008)
8 or high (Salam et al., 2009) Os communities. Gene-environment interaction also was
9 demonstrated with findings that greater outdoor exercise increased risk of asthma in GSTP1
10 lie/lie children living in high Oi communities (Islam et al., 2009). Biological plausibility for
11 these gene-Os environment interactions is provided by evidence that these enzymes have
12 antioxidant and/or anti-inflammatory activity and participate in well recognized modes of action
13 in asthma pathogenesis. As Os is a source of oxidants in the airways, oxidative stress serves as
14 the link among O^ exposure, enzyme activity, and asthma. Cross-sectional studies by Akinbami
15 et al. (2010) and Hwang et al. (2005) provide further evidence relating 63 exposures with asthma
16 prevalence.
17 Studies using a cross-sectional design provide support for a relationship between long-
18 term O^ exposure and adverse health effects in asthmatics, including: bronchitic symptoms
19 (related to TNF-308 genotype in asthmatic children) (Lee et al., 2009b); asthma severity (Rage et
20 al., 2009b) and asthma control (Jacquemin et al., 2012) in an adult cohort; respiratory-related
21 school absences (related to CAT and MPO variant genes) (Wenten et al., 2009); asthma ED
22 visits in adults (Meng et al., 2010); and, asthma hospital admissions in adults and children (Lin et
23 al., 2008; Meng et al., 2010; Moore et al., 2008). Several studies, shown in Table 7-3 (ISA, U.S.
24 EPA, 2013, p. 7-35), provide results adjusted for potential confounders presenting results for
25 both Os and PM (in single and multipollutant models) as well as other pollutants where PM
26 effects were not provided. As shown in this table, 03 associations were generally robust to
27 adjustment by potential confounding by PM.
28 Information from toxicological studies in nonhuman primates indicates that long term
29 exposure to OT, during gestation or development can result in irreversible morphological changes
30 in the lung, which in turn can influence the function of the respiratory tract. This nonhuman
31 primate evidence of an Os-induced change in airway responsiveness supports the biologic
32 plausibility of long term exposure to Os contributing to effects of asthma in children. However,
33 results from epidemiologic studies examining long-term 63 exposure and pulmonary function
34 effects are inconclusive with some new studies relating effects at higher exposure levels.
3-40
-------
1 The ISA (U.S. EPA, 2013, p. 7-31) concludes that there is limited evidence for an
2 association between long-term exposure to ambient Os concentrations and respiratory mortality
3 in adults (Jerrett et al., 2009). This effect was robust to the inclusion of PM2.5 and insensitive to a
4 number of different model specifications. Moreover, there is evidence that long-term exposure to
5 Os is associated with mortality among individuals that had previously experienced an emergency
6 hospital admission due to COPD (Zanobetti and Schwartz, 2011).
7 In conclusion, since the last review, the body of evidence about the effects of long-term
8 Os exposure has been considerably strengthened. The scientific evidence available for this
9 review, including related uncertainties, provides an overall strong body of evidence of adverse
10 health effects attributable to long-term 63 exposures. These include a coherent range of asthma
11 morbidity effects such as new-onset asthma, asthma prevalence, symptoms, school absences, ED
12 visits and hospital admissions. There is also new evidence of respiratory mortality associated
13 with long-term Os exposure. Further discussion of key studies is below.
14 New-onset Asthma and Asthma Prevalence
15 Asthma is a heterogeneous disease with a high degree of temporal variability. The on-set,
16 progression, and symptoms can vary within an individual's lifetime, and the course of asthma
17 may vary markedly in young children, older children, adolescents, and adults. In the previous
18 review, longitudinal cohort studies that examined associations between long-term Os exposures
19 and the onset of asthma in adults and children indicated a direct effect of long-term Os exposures
20 on asthma risk in adults (McDonnell et al., 1999a, 15-year follow-up; Greer et al., 1993, 10-year
21 follow-up) and effect modification by Os in children (McConnell et al., 2002). Since that review,
22 important new evidence has become available about the association between long-term
23 exposures to Os and new-onset asthma that has increased our understanding of the gene-
24 environment interaction and the mechanisms and biological pathways most relevant to assessing
25 Os-related effects.
26 In children, the relationship between long-term Os exposure and new-onset asthma has
27 been extensively studied in the CHS; a long-term study that was initiated in the early!990's
28 which has evaluated effects in several cohorts of children. The CHS was initially designed to
29 examine whether long-term exposure to ambient pollution was related to chronic respiratory
30 outcomes in children in 12 communities in southern California. In the CHS, new-onset asthma
31 was classified as having no prior history of asthma at study entry with subsequent report of
32 physician-diagnosed asthma at follow-up, with the date of onset assigned to be the midpoint of
33 the interval between the interview date when asthma diagnosis was first reported and the
34 previous interview date. The results of one study (McConnell et al., 2002) available in the
3-41
-------
1 previous review indicated that within high O?, communities, asthma risk was 3.3 times greater for
2 children who played three or more outdoor sports as compared with children who played no
3 sports.
4 For this review, as discussed in section 7.2.1.1 of the ISA (U.S. EPA 2013), recent
5 studies from the CHS provide evidence for gene-environment interactions in effects on new-
6 onset asthma by indicating that the lower risks associated with specific genetic variants are found
7 in children who live in lower O^ communities. These studies indicate that the risk for new-onset
8 asthma is related in part to genetic susceptibility, as well as behavioral factors and environmental
9 exposure. The onset of a chronic disease, such as asthma, is partially the result of a sequence of
10 biochemical reactions involving exposures to various environmental agents metabolized by
11 enzymes related to a number of different genes. Oxidative stress has been proposed to underlie
12 the mechanistic hypotheses related to Os exposure. Genetic variants may impact disease risk
13 directly, or modify disease risk by affecting internal dose of pollutants and other environmental
14 agents and/or their reaction products, or by altering cellular and molecular modes of action.
15 Understanding the relation between genetic polymorphisms and environmental exposure can
16 help identify high-risk subgroups in the population and provide better insight into pathway
17 mechanisms for these complex diseases.
18 The CHS analyses (Islam et al., 2008; Islam et al. 2009; Salam et al., 2009) have found
19 that asthma risk is related to interactions between O^ and variants in genes for enzymes such as
20 heme-oxygenase (HO-1), arginases (ARG1 and 2), and glutathione S transferase PI (GSTP1).
21 Biological plausibility for these findings is provided by evidence that these enzymes have
22 antioxidant and/or anti-inflammatory activity and participate in well-recognized modes of action
23 in asthma pathogenesis. Further, several lines of evidence demonstrate that secondary oxidation
24 products of Os initiate the key modes of action that mediate downstream health effects (ISA,
25 Section 5.3, U.S. EPA, 2013). For example, HO-1 responds rapidly to oxidants, has anti-
26 inflammatory and anti-oxidant effects, relaxes airway smooth muscle, and is induced in the
27 airways during asthma. Gene-environment interactions are discussed in detail in Section 5.4.2.1
28 in the ISA (U.S. EPA, 2013).
29 Asthma Hospital Admissions
30 In the 2006 AQCD, studies on (Vrelated hospital discharges and emergency department
31 (ED) visits for asthma and respiratory disease mainly looked at short-term (daily) metrics. The
32 short-term Os studies presented in section 6.2.7.5 of the ISA (U.S. EPA 2013) and discussed
33 above in section 3.1.2.1 continue to indicate that there is evidence for increases in both hospital
34 admissions and ED visits in children and adults related to all respiratory outcomes, including
3-42
-------
1 asthma, with stronger associations in the warm months. New studies, discussed in section 7.2.2
2 of the ISA (U.S. EPA, 2013) also evaluated long-term 63 exposure metrics, providing a new line
3 of evidence that suggests a positive exposure-response relationship between the first hospital
4 admission for asthma and long-term 63 exposure, although the ISA cautions in attributing the
5 associations in that study to long-term exposures since there is potential for short-term exposures
6 to contribute to the observed associations.
7 Evidence associating long-term 03 exposure to first asthma hospital admission in a
8 positive concentration-response relationship is provided in a retrospective cohort study (Lin et
9 al., 2008b). This study investigated the association between chronic exposure to Os and
10 childhood asthma admissions by following a birth cohort of more than 1.2 million babies born in
11 New York State (1995-1999) to first asthma admission or until 31 December 2000. Three annual
12 indicators (all 8-hour maximum from 10:00 a.m. to 6:00 p.m.) were used to define chronic Os
13 exposure: (1) mean concentration during the follow-up period (41.06 ppb); (2) mean
14 concentration during the 63 season (50.62 ppb); and (3) proportion of follow-up days with 63
15 levels >70 ppb. The effects of co-pollutants were controlled, and interaction terms were used to
16 assess potential effect modifications. A positive association between chronic exposure to 63 and
17 childhood asthma hospital admissions was observed, indicating that children exposed to high Os
18 levels over time are more likely to develop asthma severe enough to be admitted to the hospital.
19 The various factors were examined and differences were found for younger children (1-2 years),
20 poor neighborhoods, Medicaid/self-paid births, geographic region and others. As shown in the
21 ISA, Figure 7-3 (EPA 2013, p. 7-16), positive concentration-response relationships were
22 observed. Asthma admissions were significantly associated with increased 63 levels for all
23 chronic exposure indicators.
24 In considering the relationship between long-term pollutant exposures and chronic
25 disease heath endpoints, where chronic pathologies are found with acute expression of chronic
26 disease, Kiinzli (2012) hypothesizes that if the associations of pollution with events are much
27 larger in the long-term studies, it provides some indirect evidence that air pollution increases the
28 pool of subjects with chronic disease, and that more acute events are to be expected to be seen
29 for higher exposures. The results of Lin et al (2008) for first asthma hospital admission,
30 presented in Figure 7-3 (EPA 2013, p. 7-16), show effects estimates that are larger than those
31 reported in a study of childhood asthma hospital admission in New York state (Silverman and
32 Ito, 2010), discussed in section 3.1.2.1 and 3.1.2.2 above. The ISA (U.S. EPA, 2013, p. 7-16)
33 notes that this provides some support for the hypothesis that 63 exposure may not only have
34 triggered the events but also increased the pool of asthmatic children, but cautions in attributing
3-43
-------
1 the associations in Lin et al. (2008) study to long-term exposures since there is potential for
2 short-term exposures to contribute to the observed associations.
3 Pulmonary structure and function
4 In the 2006 Os AQCD, few epidemiologic studies had investigated the effect of chronic
5 Os exposure on pulmonary function. The strongest evidence was for medium-term effects of
6 extended Os exposures over several summer months on lung function (FEVi) in children, i.e.,
7 reduced lung function growth being associated with higher ambient Os levels. Short-term Os
8 exposure studies presented in ISA (EPA, 2013, Section 6.2.1.2), and above in section 3.1.2.1,
9 provide a cumulative body of epidemiologic evidence that strongly supports associations
10 between ambient Os exposure and decrements in lung function among children. A recent study
11 (Rojas-Martinez et al., 2007) of long-term exposure to Os, described in section 7.2.3.1 of the ISA
12 (U.S. EPA, 2013, p. 7-19), observed a relationship with pulmonary function declines in school-
13 aged children where Os and other pollutant levels were higher (90 ppb at high end of the range)
14 than those in the CHS. Two studies of adult cohorts provide mixed results where long-term
15 exposures were at the high end of the range.
16 Long-term studies in animals allow for greater insight into the potential effects of
17 prolonged exposure to Os that may not be easily measured in humans, such as structural changes
18 in the respiratory tract. Despite uncertainties, epidemiologic studies observing associations of Os
19 exposure with functional changes in humans can attain biological plausibility in conjunction with
20 long-term toxicological studies, particularly Os-inhalation studies performed in non-human
21 primates whose respiratory systems most closely resembles that of the human. An important
22 series of studies, discussed in section 7.2.3.2 of the ISA (U.S. EPA, 2013), have used nonhuman
23 primates to examine the effect of Os alone, or in combination with an inhaled allergen, house
24 dust mite antigen (HDMA), on morphology and lung function. These animals exhibit the
25 hallmarks of allergic asthma defined for humans, including: a positive skin test for HDMA with
26 elevated levels of IgE in serum and IgE-positive cells within the tracheobronchial airway walls;
27 impaired airflow which is reversible by treatment with aerosolized albuterol; increased
28 abundance of immune cells, especially eosinophils, in airway exudates and bronchial lavage; and
29 development of nonspecific airway responsiveness (NHLBI, 2007). These studies and others
30 have demonstrated changes in pulmonary function and airway morphology in adult and infant
31 nonhuman primates repeatedly exposed to environmentally relevant concentrations of Os (ISA,
32 section 7.2.3.2, 2013).
33 The initial observations in adult nonhuman primates have been expanded in a series of
34 experiments using infant rhesus monkeys repeatedly exposed to 0.5 ppm Os starting at 1 month
3-44
-------
1 of age (Plopper et al., 2007). The purpose of these studies was to determine if a cyclic regimen of
2 63 inhalation would amplify the allergic responses and structural remodeling associated with
3 allergic sensitization and inhalation in the infant rhesus monkey. After several episodic
4 exposures of infant monkeys to 63, they observed a significant increase in the baseline airway
5 resistance, which was accompanied by a small increase in airway responsiveness to inhaled
6 histamine (Schelegle et al., 2003), although neither measurement was statistically different from
7 filtered air control values. Exposure of animals to inhaled house dust mite antigen alone also
8 produced small but not statistically significant changes in baseline airway resistance and airway
9 responsiveness, whereas the combined exposure to both (63 + antigen) produced statistically
10 significant and greater than additive changes in both functional measurements. This nonhuman
11 primate evidence of an Os-induced change in airway resistance and responsiveness provides
12 biological plausibility of long-term exposure, or repeated short-term exposures, to 63
13 contributing to the effects of asthma in children.
14 To understand which conducting airways and inflammatory mechanisms are involved in
15 Os-induced airway hyperresponsiveness in the infant rhesus monkey, results of a follow-up study
16 (load et al., 2006) suggest that effect of 63 on airway responsiveness occurs predominantly in
17 the smaller bronchioles, where dosimetric models indicate the dose would be higher.
18 The functional changes in the conducting airways were accompanied by a number of cellular and
19 morphological changes, including a significant 4-fold increase in eosinophils. Thus, these studies
20 demonstrate both functional and cellular changes in the lung of infant monkeys after cyclic
21 exposure to 0.5 ppm 63, providing relevant information to understanding the potentially
22 damaging effects of ambient 63 exposure on the respiratory tract of children.
23 In addition, significant structural changes in the respiratory tract development, during
24 which conducting airways increase in diameter and length, have been observed in infant rhesus
25 monkeys after cyclic exposure to 63 (Fanucchi et al., 2006). Observed changes included more
26 proximal first alveolar outpocketing, decreases in the diameter and length of the terminal and
27 respiratory bronchioles, increases in mucus-producing goblet cell mass, alterations in smooth
28 muscle orientation in the respiratory bronchioles, epithelial nerve fiber distribution, and
29 basement membrane zone morphometry. The latter effects are noteworthy because of their
30 potential contribution to airway obstruction and airway hyperresponsiveness which are central
31 features of asthma. A number of studies in both non-human primates and rodents demonstrate
32 that Os exposure can increase collagen synthesis and deposition, including fibrotic-like changes
33 in the lung (ISA, section 7.2.3.2, U.S. EPA, 2013).
34 Collectively, evidence from animal studies strongly suggests that chronic O?, exposure is
35 capable of damaging the distal airways and proximal alveoli, resulting in lung tissue remodeling
3-45
-------
1 and leading to apparent irreversible changes. Potentially, persistent inflammation and interstitial
2 remodeling play an important role in the progression and development of chronic lung disease.
3 Further discussion of the modes of action that lead to Os-induced morphological changes can be
4 found in Section 5.3.7 of the ISA (U.S. EPA, 2013). Discussion of mechanisms involved in
5 lifestage susceptibility and developmental effects can be found in Section 5.4.2.4 of the ISA
6 (U.S. EPA, 2013). The findings reported in chronic animal studies offer insight into potential
7 biological mechanisms for the suggested association between seasonal 63 exposure and reduced
8 lung function development in children as observed in epidemiologic studies (see Section 7.2.3.1).
9 Respiratory Mortality
10 A limited number of epidemiologic studies have assessed the relationship between long-
11 term exposure to 63 and mortality in adults. The 2006 63 AQCD concluded that an insufficient
12 amount of evidence existed "to suggest a causal relationship between chronic 63 exposure and
13 increased risk for mortality in humans" (U.S. EPA, 2006b). Though total and cardio-pulmonary
14 mortality were considered in these studies, respiratory mortality was not specifically considered.
15 In the most recent follow-up analysis of the ACS cohort (Jerrett et al., 2009), cardiopulmonary
16 deaths were separately subdivided into respiratory and cardiovascular deaths, rather than
17 combined as in the Pope et al. (2002) work. Increased 63 exposure was associated with the risk
18 of death from respiratory causes, and this effect was robust to the inclusion of PM2.5. The
19 association between increased 63 concentrations and increased risk of death from respiratory
20 causes was insensitive to the use of different models and to adjustment for several ecologic
21 variables considered individually. Additionally, a recent multi-city time series study (Zanobetti
22 and Schwartz, 2011), which followed (from 1985 to 2006) four cohorts of Medicare enrollees
23 with chronic conditions that might predispose to Cb-related effects, observed an association
24 between long-term (warm season) exposure to 03 and elevated risk of mortality in the cohort that
25 had previously experienced an emergency hospital admission due to COPD. A key limitation of
26 this study is the inability to control for PM2.5, because data were not available in these cities until
27 1999.
28 3.1.2.3 Total Mortality - Short-term Exposures
29 • To what extent does the currently available scientific evidence, including related
30 uncertainties, strengthen or alter our understanding from the last review of
31 mortality attributable to short-term Os exposures?
32 The 2006 63 AQCD concluded that the overall body of evidence was highly suggestive
33 that short-term exposure to OT, directly or indirectly contributes to nonaccidental and
34 cardiopulmonary-related mortality in adults, but additional research was needed to more fully
3-46
-------
1 establish underlying mechanisms by which such effects occur (U.S. EPA, 2013, p. 2-18). In
2 building on the 2006 evidence, the ISA states the following (U.S. EPA, 2013, p. 6-261).
3 The evaluation of new multicity studies that examined the association between
4 short-term Os exposures and mortality found evidence that supports the
5 conclusions of the 2006 AQCD. These new studies reported consistent positive
6 associations between short-term 0j exposure and all-cause (non-accidental)
1 mortality, with associations persisting or increasing in magnitude during the
8 warm season, and provide additional support for associations between Oj
9 exposure and cardiovascular and respiratory mortality
10 The 2006 Os AQCD reviewed a large number of time-series studies of associations
11 between short-term 63 exposures and total mortality including single- and multicity studies, and
12 meta-analyses. In the large U.S. multicity studies that examined all-year data, effect estimates
13 corresponding to single-day lags ranged from a 0.5-1% increase in all-cause (nonaccidental) total
14 mortality per a 20 ppb (24-hour), 30 ppb (8-hour maximum), or 40 ppb (1-hour maximum)
15 increase in ambient Os (U.S. EPA, 2013, section 6.6.2). Available studies reported some
16 evidence for heterogeneity in 63 mortality risk estimates across cities and across studies. Studies
17 that conducted seasonal analyses reported larger Os mortality risk estimates during the warm
18 season. Overall, the 2006 63 AQCD identified robust associations between various measures of
19 daily ambient 63 concentrations and all-cause mortality, which could not be readily explained by
20 confounding due to time, weather, or copollutants. With regard to cause-specific mortality,
21 consistent positive associations were reported between short-term 63 exposure and
22 cardiovascular mortality, with less consistent evidence for associations with respiratory
23 mortality. The majority of the evidence for associations between 03 and cause-specific mortality
24 were from single-city studies, which had small daily mortality counts and subsequently limited
25 statistical power to detect associations. The 2006 Os AQCD concluded that "the overall body of
26 evidence is highly suggestive that 63 directly or indirectly contributes to non-accidental and
27 cardiopulmonary-related mortality" (U.S. EPA, 2012a, section 6.6.1).
28 Recent studies have strengthened the body of evidence that supports the association
29 between short-term Os concentrations and mortality in adults. This evidence includes a number
30 of studies reporting associations with non-accidental as well as cause-specific mortality. Multi-
31 continent and multicity studies have consistently reported positive and statistically significant
32 associations between short-term 03 concentrations and all-cause mortality, with evidence for
33 larger mortality risk estimates during the warm or summer months (Figure 3-3 below, reprinted
34 from the ISA) (U.S. EPA, 2013, Figure 6-27; Table 6-42). Similarly, evaluations of cause-
3-47
-------
1 specific mortality have reported consistently positive associations with Os, particularly in
2 analyses restricted to the warm season (U.S. EPA, 2013, Figure 6-37; Table 6-53).24
4Respiratory mortality is discussed in more detail above.
3-48
-------
1
2
Study
Gryparisetal.{2004;57276)
Bell etal. (2007; 93256)
Schwartz (2005; 57333)
Bell and Dominici(2008; 193828)
Bell etal. (2004; 94417)a
Levy etal. (2005; 74 34 7)a
Katsouyanni et al. (2009; 199899)
Bell etal. (2005; 74345 )a
ltd etal. (2005; 74346)a
Wong etal. (2010; 732535)
Katsouyanni et al. (2009; 199899)
Cakmaketal. (2011, 699135)
Katsouyanni et al. (2009; 199899)
Katsouyanni et al. (2009; 199899)b
Samoli etal. (2009; 195855)
Bell etal. (2004; 94417)a
Schwartz (2005; 57333)
Zanobetti and Schwartz(2008; 195755)
Zanobetti and Schwartz(2008; 101596)
Franklin and Schwartz (2008; 156448)
Gryparisetal.(2004;57276)
Medina-Ramon and Schwartz (2008)
Katsouyanni et al. (2009; 199899)
Bell etal. (2005; 74345)a
Katsouyanni et al. (2009; 199899)
Katsouyanni et al. (2009; 199899Jb
Levy etal. (2005; 74 34 7)a
Ito etal. (2005; 74346)a
Katsouyanni et al. (2009; 199899)
Stafoggia et al. (2010; 625034)
Location
APHEA2 (23 cities)
98 U.S. communities
14 U.S. cities
98 U.S. communities
95 U.S. communities
U.S. and Non-U.S.
APHENA-Europe
U.S. and Non-U.S.
U.S. and Non-U.S.
PAPA (4 cities
APHENA-U.S.
7 Chilean cities
APHENA-Canada
APHENA-Canada
21 European cities
95 U.S. communities
14 U.S. cities
48 U.S. cities
48 U.S. cities
18 U.S. communities
APHEA2 (21 cities)
48 U.S. cities
APHENA-Europe
U.S. and Non-U.S.
APHENA-Canada
APHENA-Canada
U.S. and Non-U.S.
U.S. and Non-U.S.
APHENA-U.S.
10 Italian cities
Lag
0-1
0-1
0
0-6
0-6
DL(0-2)
All-Year
DL
DL
DL
DL
0-1
0-2)
0-6
0-2
0-2
0-1
0-6
0
0
0-3
0
0-1
DL(0-2)
DL(0-2
DLO-2
DL(0-2
DL 0-5
Summer
1
-_ J-l
w ^
^^
^^ w
.::::*.:..
1 3 5 7 9 11
4
5
% Increase
Figure 3-3. Summary of mortality risk estimates for short-term Os and all-cause (nonaccidental) mortality.
25
25Reprinted from the ISA (U.S. EPA, 2013, Figure 6-27).
December 2013
3-49
Draft - Do Not Quote or Cite
-------
1 In assessing the evidence for Os-related mortality, the 2006 AQCD also noted that
2 multiple uncertainties remained regarding the relationship between short-term 63 concentrations
3 and mortality, including the extent of residual confounding by co-pollutants; characterization of
4 the factors that modify the (Vmortality association; the appropriate lag structure for identifying
5 Os-mortality effects; and the shape of the Os-mortality concentration-response function and
6 whether a threshold exists. Many of the studies, published since the last review, have attempted
7 to address one or more of these uncertainties. The ISA (U.S. EPA, 2013; Section 6.6.2) discusses
8 the extent to which recent studies have evaluated these uncertainties in the relationship between
9 Oj, and mortality.
10 In particular, recent studies have evaluated different statistical approaches to examine the
11 shape of the (Vmortality concentration-response relationship and to evaluate whether a
12 threshold exists for Os-related mortality. In an analysis of the NMMAPS data, Bell et al. (2006)
13 evaluated the potential for a threshold in the (Vmortality relationship. The authors reported
14 positive and statistically significant associations with mortality in a variety of restricted analyses,
15 including analyses restricted to days with 24-hour area-wide average Os concentrations below
16 60, 55, 50, 45, 40, 35, and 30 ppb. In these restricted analyses 63 effect estimates were of similar
17 magnitude, were statistically significant, and had similar statistical precision. In analyses
18 restricted to days with 24-hour average O?, concentrations below 25 ppb, the Os effect estimate
19 was similar in magnitude to the effect estimates resulting from analyses with the higher cutoffs,
20 but had somewhat lower statistical precision, with the estimate approaching statistical
21 significance (i.e., based on observation of Figure 2 in Bell et al., 2006). In analyses restricted to
22 days with lower 24-hour average 63 concentrations (i.e., below 20 and 15 ppb), effect estimates
23 were similar in magnitude to analyses with higher cutoffs, but with notably less statistical
24 precision, and were not statistically significant (i.e., confidence intervals included no 63-
25 associated mortality based on observation of Figure 2 in Bell et al., 2006). Ozone was no longer
26 positively associated with mortality when the analysis was restricted to days with 24-hour 03
27 concentrations below 10 ppb. Given the relatively small number of days included in these
r\r
28 restricted analyses, especially for cut points of 20 ppb and below, statistical uncertainty is
29 increased.
30 Bell et al. (2006) also evaluated the shape of the concentration-response relationship
31 between 63 and mortality. Although the results of this analysis suggested the lack of threshold in
32 the Os-mortality relationship, the ISA noted that it is difficult to interpret such a curve because:
33 (1) there is uncertainty around the shape of the concentration-response curve at 24-hour average
26For example, Bell et al. (2006) reported that for analyses restricted to 24-hour O3 concentrations at or below 20
ppb, 73% of days were excluded on average across the 98 communities.
3-50
-------
1 Os concentrations generally below 20 ppb and (2) the concentration-response curve does not take
2 into consideration the heterogeneity in Os-mortality risk estimates across cities (U.S. EPA, 2013,
3 section 6.6.2.3).
4 Several additional studies have used the NMMAPS dataset to evaluate the concentration-
5 response relationship between short-term Oj, concentrations and mortality. For example, using
6 the same data as Bell et al. (2006), Smith et al. (2009) conducted a subset analysis, but instead of
7 restricting the analysis to days with O?, concentrations below a cutoff the authors only included
8 days above a defined cutoff. The results of this analysis were consistent with those reported by
9 Bell et al. (2006). Specifically, the authors reported consistent positive associations for all cutoff
10 concentrations up to concentrations where the total number of days available were so limited that
11 the variability around the central estimate was increased (U.S. EPA, 2013, section 6.6.2.3). In
12 addition, using NMMAPS data for 1987-1994 for Chicago, Pittsburgh, and El Paso, Xia and
13 Tong (2006) reported evidence for a threshold around a 24-hour average 63 concentration of
14 25 ppb, though the threshold values estimated in the analysis were sometimes in the range of
15 where data density was low (U.S. EPA, 2013, section 6.6.2.3). Stylianou and Nicolich (2009)
16 examined the existence of thresholds following an approach similar to Xia and Tong (2006)
17 using data from NMMAPS for nine major U.S. cities (i.e., Baltimore, Chicago, Dallas/Fort
18 Worth, Los Angeles, Miami, New York, Philadelphia, Pittsburgh, and Seattle) for the years
19 1987-2000. The authors reported that the estimated (Vmortality risks varied across the nine
20 cities, with the models exhibiting apparent thresholds in the 10-45 ppb range for Os (24-hour
21 average). However, given the city-to-city variation in risk estimates, combining the city-specific
22 estimates into an overall estimate complicated the interpretation of the results. Additional studies
23 in Europe, Canada, and Asia did not report the existence of a threshold (Katsouyanni et al.,
24 2009), with inconsistent and/or inconclusive results across cities, or a non-linear relationship in
25 the Os-mortality concentration-response curve (Wong et al., 2010).
26 3.1.2.4 Cardiovascular effects - Short-term Exposure
27 • To what extent does the currently available scientific evidence, including related
28 uncertainties, strengthen or alter our understanding from the last review of
29 cardiovascular effects attributable to short-term Os exposures?
30 A relatively small number of studies have examined the potential effect of short-term O^
31 exposure on the cardiovascular system. The 2006 63 AQCD (U.S. EPA, 2006, p.8-77) concluded
32 that "Os directly and/or indirectly contributes to cardiovascular-related morbidity" but added that
33 the body of evidence was limited. This conclusion was based on a controlled human exposure
34 study that included hypertensive adult males; a few epidemiologic studies of physiologic effects,
3-51
-------
1 heart rate variability, arrhythmias, myocardial infarctions, and hospital admissions; and
2 toxicological studies of heart rate, heart rhythm, and blood pressure.
3 More recently, the body of scientific evidence available that has examined the effect of
4 Oj, on the cardiovascular system has expanded. There is an emerging body of animal
5 toxicological evidence demonstrating that short-term exposure to Os can lead to autonomic
6 nervous system alterations (in heart rate and/or heart rate variability) and suggesting that
7 proinflammatory signals may mediate cardiovascular effects. Interactions of 63 with respiratory
8 tract components result in secondary oxidation product formation and subsequent production of
9 inflammatory mediators, which have the potential to penetrate the epithelial barrier and to initiate
10 toxic effects systemically. In addition, animal toxicological studies of long-term exposure to 63
11 provide evidence of enhanced atherosclerosis and ischemia/reperfusion (I/R) injury,
12 corresponding with development of a systemic oxidative, proinflammatory environment. Recent
13 experimental and epidemiologic studies have investigated Cb-related cardiovascular events and
14 are summarized in Section 6.3 of the ISA (U.S. EPA, 2013, Section 6.3). Overall, the ISA
15 summarized the evidence in this review as follows (U.S. EPA, 2013, p. 6-211).
16 In conclusion, animal toxicological studies demonstrate Os-induced
17 cardiovascular effects, and support the strong body of epidemiologic evidence
18 indicating O^-induced cardiovascular mortality. Animal toxicological and
19 controlled human exposure studies provide evidence for biologically plausible
20 mechanisms underlying these Os-induced cardiovascular effects. However, a lack
21 of coherence with epidemiologic studies of cardiovascular morbidity remains an
22 important uncertainty.
23 Animal toxicological studies support that short-term 63 exposure can lead to
24 cardiovascular morbidity. Animal studies provide evidence for both increased and decreased
25 heart rate (HR), however it is uncertain if (Vinduced reductions in HR are relevant to humans.
26 Animal studies also provide evidence for increased heart rate variability (HRV), arrhythmias,
27 vascular disease and injury following short-term Os exposure. In addition, a series of studies
28 highlight the role of genetic variability and age in the induction of effects and attenuation of
29 responses to Os exposure.
30 Biologically plausible mechanisms have been described for the cardiovascular effects
31 observed in animal exposure studies (U.S. EPA, 2013, Section 5.3.8). Evidence that
32 parasympathetic pathways may underlie cardiac effects is described in more detail in Section
33 5.3.2 of the ISA (U.S. EPA, 2013). Recent studies suggest that Os exposure may disrupt the
34 endothelin system that constricts blood vessels and increase blood pressure, which can result in
35 an increase in HR, HRV; and disrupt the NO system and the production of atrial natriuretic
36 factor (ANF), vasodilators that reduce blood pressure. Additionally, 63 may increase oxidative
37 stress and vascular inflammation promoting the progression of atherosclerosis and leading to
3-52
-------
1 increased susceptibility to I/R injury. As O^ reacts quickly with the ELF and does not translocate
2 to the heart and large vessels, studies suggest that the cardiovascular effects exhibited could be
3 caused by secondary oxidation products resulting from Os exposure. However, direct evidence of
4 translocation of 63 reaction products to the cardiovascular system has not been demonstrated in
5 vivo. Alternatively, extrapulmonary release of diffusible mediators (such as cytokines or
6 endothelins) may initiate or propagate inflammatory responses throughout the body leading to
7 the cardiovascular effects reported in toxicology studies. Ozone reacts within the lung to induce
8 pulmonary inflammation and the influx and activation of inflammatory cells, resulting in a
9 cascading proinflammatory state, and may lead to the extrapulmonary release of diffusible
10 mediators that could result in cardiovascular injury.
11 Controlled human exposures studies discussed in previous AQCDs have not
12 demonstrated any consistent extrapulmonary effects. In this review, evidence from controlled
13 human exposure studies suggests cardiovascular effects in response to short-term 63 exposure
14 (see ISA, U.S. EPA, 2013, Section 6.3.1) and provides some coherence with evidence from
15 animal toxicology studies. Controlled human exposure studies also support the animal
16 toxicological studies by demonstrating Os-induced effects on blood biomarkers of systemic
17 inflammation and oxidative stress, as well as changes in biomarkers that can indicate a
18 prothrombogenic response to O^. Increases and decreases in high frequency HRV have been
19 reported following relatively low (120 ppb during rest) and high (300 ppb with exercise) 63
20 exposures, respectively. These changes in cardiac function observed in animal and human studies
21 provide preliminary evidence for Os-induced modulation of the autonomic nervous system
22 through the activation of neural reflexes in the lung (see ISA,U.S. EPA 2013, Section 5.3.2).
23 Overall, the ISA concludes that the available body of epidemiologic evidence examining
24 the relationship between short-term exposures to O^ concentrations and cardiovascular morbidity
25 is inconsistent (U.S. EPA, 2013, Section 6.3.2.9). Across studies, different definitions, (i.e., ICD-
26 9 diagnostic codes) were used for both all-cause and cause-specific cardiovascular morbidity
27 (ISA, U.S. EPA, 2013, see Tables 6-35 to 6-39), which may contribute to inconsistency in
28 results. However, within diagnostic categories, no consistent pattern of association was found
29 with 03. Generally, the epidemiologic studies used nearest air monitors to assess 03
30 concentrations, with a few exceptions that used modeling or personal exposure monitors.
31 The inconsistencies in the associations observed between short-term Os and cardiovascular
32 disease (CVD) morbidities are unlikely to be explained by the different exposure assignment
33 methods used (see Section 4.6, ISA, U.S. EPA 2013). The wide variety of biomarkers considered
34 and the lack of consistency among definitions used for specific cardiovascular disease endpoints
35 (e.g., arrhythmias, HRV) make comparisons across studies difficult.
3-53
-------
1 Despite the inconsistent evidence for an association between O?, concentration and CVD
2 morbidity, mortality studies indicate a consistent positive association between short-term 63
3 exposure and cardiovascular mortality in multicity studies and in a multicontinent study. When
4 examining mortality due to cardiovascular disease, epidemiologic studies consistently observe
5 positive associations with short-term exposure to O?,. Additionally, there is some evidence for an
6 association between long-term exposure to O?, and mortality, although the association between
7 long-term ambient 63 concentrations and cardiovascular mortality can be confounded by other
8 pollutants as evident by a study of cardiovascular mortality that reported no association after
9 adjustment for PM2.5 concentrations. The ISA (U.S. EPA 2013, section 6.3.4) states that taken
10 together, the overall body of evidence across the animal and human studies is sufficient to
11 conclude that there is likely to be a causal reationship between relevant short-term exposures to
12 O3 and cardiovascular system effects.
13 3.1.3 Adversity of Effects
14 In this section we address the following question:
15* To what extent does the currently available scientific evidence expand our
16 understanding of the adversity of Os-related health effects?
17 In making judgments as to when various (Vrelated effects become regarded as adverse
18 to the health of individuals, in previous NAAQS reviews staff has relied upon the guidelines
19 published by the American Thoracic Society (ATS) and the advice of CAS AC. In 2000, the ATS
20 published an official statement on "What Constitutes an Adverse Health Effect of Air
21 Pollution?" (ATS, 2000), which updated and built upon its earlier guidance (ATS, 1985). The
22 earlier guidance defined adverse respiratory health effects as "medically significant physiologic
23 changes generally evidenced by one or more of the following: (1) interference with the normal
24 activity of the affected person or persons, (2) episodic respiratory illness, (3) incapacitating
25 illness, (4) permanent respiratory injury, and/or (5) progressive respiratory dysfunction", while
26 recognizing that perceptions of "medical significance" and "normal activity" may differ among
27 physicians, lung physiologists and experimental subjects (ATS, 1985). The 2000 ATS guidance
28 builds upon and expands the 1985 definition of adversity in several ways. The guidance
29 concludes that transient, reversible loss of lung function in combination with respiratory
30 symptoms should be considered adverse. There is also a more specific consideration of
31 population risk (ATS, 2000). Exposure to air pollution that increases the risk of an adverse effect
32 to the entire population is adverse, even though it may not increase the risk of any individual to
33 an unacceptable level. For example, a population of asthmatics could have a distribution of lung
34 function such that no individual has a level associated with significant impairment. Exposure to
35 air pollution could shift the distribution to lower levels that still do not bring any individual to a
3-54
-------
1 level that is associated with clinically relevant effects. However, this would be considered to be
2 adverse because individuals within the population would have diminished reserve function, and
3 therefore would be at increased risk to further environmental insult (U.S. EPA, 2013, p. Ixxi; and
4 SO2 NAAQS review, 75 FR at 35526/2, June 22, 2010).
5 The ATS also concluded that elevations of biomarkers such as cell types, cytokines and
6 reactive oxygen species may signal risk for ongoing injury and more serious effects or may
7 simply represent transient responses, illustrating the lack of clear boundaries that separate
8 adverse from nonadverse events. More subtle health outcomes also may be connected
9 mechanistically to health effects that are clearly adverse, so that small changes in physiological
10 measures may not appear clearly adverse when considered alone, but may be part of a coherent
11 and biologically plausible chain of related health outcomes that include responses that are clearly
12 adverse, such as mortality (section 3.1.2.1, above).
13 In this review, the new evidence provides further support for relationships between Os
14 exposures and a spectrum of health effects, including effects that meet the ATS criteria for being
15 adverse (ATS, 1985 and 2000). The ISA judgment that there is a causal relationship between
16 short-term Os exposure and a full range of respiratory effects, including respiratory morbidity
17 (e.g., lung function decrements, respiratory symptoms, inflammation, hospital admissions, and
18 emergency department visits) and mortality, provides support for concluding that short-term 63
19 exposure is associated with adverse effects (U.S. EPA 2013, section 2.5.2). Overall, including
20 new evidence of cardiovascular system effects, the evidence supporting an association between
21 short-term Os exposures and total (non-accidental, cardiopulmonary) respiratory mortality is
22 stronger in this review (U.S. EPA 2013, section 2.5.2). And the judgment of likely causal
23 associations between long-term measures of 63 exposure and respiratory effects such as new-
24 onset asthma, prevalence of asthma, asthma symptoms and control, and asthma hospital
25 admissions provides support for concluding that long-term 63 exposure is associated with
26 adverse effects ranging from episodic respiratory illness to permanent respiratory injury or
27 progressive respiratory decline (U.S. EPA 2013, section 7.2.8).
28 This review provides additional evidence of (Vattributable effects that are clearly
29 adverse, including premature mortality. Application of the ATS guidelines to the least serious
30 category of effects related to ambient 03 exposures, which are also the most numerous and
31 therefore are also important from a public health perspective, involves judgments about which
32 medical experts on CASAC panels and public commenters have in the past expressed diverse
33 views. To help frame such judgments, EPA staff defined gradations of individual functional
34 responses (e.g., decrements in FEVi and airway responsiveness) and symptomatic responses
35 (e.g., cough, chest pain, wheeze), together with judgments as to the potential impact on
36 individuals experiencing varying degrees of severity of these responses. These gradations were
3-55
-------
1 used in the 1997 O3 NAAQS review and slightly revised in the 2008 review (U.S. EPA, 1996, p.
2 59; 2007, p.3-72; 72 FR 37849, July 11, 2007). These gradations and impacts are summarized in
3 Tables 3-2 and 3-3 in the 2007 O3 Staff Paper (U.S. EPA, 2007, pp.3-74 to 3-75).
4 For active healthy people, including children, moderate levels of functional responses
5 (e.g., FEVi decrements of > 10% but < 20%, lasting 4 to 24 hours) and/or moderate symptomatic
6 responses (e.g., frequent spontaneous cough, marked discomfort on exercise or deep breath,
7 lasting 4 to 24 hours) would likely interfere with normal activity for relatively few sensitive
8 individuals (U.S. EPA, 2007, p.3-72; 72 FR 37849, July 11, 2007); whereas large functional
9 responses (e.g., FEVi decrements > 20%, lasting longer than 24 hours) and/or severe
10 symptomatic responses (e.g., persistent uncontrollable cough, severe discomfort on exercise or
11 deep breath, lasting longer than 24 hours) would likely interfere with normal activities for many
12 sensitive individuals (U.S. EPA, 2007, p.3-72; 72 FR 37849, July 11, 2007) and therefore would
13 be considered adverse under ATS guidelines. For the purpose of estimating potentially adverse
14 lung function decrements in active healthy people in the 2008 O3 NAAQS review, the CASAC
15 panel for that review indicated that a focus on the mid to upper end of the range of moderate
16 levels of functional responses is most appropriate (e.g., FEVi decrements > 15% but < 20%)
17 (Henderson, 2006; 2007 Staff Paper, p. 3-76). However, for children and adults with lung
18 disease, even moderate functional (e.g., FEVi decrements > 10% but < 20%, lasting up to 24
19 hours) or symptomatic responses (e.g., frequent spontaneous cough, marked discomfort on
20 exercise or with deep breath, wheeze accompanied by shortness of breath, lasting up to 24 hours)
21 would likely interfere with normal activity for many individuals, and would likely result in
22 additional and more frequent use of medication (U.S. EPA, 2007, p.3-72; 72 FR 37849, July 11,
23 2007). For people with lung disease, large functional responses (e.g., FEVi decrements > 20%,
24 lasting longer than 24 hours) and/or severe symptomatic responses (e.g., persistent
25 uncontrollable cough, severe discomfort on exercise or deep breath, persistent wheeze
26 accompanied by shortness of breath, lasting longer than 24 hours) would likely interfere with
27 normal activity for most individuals and would increase the likelihood that these individuals
28 would seek medical treatment (U.S. EPA, 2007, p.3-72; 72 FR 37849, July 11, 2007). In the last
29 O3 NAAQS review, for the purpose of estimating potentially adverse lung function decrements
30 in people with lung disease the CASAC panel indicated that a focus on the lower end of the
31 range of moderate levels of functional responses is most appropriate (e.g., FEVi decrements
32 >10%) (Henderson, 2006; 2007 Staff Paper, p. 3-76). In addition, in the reconsideration of the
33 2008 final decision, CASAC stated that "[a] 10% decrement in FEVI can lead to respiratory
34 symptoms, especially in individuals with pre-existing pulmonary or cardiac disease. For
35 example, people with chronic obstructive pulmonary disease have decreased ventilatory reserve
3-56
-------
1 (i.e., decreased baseline FEV1) such that a >10% decrement could lead to moderate to severe
2 respiratory symptoms" (Samet, 2011) (section 3.1.2.1, above).
3 In judging the extent to which these impacts represent effects that should be regarded as
4 adverse to the health status of individuals, in previous NAAQS reviews we also considered
5 whether effects were experienced repeatedly during the course of a year or only on a single
6 occasion (Staff Paper, U.S. EPA, 2007). Although some experts would judge single occurrences
7 of moderate responses to be a "nuisance," especially for healthy individuals, a more general
8 consensus view of the adversity of such moderate responses emerges as the frequency of
9 occurrence increases. Thus it has been judged that repeated occurrences of moderate responses,
10 even in otherwise healthy individuals, may be considered to be adverse since they could well set
11 the stage for more serious illness (61 FR 65723). The CAS AC panel in the 1997 NAAQS review
12 expressed a consensus view that these "criteria for the determination of an adverse physiological
13 response were reasonable" (Wolff, 1995). In the review completed in 2008, estimates of repeated
14 occurrences continued to be an important public health policy factor in judging the adversity of
15 moderate lung function decrements in healthy and asthmatic people (72 FR 37850, July 11,
16 2007).
17 Evidence new to this review indicates that 6.6-hour exposures to 60 ppb 63 during
18 moderate exertion can result in pulmonary inflammation in healthy adults. As discussed in
19 section 3.1.2 above, the initiation of inflammation can be considered as evidence that injury has
20 occurred. Inflammation induced by a single 63 exposure can resolve entirely, but continued
21 acute inflammation can evolve into a chronic inflammatory state (ISA, U.S. EPA, 2013, p. 6-76),
22 which is clearly adverse. Therefore, like moderate lung function decrements, whether
23 inflammation is experienced repeatedly during the course of a year or only on a single occasion
24 is judged by staff to be reasonable criteria for determining adverse inflammatory effects
25 attributable to 63 exposures at 60 ppb.
26 Responses measured in controlled human exposure studies indicate that the range of
27 effects elicited in humans exposed to ambient 63 concentrations include: decreased inspiratory
28 capacity; mild bronchoconstriction; rapid, shallow breathing pattern during exercise; and
29 symptoms of cough and pain on deep inspiration (EPA, 2013, section 6.2.1.1). Some young,
30 healthy adults exposed to 63 concentrations > 60 ppb, while engaged in 6.6 hours of intermittent
31 moderate exertion, develop statistically significant reversible, transient decrements in lung
32 function, symptoms of breathing discomfort, and inflammation if minute ventilation or duration
33 of exposure is increased sufficiently (EPA, 2013, section 6.2.1.1). Among healthy subjects there
34 is considerable interindividual variability in the magnitude of the FEVi responses, but averaged
35 across studies at 60 ppb (EPA, 2013, pp. 6-17 to 6-18), 10% of healthy subjects had >10% FEVi
36 decrements. The combination of lung function decrements and respiratory symptoms, which has
3-57
-------
1 been considered adverse in previous reviews, has been demonstrated in healthy adults following
2 prolonged (6.6 hour) exposures, while at intermittent moderate exertion, to 70 ppb. For these
3 types of effects, information from controlled human exposure studies, which provides an
4 indication of the magnitude and thus adversity of effects at different 63 concentrations,
5 combined with estimates of occurrences in the population from the HREA, provide information
6 about their importance from a policy perspective.
7 3.1.4 Ozone Concentrations Associated With Health Effects
8 In evaluating 63 exposure concentrations reported to result in health effects, within the
9 context of the adequacy of the current standard, we first consider the following specific question:
10 • To what extent does the currently available scientific evidence indicate morbidity
11 and/or mortality attributable to exposures to Os concentrations lower than
12 previously reported or that would meet the current standard?
13 In addressing this question, we characterize the extent to which (Vattributable effects have been
14 reported over the ranges of 03 exposure concentrations evaluated in controlled human exposure
15 studies and over the distributions of ambient 63 concentrations in locations where epidemiologic
16 studies have been conducted.
17 3.1.4.1 Concentrations in Controlled Human Exposure Studies and in Epidemiologic
18 Panel Studies
19 In considering what the currently available evidence indicates with regard to effects
20 associated with exposure concentrations lower than those identified in the last review, or that
21 could meet the current standard, we first consider the evidence from controlled human exposure
22 studies and epidemiologic panel studies. This evidence is assessed in section 6.2 of the ISA and
23 is summarized in section 3.1.2 above. Controlled human exposure studies have generally been
24 conducted with young, healthy adults, and have evaluated exposure durations less than 8 hours.
25 Epidemiologic panel studies have evaluated a wider range of study populations, including
26 children, and have generally evaluated associations with 63 concentrations averaged over several
27 hours (U.S. EPA, 2013, section 6.2.1.2).27
28 As summarized above (section 3.1.2.1), and as discussed in detail in the ISA (U.S. EPA,
29 2013, section 6.2), a large number of controlled human exposure studies have reported lung
30 function decrements, respiratory symptoms, airway inflammation, airway hyperresponsiveness,
31 and/or impaired lung host defense in young, healthy adults engaged in moderate, intermittent
27
In this section we focus on panel studies that used on-site monitoring, and that are highlighted in the ISA for the
extent to which monitored ambient O3 concentrations reflect exposure concentrations in their study populations
(U.S. EPA, 2013, section 6.2.1.2).
3-58
-------
1 exertion, following 6.6-hour 03 exposures. These studies have consistently reported such effects
2 following exposures to Oj concentrations of 80 ppb or greater. Available studies have also
3 evaluated some of these effects (i.e., lung function decrements, respiratory symptoms, airway
4 inflammation) following exposures to Os concentrations below 75 ppb. Table 3-1 highlights the
5 group mean results of individual controlled human exposure studies that have evaluated
6 exposures of healthy adults to O^ concentrations below 75 ppb.
7 Table 3-1.
9
10
11
12
13
14
Group mean results of controlled human exposure studies that have evaluated
exposures to ozone concentrations below 75 ppb in young, healthy adults.
Endpoint
FEVi decrements
O3 Exposure
Concentration
70 ppb
60 ppb
40 ppb
Study
Schelegle et al., 2009
Kim etal., 2011
Schelegle et al., 2009
Adams, 2006
Adams, 2002
Adams, 2006
Adams, 2002
Statistically
Significant O3-
Induced
Effect28
yes
yes
no
29
yes
no
no
no
Respiratory
Symptoms
70 ppb
60 ppb
40 ppb
Schelegle et al., 2009
Kim etal., 2011
Schelegle et al., 2009
Adams, 2006
Adams, 2006
Adams, 2002
yes
no
no
no30
no
no
Airway
Inflammation
(neutrophil influx)
60 ppb
Kim etal., 2011
yes
In further evaluating Os-induced FEVi decrements following exposures to 63
concentrations below 75 ppb, the ISA also combined the individual data from multiple studies of
healthy adults exposed for 6.6 hours to 60 ppb O^ (Kim et al., 2011; Brown et al., 2008; Adams,
2006a, 2002, 1998). Based on these data, the ISA reports that 10% of exposed subjects
experienced FEVi decrements of 10% or more (i.e., abnormal and large enough to be potentially
Based on study population means.
29 In an analysis of the Adams (2006) data for square-wave chamber exposures, even after removal of potential
outliers, Brown et al. (2008) reported the average effect on FEVI at 60 ppb to be statistically significant (p < 0.002)
using several common statistical tests (U.S. EPA, 2013, section 6.2.1.1) (section 3.1.2.1, above).
30Adams (2006) reported increased respiratory symptoms during a 6.6 hour exposure protocol with an average O3
exposure concentration of 60 ppb. The increase in symptoms was reported to be statistically different from initial
respiratory symptoms, though not statistically different from filtered air controls.
3-59
-------
1 adverse for people with pulmonary disease, based on past CASAC advice (section 3.1.3,
2 above))31 (U.S. EPA, 2013, section 6.2.1.1). Consistent with these findings, recently developed
3 empirical models predict that the onset of Os-induced FEVi decrements in healthy adults occurs
4 following exposures to 60 ppb 63 for 4 to 5 hours while at moderate, intermittent exertion
5 (Schelegle et al., 2012), and that 9% of healthy adults exposed to 60 ppb Os for 6.6 hours would
6 experience FEVi decrements greater than or equal to 10% (McDonnell et al., 2012) (U.S. EPA,
7 2013, section 6.2.1.1). When the evidence for Os-induced lung function decrements was taken
8 together, the ISA concluded that (1) "mean FEVi is clearly decreased by 6.6-h exposures to 60
9 ppb 63 and higher concentrations in subjects performing moderate exercise" (U.S. EPA, 2013, p.
10 6-9) and (2) although group mean decrements following exposures to 60 ppb 63 are biologically
11 small, "a considerable fraction of exposed individuals experience clinically meaningful
12 decrements in lung function" (U.S. EPA, 2013, p. 6-20).
13 In considering the specific question above, we note that controlled human exposure
14 studies have reported decreased lung function, increased airway inflammation, and increased
15 respiratory symptoms in healthy adults following exposures to 63 concentrations below 75 ppb.
16 Such impairments in respiratory function have the potential to be adverse, based on ATS
17 guidelines for adversity and based on previous advice from CASAC (section 3.1.3, above). In
18 addition, if they become serious enough, these respiratory effects could lead to the types of
19 clearly adverse effects commonly reported in 03 epidemiologic studies (e.g., respiratory
20 emergency department visits, hospital admissions). Therefore, following exposures to 63
21 concentrations lower than 75 ppb, controlled human exposure studies have reported respiratory
22 effects that could be adverse in some individuals, particularly if experienced by members of at-
23 risk populations (e.g., asthmatics, children).32
24 In further considering effects following exposures to Os concentrations below 75 ppb, we
25 also note that the ISA highlights some epidemiologic panel studies for the extent to which
26 monitored ambient 63 concentrations reflect exposure concentrations in their study populations
27 (U.S. EPA, 2013, section 6.2.1.2). Specifically, Table 3-2 below includes O3 panel studies that
28 have evaluated associations with lung function decrements for 63 concentrations at or below 75
29 ppb, and that measured Os concentrations with monitors located in the areas where study
30 subjects were active (e.g., on site at summer camps or in locations where exercise took place)
31CASAC has previously stated that "[a] 10% decrement in FEVI can lead to respiratory symptoms, especially in
individuals with pre-existing pulmonary or cardiac disease. For example, people with chronic obstructive pulmonary
disease have decreased ventilatory reserve (i.e., decreased baseline FEVI) such that a >10% decrement could lead to
moderate to severe respiratory symptoms" (Samet, 2011) (section 3.1.3, above).
32These effects were reported in healthy individuals. It is thus a reasonable inference that the effects would be
greater in magnitude and potential severity for at-risk groups. See National Environmental Development Ass 'n
Clean Air Project v. EPA, 686 F. 3d 803, 811 (D.C. Cir. (2012) (making this point).
3-60
-------
1 (U.S. EPA, 2013, section 6.2.1.2 and Table 6-6). Epidemiologic panel studies have evaluated a
2 wider range of populations and lifestages than controlled human exposure studies of 63
3 concentrations below 75 ppb (e.g., including children).
4
5
Table 3-2. Panel studies of lung function decrements with analyses restricted to
concentrations below 75 ppb.
Study
Spektor et al.
(1988)
Chan and
Wu (2005)
Korrick et al.
(1998)
Brauer et al.
(1996)
Brunekreef
etal. (1994)
Population
Children at
summer camp
Mail carriers
Adult hikers
Farm workers
Exercising
adults
O3 Concentrations
Restricted to 1 -hour concentrations
below 60 ppb
Maximum 8-hour average was 65
ppb
2- to 12-hour average from 40 to
74 ppb during hikes
Restricted to 1 -hour maximum
below 40 ppb
Restricted to 1 -hour maximum
below 30 ppb
Restricted to 10-minute to 2.4-hour
averages below 61 ppb
Restricted to 10-minute to 2.4-hour
averages below 5 1 ppb
Restricted to 10-minute to 2. 4 -hour
averages below 41 ppb
Statistically Significant Association
with Lung Function Decrements
Yes
Yes
Yes
Yes
No
No
No
No
6
7
8
9
10
11
12
13
14
15
16
17
18
19
Although these studies report health effect associations for different averaging times, and it is not
clear the extent to which specific O^ exposure conditions (i.e., concentrations, durations of
exposure, degrees of activity) were responsible for eliciting reported decrements, they are
consistent with the findings of the controlled human exposure studies discussed above.
Specifically, the epidemiologic panel studies in Table 3-2 indicate C^-associated lung function
decrements when on-site monitored concentrations (ranging from minutes to hours) were below
75 ppb, with the evidence becoming less consistent at lower Os concentrations.
3.1.4.2 Concentrations in Epidemiologic Studies - Short-term Metrics
We next consider distributions of ambient O^ concentrations in locations where
epidemiologic studies have evaluated (^-associated hospital admissions, emergency department
visits, and/or mortality. When considering epidemiologic studies within the context of the current
standard, we emphasize those studies conducted in the U.S. and Canada. Such studies reflect air
quality and exposure patterns that are likely more typical of the U.S. population than the air
quality and exposure patterns reflected in studies conducted outside the U.S. and Canada (section
3-61
-------
1 1.3.1.2, above).33 We also emphasize studies reporting associations with effects judged in the
2 ISA to be robust to confounding by other factors, including co-occurring air pollutants. In
3 addition to these factors, we consider the statistical precision of study results, the extent to which
4 studies report associations in at-risk populations, and the extent to which the biological
5 plausibility of associations at various ambient Os concentrations is supported by controlled
6 human exposure and/or animal toxicological studies. These considerations help inform the range
7 of ambient 63 concentrations over which we have the most confidence in (Vassociated health
8 effects, and the range of concentrations over which our confidence in such associations is
9 appreciably lower. We place particular emphasis on characterizing those portions of distributions
10 of ambient 63 concentrations likely to meet the current standard.
11 In our consideration of these issues, we first address the following question:
12 • To what extent have U.S. and Canadian epidemiologic studies reported
13 associations with mortality or morbidity in locations that would have met the
14 current Os standard during the study period?
15 Addressing this question can provide important insights into the extent to which Os-health effect
16 associations are present for distributions of ambient Os concentrations that would be allowed by
17 the current standard. To the extent O^ health effect associations are reported in study areas that
18 would have met the current standard, we have greater confidence that the current standard could
19 allow the clearly adverse Cb-associated effects indicated by those studies (e.g., mortality,
20 hospital admissions, emergency department visits).34
21 We note that epidemiologic studies evaluate statistical associations between variation in
22 the incidence of health outcomes and variation in ambient Os concentrations. In many of the O?,
23 epidemiologic studies assessed in the ISA, ambient concentrations are averaged across multiple
24 monitors within study areas, and in some cases over multiple days. These averages are used as
25 surrogates for the spatial and temporal patterns of O?, exposures in study populations. In this
26 second draft PA, we refer to these averaged concentrations as "area-wide" 63 concentrations.
27 We recognize that these epidemiologic studies do not identify the Os exposures that
28 population members have experienced, and do not identify the exposures that may be eliciting
29 the observed health outcomes. Thus, in considering epidemiologic studies of mortality and
30 morbidity, we are not drawing conclusions regarding single short-duration O?, concentrations in
31 ambient air that, alone, are eliciting the reported health outcomes. Rather, our focus in this
32 section is to consider what these studies convey regarding the extent to which health effects may
33Nonetheless, we recognize the importance of all studies, including international studies, in the ISA's assessment of
the weight of the evidence that informs causality determinations.
34
See ATA III, 283 F.3d at 370 (EPA justified in revising NAAQS when health effect associations are observed at
levels allowed by the NAAQS).
3-62
-------
1 be occurring (i.e., as indicated by associations) under air quality conditions meeting the current
2 standard.
3 In order to facilitate consideration of the question above, we have identified U.S. and
4 Canadian studies of respiratory hospital admissions, respiratory emergency department visits,35
5 and mortality (total, respiratory, cardiovascular) from the ISA (studies identified from U.S. EPA,
6 2013, Table 6-28, section 6.2.8, and Table 6-42) (Appendix 3-D). For each monitor in the areas
7 evaluated by these studies, we have identified the 3-year averages of the annual 4th highest daily
8 maximum 8-hour Os concentrations (Appendix 3-D).36 To provide perspective on whether study
9 cities would have met or violated the current 63 NAAQS during the study period, these 63
10 concentrations were compared to the level of the current standard. Based on this approach, a
11 study city was judged to have met the current standard during the study period if all of the 3-year
12 averages of annual 4th highest 8-hour 63 concentrations in that area were below 75 ppb.
13 Based on these analyses, the large majority of epidemiologic study areas evaluated would
14 have violated the current standard during study periods (Appendix 3-D). Table 3-3 below
15 highlights the subset of U.S. and Canadian studies reporting 63 health effect associations in
16 locations that would have met the current standard during study periods. This includes a U.S.
17 single-city study that would have met the current standard over the entire study period (Mar and
18 Koenig, 2009) and four Canadian multicity studies for which the majority of study cities would
19 have met the current standard over the entire study periods (Cakmak et al., 2006; Dales et al.,
20 2006; Katsouyanni et al., 2009; Stieb et al., 2009).37
21
35Given the inconsistency in results across cardiovascular morbidity studies (U.S. EPA, 2013, section 6.3.2.9), our
consideration of the morbidity evidence in this section focuses on studies of respiratory hospital admissions and
emergency department visits.
36These concentrations are referred to as "design values." A design value is a statistic that is calculated at individual
monitors and based on 3 consecutive years of data collected from that site. In the case of O3, the design value for a
monitor is based on the 3-year average of the annual 4th highest daily maximum 8-hour O3 concentration in parts
per billion (ppb). For U.S. study areas, we used EPA's Air Quality System (AQS)
(http://www.epa.gov/ttn/airs/airsaqs/) to identify design values. For Canadian study areas, we used publically
available air quality data from the Environment Canada National Air Pollution Surveillance Network
(http://www.etc-cte.ec.gc.ca/napsdata/main.aspx). We followed the data handling protocols for calculating design
values as detailed in 40 CFR Part 50, Appendix P.
37In addition, a study by Vedal et al. (2003) was included in the 2006 CD (U.S. EPA, 2006). This study reported
positive and statistically significant associations with mortality in Vancouver during a time period when the study
area would have met the current standard (U.S. EPA, 2007). This study was not highlighted in the ISA in the current
review (U.S. EPA, 2013).
3-63
-------
1
2
Table 3-3. U.S. and Canadian epidemiologic studies reporting Os health effect
associations in locations that would have met the current standard during
study periods.
Authors
Cakmak et al.
(2006)
Dales et al.
(2006)
Katsouyanni et
al. (2009)
Katsouyanni et
al. (2009)
Mar and
Koenig (2009)
Stieb et al.
(2009)
Study Results
Positive and statistically significant association
with respiratory hospital admissions
Positive and statistically significant association
with respiratory hospital admissions
Positive and statistically significant associations
with respiratory hospital admissions
Positive and statistically significant associations
with all -cause and cardiovascular mortality38
Positive and statistically significant associations
with asthma emergency department visits in
children (< 18 years) and adults (> 18 years)
Positive and statistically significant association
with asthma emergency department visits
Cities
10
Canadian
cities
11
Canadian
cities
12
Canadian
cities
12
Canadian
cities
Seattle
7 Canadian
cities
Number of cities meeting
the current standard over
entire study period
7
7
10
8
1
5
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
As illustrated in Table 3-3, one U.S. single-city study highlighted in the ISA has reported
health effect associations with asthma emergency department visits in a location that would have
met the current standard over the entire study period. In addition, four Canadian multicity studies
reported associations with mortality or morbidity when the majority of study locations would
have met the current standard over the entire study periods. While there is uncertainty in
ascribing the multicity effect estimates reported in these Canadian studies entirely to ambient
concentrations that would have met the current standard (i.e., given that some study locations
would have violated the current standard over at least part of the study period), the information
in Table 3-3 suggests that reported multicity effect estimates are largely influenced by locations
meeting the current standard. Together, these U.S. and Canadian epidemiologic studies suggest a
relatively high degree of confidence in the presence of associations with mortality and morbidity
for ambient 63 concentrations meeting the current standard.
We next consider the extent to which additional epidemiologic studies of mortality or
morbidity (i.e., those conducted in locations that violated the current standard) can also inform
our consideration of adequacy of the current standard. In doing so, we note that health effect
38Katsouyanni et al. (2009) report a positive and statistically significant association with cardiovascular mortality for
people aged 75 years or older.
3-64
-------
1 associations reported in epidemiologic studies are influenced by the full distributions of ambient
2 63 concentrations, including concentrations below the level of the current standard. We focus on
3 studies that have explicitly characterized such O^ health effect associations, including confidence
4 in those associations, for various portions of distributions of ambient 63 concentrations. In doing
5 so, we consider the following question:
6 • To what extent do analyses from epidemiologic studies indicate confidence in health
7 effect associations over distributions of ambient Os concentrations, including at
8 concentrations lower than previously identified or below the current standard?
9 We first focus on those studies that have reported confidence intervals around
10 concentration-response functions over distributions of ambient 63 concentrations. Confidence
11 intervals around concentration-response functions can provide insights into the range of ambient
12 concentrations over which the study indicates the most confidence in the reported health effect
13 associations (i.e., where confidence intervals are narrowest), and into the range of ambient
14 concentrations below which the study indicates that uncertainty in the nature of such associations
15 becomes notably greater (i.e., where confidence intervals become markedly wider). The
16 concentrations below which confidence intervals become markedly wider in such analyses are
17 intrinsically related to data density, and do not necessarily indicate the absence of an association.
18 The ISA identifies several epidemiologic studies that have reported confidence intervals
19 around concentration-response functions in U.S. cities. The ISA concludes that studies generally
20 indicate a linear concentration-response relationship "across the range of 8-h max and 24 h avg
21 O3 concentrations most commonly observed in the U.S. during the O3 season" and that "there is
22 less certainty in the shape of the C-R curve at the lower end of the distribution of 03
23 concentrations" (U.S. EPA, 2013, pp. 2-32 to 2-34). In characterizing the 63 concentrations
24 below which such certainty decreases, the ISA discusses area-wide Os concentrations as low as
25 20 ppb and as high as 40 ppb (U.S. EPA, 2013, section 2.5.4.4).
26 Consistent with these conclusions, the range of ambient concentrations over which the
27 evidence indicates the most certainty in concentration-response relationships can vary across
28 studies. Such variation is likely due at least in part to differences in the 63 metrics evaluated and
29 differences in the distributions of ambient concentrations and health events. Thus, although
30 consideration of confidence intervals around concentration-response functions can provide
31 valuable insights into the ranges of ambient concentrations over which studies indicate the most
32 confidence in reported health effect associations, there are limitations in the extent to which
33 these analyses can be generalized across 03 metrics, study locations, study populations, and
34 health endpoints.
3-65
-------
1 The ISA emphasizes two U.S. single-city studies that have reported confidence intervals
2 around concentration-response functions (Silverman and Ito, 2010; Strickland et al., 2010).
3 These studies, and their associated O^ air quality, are discussed below.
4 Silverman and Ito (2010) evaluated associations between 2-day rolling average 63
5 concentrations39 and asthma hospital admissions in New York City from 1999 to 2006 (a time
6 period when the study area would have violated the current standard, Appendix 3-D). As part of
7 their analysis, the authors evaluated the shape of the concentration-response relationship for 63
8 using a co-pollutant model that included PM2.5 (reprinted in Figure 3-4, below). Based on their
9 analyses, Silverman and Ito (2010) concluded a linear relationship between 63 and hospital
10 admissions is a reasonable approximation of the concentration-response function throughout
11 much of the range of ambient Os concentrations. Based on visual inspection of Figure 3-4 below
12 (Figure 3 from published study), we note that confidence in the reported concentration-response
13 relationship is highest for area-wide average OT, concentrations around 40 ppb (i.e., near the
14 reported median of 41 ppb), and decreases notably for concentrations at and below about 20 ppb.
15
39
2-day rolling averages of 8-hour daily maximum O3 concentrations were calculated throughout the study period,
averaged across study monitors.
3-66
-------
Ozone: All Ages
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
<\J
°2
be
^
cu
DC
o>
o
Hill
20 40 60 80
03 Concentration (ppb)
100
Figure 3-4. Concentration-response function for asthma hospital admissions over the
distribution of area-wide averaged Os concentrations (adapted from Silverman
and Ito, 2010).40
In considering the concentration-response function presented by Silverman and Ito (2010)
within the context of the adequacy of the current standard, we evaluate the extent to which the O^
concentrations contributing to various portions of the function would likely have been allowed
by the current standard. In doing so, we recognize that true design values cannot be identified for
the subsets of air quality data contributing to various portions of the concentration-response
function.41 Therefore, to use this analysis to inform our consideration of the adequacy of the
current standard we evaluate the extent to which the concentration-response function indicates a
relatively high degree of confidence in the reported health effect association on days when all
monitored 8-hour 63 concentrations were below 75 ppb (Table 3-4, below). This approach can
provide insight into the extent to which the reported O^ health effect association is present when
all monitored 63 concentrations are below the level of the current standard.
Based on the information in Table 3-4 below, when 2-day averaged O?, concentrations
ranged from 36 to 45 ppb (i.e., around the median, where confidence intervals are narrowest),
there was 1 day (out of 236) with at least one monitor recording an 8-hour daily maximum 63
40This figure was also reprinted in the ISA (U.S. EPA, 2013; Figure 6-16).
41
As discussed above, O3 design values are calculated using all data available from a monitor.
3-67
-------
2
O
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
concentration above the level of the current standard (approximately 0.4% of days). When 2-day
averaged 63 concentrations ranged from 26 to 45 ppb (i.e., extending to concentrations below the
median, but still above the concentrations where confidence intervals widen notably), there were
4 days (out of 816) with at least one monitor recording an 8-hour daily maximum Os
concentration above the level of the current standard (approximately 0.5% of days). Thus, on
over 99% of the days when area-wide "averaged" Os concentrations were between 26 and 45
ppb, the highest 8-hour daily maximum 63 concentrations were below 75 ppb. For comparison,
the annual 4* highest 8-hour daily maximum Os concentration generally corresponds to the 98*
or 99th percentile of the seasonal distribution, depending on the length of the 63 season.
Table 3-4. Distributions of daily 8-hour maximum ozone concentrations from highest
monitors over range of 2-day moving averages from composite monitors (for
study area evaluated by Silverman and Ito, 2010)
2-day moving average across monitors (ppb)
8-hr max from
st monitor
those 2 days
Min
5th
25th
50th
75th
95th
98th
99th
Max
11 to 20
(62 days)
16
20
26
29
35
39
41
42
42
21 to 25
(92 days)
27
28
31
35
38
50
54
55
59
26 to 30
(178 days)
30
34
37
42
46
54
60
67
80
31 to 35
(206 days)
36
39
43
46
50
60
70
72
75
36 to 40
(236 days)
41
44
47
51
55
63
68
71
79
41 to 45
(196 days)
45
49
53
59
64
74
82
87
100
46 to 50
(153 days)
52
56
61
64
69
80
85
87
97
51 to 55
(111 days) (
58
61
67
71
78
85
93
94
96
56 to(
71 da^
62
67
72
76
85
96
99
103
108
Days > 75 ppb
15
20
In a separate study, Strickland et al. (2010) evaluated associations between 3-day rolling
j\r\
average Oj, concentrations and asthma hospital admissions in Atlanta during the warm season
from 1994 to 2004 (a time period when the study area would have violated the current standard,
Appendix 3-D). As part of this analysis, Strickland et al. (2010) evaluated the concentration-
response relationship for Os and pediatric asthma emergency department visits. The authors
reported the shape of the concentration-response function to be approximately linear with no
evidence of a threshold when 3-day averaged 8-hour daily maximum Os concentrations were
approximately 30 to 80 ppb (Figure 3-5 below and U.S. EPA, 2013, Figure 6-18). Figure 3-5
below illustrates that the confidence intervals around the concentration-response function are
42
Three-day rolling averages of population-weighted 8-hour daily maximum O3 concentrations were calculated
throughout the study period (Strickland et al., 2010).
3-68
-------
1
2
4
5
6
9
10
11
12
13
14
15
16
17
narrowest around the study mean (i.e., 55 ppb), and that these confidence intervals do not widen
notably for "averaged" 63 concentrations as low as about 30 ppb.
Ozone Warm Season
LT>
CM -
in
o ^J
'•4-1
CO
0)
4-1
(0 LD
a: q
ID
O) -
30 40 50 60 70
Concentration (ppb)
80
Figure 3-5. Concentration-response function for pediatric asthma emergency department
visits over the distribution of averaged, population-weighted 8-hour Os
concentrations (reprinted from Strickland et al., 2010). 3
Similar to the study by Silverman and Ito (2010), we consider the extent to which the
reported concentration-response function indicates a relatively high degree of confidence in
health effect associations on days when all monitored 8-hour Os concentrations are below 75 ppb
(Tabe 3-5, below).44 In considering the information presented in Table 3-5, we first note that
when 3-day averaged 63 concentrations were in the range of the mean (i.e., 51 to 60 ppb), there
were 77 days (out of 516; 14.9%) with at least one monitor recording an 8-hour daily maximum
63 concentration above the level of the current standard. In contrast, during the 519 days when
averaged 63 concentrations were in the lower portion of the distribution where study authors
indicate relatively high confidence in the reported concentration-response relationship (i.e.,
between 31 and 45 ppb), there were 4 days with at least one monitor in the study area measuring
an 8-hour daily maximum Os concentration greater than 75 ppb (approximately 0.8% of days).
43 This figure was also reprinted in the ISA (U.S. EPA, 2013; Figure 6-18).
44
The study by Strickland et al. (2010) used five monitors. For our evaluation of highest 8-hour daily maximum
concentrations (i.e., from the individual monitor recording the highest such concentration), we obtained information
from the four of these study area monitors that report data to AQS (Appendix 3-D).
3-69
-------
1 Thus, on over 99% of the days when "averaged" O^ concentrations were between 31 and 45 ppb,
2 all monitors measured 8-hour daily maximum 63 concentrations below 75 ppb.
3 Table 3-5. Distribution of daily 8-hour maximum ozone concentrations from highest
4 monitors over range of 3-day moving averages of population-weighted
5 concentrations (for study area evaluated by Strickland et al., 2010)
rmax 3-day moving average across monitors (ppb)
from
highest
monitor
during
Min
5th
25th
50th
75th
95th
98th
99th
Max
26-30
(75 days)
27
29
36
38
44
53
59
66
70
31-35
(144 days)
24
33
40
44
49
56
59
61
64
36-40
(165 days)
33
36
45
50
54
70
75
78
83
41-45
(210 days)
30
45
51
58
62
74
84
89
99
46-50
(235 days)
36
48
56
62
70
81
86
88
97
51-55
(244 days)
38
54
63
68
74
84
93
95
107
56-60
(272 days)
49
60
68
73
81
95
100
103
106
61-65
(234 days)
52
64
73
78
85
95
98
102
122
66-70
(169 days)
57
69
79
86
96
104
107
108
129
71 to 75
(124 days)
67
75
85
92
101
109
110
112
120
76 to 80
(106 days)
69
85
95
101
108
120
124
124
138
g Days > 75 00 2 2 10 24 53 80 89 87 87
7 In summary, analyses of air quality data from the study locations evaluated by Silverman
8 and Ito (2010) and Strickland et al. (2010) indicate a relatively high degree of confidence in
9 reported statistical associations with respiratory health outcomes on days when all monitors
10 recorded 8-hour average Os concentrations of 75 ppb or below. Though these analyses do not
11 identify true design values, the presence of (^-associated respiratory effects on such days
12 provides insight into the types of health effects that could occur in locations that meet the current
13 standard.
14 There are several important uncertainties that are specifically related to our analyses of
15 distributions of Os air quality in the study locations evaluated by Silverman and Ito (2010) and
16 Strickland et al. (2010). Although these studies report health effect associations with two-day
17 (Silverman and Ito) and three-day (Strickland) averages of daily 63 concentrations, it is possible
18 that the respiratory morbidity effects reported in these studies were also at least partly
19 attributable to the days immediately preceding these two- and three-day periods. In support of
20 this possibility, Strickland et al. reported positive and statistically significant associations with
21 emergency department visits for multiple lag periods, including lag periods exceeding three days.
22 Our analysis of highest monitored concentrations focuses on two- and three- day periods, as used
23 in the published study to generate concentration-response functions. This could have important
24 implications for our interpretation of the reported concentration-response functions if a 2-day
25 period with no monitors measuring 8-hour concentrations at or above 75 ppb is immediately
26 preceded by one or more days with monitors that do exceed 75 ppb. Although we do not know
27 the extent to which 63 concentrations on a larger number of days could have contributed to
3-70
-------
1 reported health effect associations, we note this as a potentially important uncertainty in our
2 consideration of concentration-response functions within the context of the current standard.
3 In addition, an important uncertainty that applies to epidemiologic studies in general is
4 the extent to which reported health effects are caused by exposures to 63 itself, as opposed to
5 other factors such as co-occurring pollutants or other pollutant mixtures. Although both of the
6 studies evaluated above reported health effect associations in co-pollutant models, this
7 uncertainty becomes an increasingly important consideration as health effect associations are
8 evaluated at lower ambient Os concentrations (i.e., resulting in lower exposure concentrations).
9 One approach to considering the potential importance of this uncertainty in
10 epidemiologic studies is to evaluate the extent to which there is coherence with the results of
11 experimental studies (i.e., in which the study design dictates that exposures to O^ itself are
12 responsible for reported effects). Therefore, in further considering uncertainties associated with
13 the above air quality analyses for the study areas evaluated by Silverman and Ito (2010) and
14 Strickland et al. (2010), we evaluate the following question:
15 • To what extent is there coherence between evidence from controlled human
16 exposure studies and epidemiologic studies supporting the occurrence of Os-
17 attributable respiratory effects when 8-hour daily maximum ambient Os
18 concentrations are at or below 75 ppb?
19 As summarized above and as discussed in the ISA (U.S. EPA, 2013, section 6.2),
20 controlled human exposure studies demonstrate the occurrence of respiratory effects in an
21 appreciable percentage of healthy adults following single short-term exposures to Os
22 concentrations as low as 60 ppb. In addition, as 63 exposure concentrations exceed 60 ppb: 1)
23 effects in healthy adults become larger and more serious; 2) a broader range of effects are
24 observed in a greater percentage of exposed individuals; and 3) effects are reported more
25 consistently across studies. Thus, as the potential increases for exposures to 63 concentrations
26 approaching or exceeding 60 ppb, our confidence increases that reported respiratory health
27 effects could be caused by exposures to the ambient 63 concentrations present in study locations.
28 In considering coherence with results of controlled human exposure studies in this way, it
29 is important to note the relative lack of experimental data for exposure concentrations below 60
30 ppb. It is possible that lower exposure concentrations can result in respiratory effects serious
31 enough to lead to hospital admissions or emergency department visits, particularly in at-risk
32 populations such as children and asthmatics. Thus, although we consider coherence between
33 epidemiologic and controlled human exposure study results, we also acknowledge that an 63
34 exposure concentration of 60 ppb does not constitute a bright line below which we are confident
35 that effects no longer occur, particularly for at-risk groups.
3-71
-------
1 As discussed above, for the study by Silverman and Ito (2010), 26 to 45 ppb represents
2 the lower end of the range of "averaged" daily maximum 8-hour concentrations over which the
3 study indicates a relatively high degree of confidence in the statistical association with
4 respiratory hospital admissions (and for which virtually all monitored concentrations were 75
5 ppb or below). As averaged concentrations increase from 26 to 45 ppb, the number of days with
6 maximum monitored concentrations exceeding 60 ppb increases dramatically (Table 3-4,
7 above).45 For example, of the 178 days with area-wide average daily maximum 8-hour
8 concentrations from 26 to 30 ppb, only about 2% had any monitors recording ambient
9 concentrations of 60 ppb or greater. In contrast, of the 196 days with area-wide average
10 concentrations from 41 to 45 ppb, about half had at least one monitor recording an ambient
11 concentration near or above 60 ppb, with monitors on some days measuring concentrations
12 greater than 80 ppb. Thus, as averaged concentrations approach 45 ppb there is an increasing
13 likelihood that at least some portion of the study population could have been exposed to Os
14 concentrations approaching or exceeding those reported in controlled human exposure studies to
15 cause respiratory effects in healthy adults.
16 For the study by Strickland et al (2010), "averaged" concentrations from 30 to 45 ppb
17 represent the lower end of the range of concentrations over which the study indicates a relatively
18 high degree of confidence in the statistical association with respiratory emergency department
19 visits (and for which virtually all monitored concentrations were 75 ppb or below). Similar to the
20 study area evaluated by Silverman and Ito, as 8-hour area-wide average 63 concentrations
21 approach 45 ppb, maximum monitored concentrations exceed 60 ppb more frequently. On most
22 days contributing to averaged Os concentrations from 41 to 45 ppb, maximum monitor
23 concentrations were near or above 60 ppb. On a small number of these days, maximum
24 monitored concentrations were greater than 80 ppb. Therefore, similar to the study by Silverman
25 and Ito (2010), at least some portion of the study population on these days are likely to have been
26 exposed to Oj concentrations exceeding those reported in controlled human exposure studies to
27 cause respiratory effects in some healthy adults.
28 Thus, consideration of distributions of individual monitored concentrations, in
29 conjunction with the results of Os controlled human exposure studies, supports the role of
30 ambient O?, concentrations in eliciting the reported respiratory hospital admissions and
31 emergency department visits. Specifically, these analyses support the occurrence of Oj-
32 attributable hospital admissions and emergency department visits on days when virtually all
33 monitors measure 8-hour ambient concentrations below 75 ppb.
45
Though, as noted above, the epidemiologic studies by Silverman and Ito (2010) and Strickland et al. (2010) do not
provide information on the extent to which reported health effects result from exposures to any specific O3
concentrations.
3-72
-------
1 In further evaluating Os concentration-response relationships within the context of the
2 adequacy of the current standard, we note that some epidemiologic studies report health effect
3 associations for air quality distributions restricted to ambient pollutant concentrations below one
4 or more predetermined cut points. Such "cut point" analyses can provide information on the
5 magnitude and statistical precision of effect estimates for defined distributions of ambient
6 concentrations, which may in some cases include distributions that would meet the current
7 standard. Therefore, we next consider the following question:
8 • To what extent do cut-point analyses from epidemiologic studies report health effect
9 associations at ambient Os concentrations lower than previously identified or that
10 would likely meet the current standard?
11 By considering the magnitude and statistical significance of effect estimates for restricted
12 air quality distributions, cut-point analyses can provide insight into the extent to which health
13 effect associations are driven by ambient concentrations above the cut point, versus
14 concentrations below the cut point. For studies that evaluate multiple cut points, these analyses
15 can provide insights into the magnitude and statistical precision of health effect associations for
16 different portions of the distribution of ambient concentrations, including insights into the
17 ambient concentrations below which uncertainty in reported associations becomes notably
18 greater. As with analyses of concentration-response functions, discussed above, the cut points
19 below which confidence intervals become notably wider depend in large part on data density.46
20 In the U.S. multicity study by Bell et al. (2006), study authors used the NMMAPS data
21 set to evaluate associations between 2-day rolling average 63 concentrations47 and total (non-
22 accidental) mortality in 98 U.S. cities from 1987 to 2000. Based on the full distributions of
23 ambient 63 concentrations in study cities, the large majority of the NMMAPS cities would have
24 violated the current standard during the study period (Appendix 3-D). However, Bell et al.
25 (2006) also reported health effect associations in a series of cut-point analyses, with effect
26 estimates based only on the subsets of days contributing to "averaged" 63 concentrations below
27 cut points ranging from 5 to 60 ppb (see Figure 2 in Bell et al., 2006). The lowest cut-point for
28 which the association between 63 and mortality was reported to be statistically significant was
29 30 ppb (based on visual inspection of Figure 2 in published study). As with the studies by
30 Silverman and Ito (2010) and Strickland et al. (2010), we consider what these cut point analyses
46As such, these analyses provide insight into the ambient concentrations below which the available air quality
information becomes too sparse to support conclusions about the nature of concentration-response relationships,
with a high degree of confidence.
47Two-day rolling averages of 24-hour average O3 concentrations were calculated throughout the study period. This
calculation was done across study monitors in study cities with multiple monitors.
3-73
-------
1 indicate with regard to the potential for health effect associations to extend to ambient O^
2 concentrations likely to be allowed by the current 63 NAAQS.
3 We attempted to recreate the subsets of air quality data used in the cut point analyses
4 presented by Bell et al. (2006). In doing so, we applied the criteria described in the published
5 study to generate air quality subsets corresponding to those defined by the cut points evaluated
,10
6 by study authors. From the days with averaged Oj, concentrations below each cut point, we
7 identified 3-year averages of annual 4th highest 8-hour daily maximum 63 concentrations in each
8 study area. We then compared these 4* highest Os concentrations to the level of the current
9 standard in order to provide insight into the extent to which the air quality distributions included
10 in various cut point analyses would likely have met the current standard.
11 We particularly focus on the lowest cut-point for which the association between O^ and
12 mortality was reported to be statistically significant (i.e., 30 ppb, as noted above). Based on the
13 Os air quality concentrations that met the criteria for inclusion in the 30 ppb cut point analysis,
14 95% of study areas had 3-year averages of annual 4 highest 8-hour daily maximum Oj,
15 concentration at or below 75 ppb over the entire study period. For the 35 ppb cut point, which
16 also resulted in a statistically significant association with mortality, 68% of study areas had 3-
17 year averages of annual 4th highest 8-hour daily maximum 63 concentration at or below 75 ppb.
18 This suggests that the large maj ority of air quality distributions that provided the basis for
19 positive and statistically significant associations with mortality (i.e., for the 30 and 35 ppb cut
20 points) would likely have met the current 63 standard. For higher cut points, all of which also
21 resulted in statistically significant associations with mortality, the majority of study cities had 3-
22 year averages of annual 4* highest 8-hour daily maximum concentrations greater than 75 ppb.
23
48We were unable to obtain the air quality data used to generate the cut-point analyses in the study published by Bell
et al. (2006). Therefore, we generated 2-day averages of 24-hour O3 concentrations in study locations using the air
quality data available in AQS, combined with the published description of study area definitions. In doing so, we did
not recreate the trimmed means used by Bell. As discussed below, this represents an important uncertainty in our
analysis.
3-74
-------
1 Table 3-6.
2
Number of study cities with 4 highest 8-hour daily maximum concentrations
greater than 75 ppb, for various cut-point analyses presented in Bell et al.
(2006)
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
Number (%) of
Cities with 4th
highest >75 (any
3 -yr period;
1987-2000)
Cut-point for 2-day moving average across monitors and cities (24-hour average)
25
0 (0%)
30
5 (5%)
35
31
(32%)
40
70
(71%)
45
86
(88%)
50
88
(90%)
55
92
(94%)
60
92
(94%)
All
92
(94%)
In addition to the uncertainties noted above for our analysis of the single-city studies by
Silverman and Ito (2010) and Strickland et al. (2010) (e.g., attributing effects specifically to air
quality included in various subsets), an important uncertainty related to this analysis is that we
were unable to obtain the air quality data used to generate the cut-point analyses in the study
published by Bell et al. (2006). Therefore, as noted above, we generated 2-day averages of 24-
hour Oj, concentrations in study locations using the air quality data available in AQS, combined
with the published description of study area definitions. In doing so, we did not recreate the
trimmed means used by Bell. An important uncertainty in this approach is the extent to which we
were able to appropriately recreate the cut-point analyses in the published study.
The ISA also notes important uncertainties inherent in multicity studies that evaluate the
potential for thresholds to exist, as was done in the study by Bell et al. (2006). Specifically, the
ISA highlights the regional heterogeneity in 63 health effect associations as a factor that could
obscure the presence of thresholds, should they exist, in multicity studies (U.S. EPA, 2013,
sections 2.5.4.4 and 2.5.4.5). The ISA notes that community characteristics (e.g., activity
patterns, housing type, age distribution, prevalence of air conditioning) could be important
contributors to reported regional heterogeneity (U.S. EPA, 2013, section 2.5.4.5). Given this
heterogeneity, the ISA concludes that "a national or combined analysis may not be appropriate to
identify whether a threshold exists in the Os-mortality C-R relationship" (U.S. EPA, 2013, p. 2-
33). This represents an important source of uncertainty when characterizing our confidence in
reported concentration-response relationships over distributions of ambient 63 concentrations,
based on multicity studies. This uncertainty becomes increasingly important when interpreting
concentration-response relationships at lower ambient 63 concentrations, particularly those
concentrations corresponding to portions of distributions where data density decreases notably.
3.1.4.3 Concentrations in Epidemiologic Studies - "Long-term" Metrics
We next consider the extent to which epidemiologic studies employing longer-term
ambient Os concentration metrics inform our understanding of the air quality conditions
3-75
-------
1 associated with Os-attributable health effects, and specifically inform consideration of the extent
2 to which such effects could occur under air quality conditions meeting the current standard.
3 Unlike for the studies of short-term O^ discussed above, the available U.S. and Canadian
4 epidemiologic studies evaluating long-term ambient 63 concentration metrics have not been
5 conducted in locations likely to have met the current 8-hour Os standard during the study period
6 (Appendix 3-D). Therefore, although these studies contribute to our understanding of health
7 effects associated with long-term or repeated exposures to ambient 63 (as summarized in section
8 3.1.2 above), consideration of study area design values does not inform our consideration of the
9 extent to which those health effects may be occurring in locations that met the current standard.
10 In further considering epidemiologic studies of long-term 63 concentrations, we also
11 evaluate the extent to which concentration-response functions have been reported over
12 distributions of ambient concentrations, and what those functions can tell us about health effect
13 associations with Os concentrations likely to be allowed by the current standard. Specifically, we
14 consider the following question:
15 • To what extent do confidence intervals around concentration-response functions
16 indicate Os-associated health outcomes at ambient concentrations meeting the
17 current Os standard?
18 The ISA identifies a single epidemiologic study reporting confidence intervals around a
19 concentration-response function for "long-term" 63 concentrations and respiratory mortality
20 (Jerrett et al., 2009; U.S. EPA, 2013, sections 7.2.7, 7.2.8 and 7.7). Jerrett et al. (2009) reported
21 that when seasonal averages of 1-hour daily maximum 03 concentrations49 ranged from 33 to
22 104 ppb, there was no statistical deviation from a linear concentration-response relationship
23 between Os and respiratory mortality across 96 U.S. cities (U.S. EPA, 2013, section 7.7).
24 However, the authors reported "limited evidence" for an effect threshold at an 63 concentration
25 of 56 ppb (p=0.06). Visual inspection of this concentration-response function (Figure 3-6)
26 confirms the possibility of an inflection point just below 60 ppb, which is close to the median
27 concentration across cities (i.e., 57 ppb).
49
Jerrett et al. (2009) evaluated the April to September averages of 1 -hour daily maximum O3 concentrations across
96 U.S. metropolitan areas from 1977- 2000. In urban areas with multiple monitors, April to September 1-hour daily
maximum concentrations from each individual monitor were averaged. This step was repeated for each year in the
study period. Finally, each yearly averaged O3 concentrations was then averaged again to yield the single averaged
1-hour daily maximum O3 concentration depicted on the x axis of Figure 3-6 below.
3-76
-------
0.2-
0.0-
lii i ijanniiiiiiiii
40 60 SO 100
l!i£HJl^^
2 Figure 3-6. Exposure-Response relationship between risk of death from respiratory causes
3 and ambient Os concentration study metric (Jerrett et al., 2009).
4 We consider the extent to which this concentration-response function indicates
5 confidence in the reported health effect association at various ambient Os concentrations. In
6 doing so, we note the following: (1) most of the study cities had O?, concentrations above 53.1
7 ppb (i.e., the upper bound of the first quartile), accounting for approximately 72% of the
8 respiratory deaths in the cohort (Table 2 in Jerrett et al. 2009); (2) confidence intervals widen
9 notably for 63 concentrations in the first quartile (based on visual inspection of Figure 3-6); and
10 (3) study authors noted limited evidence for a threshold at 56 ppb.50 In considering this
1 1 information, we conclude that the analysis reported by Jerrett indicates the greatest confidence in
12 the linear concentration-response function for "long-term" 63 concentrations in the upper three
13 quartiles (i.e., above about 53 ppb).
14 Based on information in the published study (Figure 1 in Jerrett et al., 2009), we
15 identified 72 of the 96 study cities as having ambient Oi concentrations in the highest three
16 quartiles (Appendix 3-D). As noted above, these 72 cities account for approximately 72% of the
17 respiratory deaths in the cohort (Table 2 in Jerrett et al. 2009). Of these 72 cities, 71 had 3 -year
18 averages of annual 4th highest 8-hour daily maximum 63 concentrations above 75 ppb (Appendix
19 3-D). Thus, the current 8-hour NAAQS would have been violated during the study period in
20 virtually all of the study cities that contribute to the range of long-term 63 concentrations over
21 which we have the greatest confidence in the reported relationship with respiratory mortality.
22 Thus, while the study by Jerrett et al. (2009) contributes to our understanding of health effects
23 associated with ambient 63 (as summarized in section 3.1.2 above), it is less informative
50The ISA does not reach conclusions regarding the potential for a threshold in the association between "long-term"
O3 concentrations and respiratory mortality.
3-77
-------
1 regarding the extent to which those health effects may be occurring under air quality conditions
2 allowed by the current standard.
3 3.1.5 Public Health Implications
4 In this section, we address the public health implications of Os exposures with respect to
5 the factors that put populations at increased risk from exposures (section 3.1.5.1), the size of at-
6 risk populations (section 3.1.5.2), and the potential effects of averting behavior on reducing 63
7 exposures and associated health effects (section 3.1.5.3). Providing appropriate public health
8 protection requires consideration of the factors that put populations at greater risk from 63
9 exposure. In order to estimate potential overall for public health impacts, it is important to
10 consider not only the adversity of the health effects, but also the populations at greater risk and
11 potential behaviors that may reduce exposure.
12 3.1.5.1 At-Risk Populations
13 In this section we address the following question:
14 • To what extent does the currently available scientific evidence expand our
15 understanding of at-risk populations?
16 The currently available evidence expands our understanding of populations that were
17 identified to be at greater risk of Os-related health effects at the time of the last review (i.e.,
18 people who are active outdoors, people with lung disease, children and older adults and people
19 with increased responsiveness to 63) and supports the identification of additional factors that
20 may lead to increased risk (U.S. EPA 2006, section 3.6.2; U.S. EPA, 2013, chapter 8).
21 Populations and lifestages may be at greater risk for Os-related health effects due to factors
22 contribute to their susceptibility and/or vulnerability to ozone. The definitions of susceptibility
23 and vulnerability have been found to vary across studies, but in most instances "susceptibility"
24 refers to biological or intrinsic factors (e.g., lifestage, sex, preexisting disease/conditions) while
25 "vulnerability" refers to non-biological or extrinsic factors (e.g., socioeconomic status [SES])
26 (U.S. EPA, 2013, p. 8-1; U.S. EPA, 2010c, 2009d). In some cases, the terms "at-risk" and
27 "sensitive" have been used to encompass these concepts more generally. In the ISA and this PA,
28 "at-risk" is the all-encompassing term used for groups with specific factors that increase their
29 risk of Os-related health effects. Further discussion of at-risk populations can be found in
30 Appendix 3C below.
31 There are multiple avenues by which groups may experience increased risk for 63-
32 induced health effects. A population or lifestage51 may exhibit greater effects than other
51 Lifestages, which in this case includes childhood and older adulthood, are experienced by most people over the
course of a lifetime, unlike other factors associated with at-risk populations.
3-78
-------
1 populations or lifestages exposed to the same concentration or dose, or they may be at greater
2 risk due to increased exposure to an air pollutant (e.g., time spent outdoors). A group with
3 intrinsically increased risk would have some factor(s) that increases risk through a biological
4 mechanism and, in general, would have a steeper concentration-risk relationship, compared to
5 those not in the group. Factors that are often considered intrinsic include asthma, genetic
6 background, and lifestage. A group of people could also have extrinsically increased risk, which
7 would be through an external, non-biological factor, including, for example, socioeconomic
8 status (SES) and diet. Some groups are at risk of increased internal dose at a given exposure
9 concentration, for example, because of breathing patterns. This category would include people
10 who work or exercise outdoors. Finally, there are those who might be placed at increased risk for
11 experiencing greater exposures by being exposed at higher concentrations. This would include,
12 for example, groups of people with greater exposure to ambient 63 due to less availability or use
13 of home air conditioners such that they are more likely to be in locations with open windows on
14 high ozone days. Some groups may be at increased risk of Os-related health effects through a
15 combination of factors. For example, children tend to spend more time outdoors when 63 levels
16 are high, and at higher levels of activity than adults, which leads to increased exposure and dose,
17 and they also have biological, or intrinsic, risk factors (e.g., their lungs are still developing)
18 (U.S. EPA, 2013, Chapter 8). An at-risk population or lifestage is more likely to experience
19 adverse health effects related to O^ exposures and/or, develop more severe effects from exposure
20 than the general population.
21 Based on the currently available evidence, the at-risk populations for 63- related health
22 effects are based on factors that include: asthma, lifestages (children and older adults), genetic
23 variability, dietary factors, and working outdoors (U.S. EPA, 2013, section 8.5, Table 8-6). This
24 conclusion is supported by consistency in findings across studies and evidence of coherence in
25 results from different scientific disciplines. The current evidence is suggestive of a potential for
26 three other factors to influence risk of Os-related health effects. The evidence suggests that
27 women may be at greater risk than men, groups with low SES or living in neighborhoods with
28 low SES may be at greater risk than other socioeconomic groups, and obesity may be a potential
29 risk factor. Further studies are needed, however, on these factors. Overall, the factors most
30 strongly supported as contributing to increased risk of Os-related effects are related to asthma,
31 lifestage (children and older adults), genetic variability, dietary factors, and working outdoors
32 (U.S. EPA, 2013; chapter 8).
33 In summary, the evidence available in this review supports the identification of the
34 following populations and lifestages as having increased risk for Os-related health effects, based
35 on consistency in findings across studies and evidence of coherence in results from different
3-79
-------
1 scientific disciplines (U.S. EPA, 2013, section 8.5): individuals with certain genotypes,
2 individuals with asthma, younger and older age groups, individuals with reduced intake of
3 certain nutrients, and outdoor workers. Multiple genetic variants have been observed in
4 epidemiologic and controlled human exposure studies to affect the risk of Ch-related respiratory
5 outcomes and support is provided by toxicological studies of genetic factors (U.S. EPA, 2013,
6 section 8.1). Asthma has been recognized in past reviews and continues to be well established in
7 this review as a risk factor for (Vrelated health effects based on multiple lines of evidence,
8 included controlled human exposure and toxicological studies in animal models, as well as some
9 evidence from epidemiologic studies. This evidence supports our understanding of the
10 biological (intrinsic) factors that put individuals with asthma at greater risk than other groups
11 (U.S. EPA, 2013, section 8.2.2). As noted above, some extrinsic (exposure-related) and intrinsic
12 factors contribute to the identification of children as an at-risk lifestage. Children have higher
13 exposure and dose due to increased time spent outdoors and ventilation rate, their lungs are still
14 developing, and they are more likely than adults to have asthma (U.S. EPA, 2013, section
15 8.3.1.1). Older adults may also withstand greater 63 exposure and not seek relief as quickly as
16 younger adults. Multiple epidemiologic, controlled human exposure and toxicological studies
17 reported that diets deficient in vitamins E and C are associated with risk of Ch-related health
18 effects for all lifestages. Previous studies have shown that increased exposure to 63 due to
19 outdoor work leads to increased risk of Os-related health effects and it is clear that outdoor
20 workers have higher exposures, and possibly greater internal doses, of 63, which may lead to
21 increased risk of (Vrelated health effects (U.S. EPA, 2013, section 8.5).
22 In some cases, it is difficult to determine a factor that results in increased risk of effects.
23 For example, previous assessments have included controlled human exposure studies in which
24 some healthy individuals demonstrate greater Os-related health effects compared to other healthy
25 individuals. Intersubject variability has been observed for lung function decrements,
26 symptomatic responses, pulmonary inflammation, AHR, and altered epithelial permeability in
27 healthy adults exposed to 63 and these results tend to be reproducible within a given individual
28 over a period of several months indicating differences in the intrinsic responsiveness. In many
29 cases the reasons for the variability is not clear. This may be because one or some of the factors
30 described above have not been evaluated in studies, or it may be that additional, unidentified
31 factors influence individual responses to O^ (U.S. EPA, 2013, section 8.5).
32 As discussed in chapter 8 of the ISA, and further in Appendix 3C below, the challenges
33 and limitations in evaluating the factors that can increase risk for experiencing (Vrelated health
34 effects may contribute to a lack of information about the factors that may increase risk from 63
35 exposures. This lack of information may contribute to conclusions that evidence for some
3-80
-------
1 factors, such as sex, SES, and obesity provided "suggestive" evidence of increased risk, or that
2 for a number of factors the evidence was inadequate to draw conclusions about potential increase
3 in risk of effects. Overall, the factors most strongly supported as contributing to increased risk of
4 populations for experience (Vrelated effects were related to asthma, lifestage (children and older
5 adults), genetic variability, dietary factors, and working outdoors.
6 3.1.5.2 Size of At-Risk Populations and Lifestages in the United States
7 One consideration in the assessment of potential public health impacts is the size of
8 various population groups for which there is adequate evidence of increased risk for health
9 effects associated with Os-related air pollution exposure. The factors for which the ISA judged
10 the evidence to be "adequate" with respect to contributing to increased risk of Os-related effects
11 among various populations and lifestages included: asthma; childhood and older adulthood; diets
12 lower in vitamins C and E; certain genetic variants and, working outdoors (EPA, 2013, section
13 8.5).
14 With regard to asthma, Table 3-12 below summarizes information on the prevalence of
15 current asthma by age in the U.S. adult population in 2010 (Schiller et al. 2012; children - Bloom
16 et al., 2011). Individuals with current asthma constitute a fairly large proportion of the
17 population, including more than 25 million people. Asthma prevalence tends to be higher in
18 children than adults.
19 Within the U.S., approximately 8.2% of adults have reported currently having asthma
20 (Schiller et al., 2012) and 9.5% of children have reported currently having asthma (Bloom et al.,
21 2011). Table 3-12 below provides more detailed information on prevalence of asthma by age in
22 the U.S.
23
3-81
-------
1 Table 3-7. Prevalence of asthma by age in the U.S.
Age (years)
0-4
5-11
12-17
18-44
45-64
65-74
75+
Asthma prevalence is reported for "still
Source: Statistics for adults: Schiller et
N (in thousands)
1,285
3,020
2,672
8,902
6,704
1,849
1,279
has asthma"
al. (2012); Statistics for children:
Percent
6.0
10.5
10.9
8.1
8.4
8.7
7.4
Bloom etal. (2011)
3 With regard to lifestages, based on U.S. census data from 2010 (Howden and Meyer,
4 2011), about 74 million people, or 24% of the U.S. population, are under 18 years of age and
5 more than 40 million people, or about 13% of the U.S. population, are 65 years of age or older.
6 Hence, a large proportion of the U.S. population, more than 33%, is included in age groups that
7 are considered likely to be at increased risk for health effects from ambient O?, exposure.
8 With regard to dietary factors, no statistics are available to estimate the size of an at-risk
9 population based on nutritional status.
10 With regard to outdoor workers, in 2010 approximately 11.7% of the total number of
11 people (143 million people) employed, or about 16.8 million people, worked outdoors one or
CO
12 more day per week (based on worker surveys). Of these approximately 7.4% of the workforce,
13 or about 7.8 million people, worked outdoors three or more days per week.
14 The health statistics data illustrate what is known as the "pyramid" of effects. At the top
15 of the pyramid, there are approximately 2.5 million deaths from all causes per year in the U.S.
16 population, with about 250 thousand respiratory-related deaths (CDC-WONDER, 2008). For
17 respiratory health diseases, there are nearly 3.3 million hospital discharges per year (HCUP,
18 2007), 8.7 million respiratory ED visits (HCUP, 2007), 112 million ambulatory care visits
52 The O*NET program is the nation's primary source of occupational information. Central to the project is the
O*NET database, containing information on hundreds of standardized and occupation-specific descriptors. The
database, which is available to the public at no cost, is continually updated by surveying a broad range of workers
from each occupation, http://www.onetcenter.org/overview.html
http://www.onetonline.org/find/descriptor/browse/Work_Context/4.C.2/
3-82
-------
1 (Woodwell and Cherry, 2004), and an estimated 700 million restricted activity days per year due
2 to respiratory conditions (Adams et al., 1999). Combining small risk estimates with relatively
3 large baseline levels of health outcomes can result in quite large public health impacts. Thus,
4 even a small percentage reduction in 63 health impacts on cardiopulmonary diseases would
5 reflect a large number of avoided cases.
6 3.1.5.3 Averting Behavior
7 The activity pattern of individuals is an important determinant of their exposure (ISA,
8 U.S. EPA, 2013, section 4.4.1). Variation in Os concentrations among various
9 microenvironments means that the amount of time spent in each location, as well as the level of
10 activity, will influence an individual's exposure to ambient 63. Activity patterns vary both
11 among and within individuals, resulting in corresponding variations in exposure across a
12 population and over time. Individuals can reduce their exposure to 63 by altering their behaviors,
13 such as by staying indoors, being active outdoors when air quality is better, and by reducing their
14 activity levels or reducing the time being active outdoors on high-Os days (U.S. EPA, 2013,
15 section 4.4.2). The evidence in this topic area, while not addressed in the 2006 AQCD, is
16 evaluated in the ISA for this review.
17 The widely reported Air Quality Index (AQI) conveys advice to the public, and
18 particularly at-risk populations, on reducing exposure on days when ambient levels of common
19 air pollutants are elevated (www.airnow.gov). The AQI describes the potential for health effects
20 from Os (and other individual pollutants) in six color-coded categories of air-quality, ranging
21 from Good (green), Moderate (yellow), Unhealthy for Sensitive Groups (orange), Unhealthy
22 (red), and Very Unhealthy (purple), and Hazardous (maroon). Levels in the unhealthy ranges
23 (i.e., Unhealthy for Sensitive Groups and above) come with recommendations about reducing
24 exposure. Forecasted and actual AQI values for 63 are reported to the public during the 63
25 season. The AQI advisories explicitly state that children, older adults, people with lung disease,
26 and people who are active outdoors, may be at greater risk from exposure to 03. People are
27 advised to reduce exposure depending on the predicted 63 levels and the likelihood of risk. This
28 advice includes being active outdoors when air quality is better, and reducing activity levels or
29 reducing the time being active outdoors on high-Os days. Staying indoors to reduce exposure is
30 not recommended until air quality reaches the Very Unhealthy or Hazardous categories.
31 Evidence of individual averting behaviors in response to AQI advisories has been found
32 in several studies, including activity pattern and epidemiologic studies, especially for the at-risk
33 populations, such as children, older adults, and people with asthma, who are targeted by the
34 advisories. Such effects are less pronounced in the general population, possibly due to the
35 opportunity cost of behavior modification. Epidemiologic evidence from a study (Neidell and
3-83
-------
1 Kinney, 2010) conducted in the 1990's in Los Angeles, CA reports increased asthma hospital
2 admissions among children and older adults when O3 alert days (1-hour max O3 concentration
3 >200 ppb) were excluded from the analysis of daily hospital admissions and O3 concentrations
4 (presumably thereby eliminating averting behavior based on high O3 forecasts). The lower rate
5 of admissions observed when alert days were included in the analysis suggests that estimates of
6 health effects based on concentration-response functions that do not account for averting
7 behavior may be biased towards the null (U.S. EPA, 2013, section 4.4.2).
8 3.2 AIR QUALITY-, EXPOSURE-, AND RISK-BASED CONSIDERATIONS
9 In order to inform judgments about the public health impacts of Os-related health
10 effects, the second draft HREA has developed and applied models to estimate human exposures
11 to Os and Os-associated health risks across the United States, with a specific focus on urban case
12 study areas (U.S. EPA, 2014).53 The second draft HREA uses photochemical modeling to adjust
13 air quality from the 2006-2010 Os seasons to just meet the current and alternative standards for
14 the 2006-2008 and 2008-2010 periods.54 In this section, staff considers estimates of short-term
15 Os exposures and estimates of health risks associated with short- and long-term Os exposures, for
16 air quality adjusted to just meet the current Os standard. In section 3.2.1, we consider the
17 implications for exposure and risk estimates of the approach used in the second draft HREA to
18 adjust air quality. Sections 3.2.2 and 3.2.3 discuss our exposure-based and risk-based
19 considerations, respectively. In these sections we specifically consider the following question:
20 • What are the nature and magnitude of Os exposures and health risks remaining
21 upon adjusting recent air quality to just meet the current Os standard, and what are
22 the important uncertainties associated with those exposure and risk estimates?
23 3.2.1 Consideration of the Adjusted Air Quality Used in Exposure and Risk
24 Assessments
25 In the first draft HREA for this review, as in the last review, the EPA relied upon
26 quadratic rollback to adjust hourly Os concentrations in urban case study areas to just meet the
27 current Os standard (U.S. EPA, 2012b). Although the quadratic rollback method reproduces
28 historical patterns of air quality changes better than some alternative methods, it relies on
The 15 urban case study areas analyzed for exposures are Atlanta, Baltimore, Boston, Chicago, Cleveland, Dallas,
Denver, Detroit, Houston, Los Angeles, New York, Philadelphia, Sacramento, St. Louis, and Washington, DC.
Morbidity and mortality risk estimates are presented for these same areas, with the exception of Chicago, Dallas,
and Washington, DC. The second draft HREA also presents a national scale mortality risk assessment for unadjusted
(recent) air quality. This national-scale assessment, which focuses on existing air quality conditions and does not
estimate the health risks associated with just meeting the current or alternative standards, can provide perspective on
the relationship between national-scale O3 public health impacts and impacts estimated in specific urban areas.
54 Three-year periods are used recognizing that the current standard is the average across three years of the annual
fourth-highest daily maximum 8-hour average concentration.
3-84
-------
1 statistical relationships without explicitly accounting for atmospheric chemistry and precursor
2 emissions (U.S. EPA, 2014, Chapter 4). An important drawback of the quadratic rollback
3 approach, recognized in the first draft HREA (U.S. EPA, 2012b), is that it forces all monitors in
4 an assessment area to exhibit the same response when air quality is adjusted. It does not allow for
5 the spatial or temporal heterogeneity in responses that result from the non-linear atmospheric
6 chemistry that influences ambient Os concentrations (U.S. EPA, 2014, Chapter 4). Because
7 quadratic rollback does not account for physical and chemical atmospheric processes, or the
8 sources of emissions precursors that lead to Os formation, a backstop or "floor" must be used
9 when applying quadratic rollback to just meet current or alternative standards to ensure that
10 estimated 63 is not reduced in a manner inconsistent with 63 chemistry, such as to reduce
11 concentrations below that associated with background sources (U.S. EPA, 2014, Chapter 4).
12 Consistent with recommendations from the National Research Council of the National
13 Academies (NRC, 2008), the second draft HREA uses a photochemical model to estimate
14 sensitivities of Os to changes in precursor emissions, in order to estimate ambient O?,
15 concentrations that would just meet the current and alternative standards (U.S. EPA, 2014,
16 Chapter 4).55 For the urban case study areas evaluated in the second draft HREA, this model-
17 based adjustment approach was set up to estimate hourly 63 concentrations at each monitor
18 location when modeled U.S. anthropogenic precursor emissions (i.e., NOx, VOC)56 were reduced
19 to estimate air quality that just meets the current and alternative 03 standards.57
20 As discussed in Chapter 4 of the second draft HREA (U.S. EPA, 2014), this approach
21 models the physical and chemical atmospheric processes that influence ambient Os
22 concentrations. Compared to the quadratic rollback approach, it provides more realistic estimates
23 of the spatial and temporal responses of 63 to reductions in precursor emissions. These improved
24 estimates avoid many of the limitations inherent in the quadratic rollback method, including the
25 requirement that all monitors in an assessment area exhibit the same response upon model
The second draft HREA uses the CMAQ photochemical model instrumented with the higher order direct
decoupled method (HDDM) to estimate ozone concentrations that would occur with the achievement of the current
and alternative O3 standards (U.S. EPA, 2014, Chapter 4).
Exposure and risk analyses for most urban case study areas focus on reducing NOX emissions alone (NOX
emissions were reduced by about 40 to 85% for the current standard, and up to 95% for alternatives). In most of the
urban case study areas, the addition of modeled reductions in VOC emissions did not alter the reductions in NOX
emissions required to simulate the current or alternative standards. However, in Chicago and Denver, the addition of
reductions in VOC emissions allowed for smaller NOX emissions reductions to simulate the current and alternative
standards. Therefore, exposure and risk analyses for Chicago and Denver focus on reductions in emissions of both
NOX and VOC (U.S. EPA, 2014, section 4.3.3.1).
Although this chapter focuses on the current standard, our overarching considerations regarding model-adjusted
air quality also apply to alternative standards simulated in the second draft HREA. Alternative standards are
discussed in chapter 4 of this second draft PA.
3-85
-------
1 adjustment to the current and/or alternative standards. Because model-adjusted air quality
2 scenarios are based on reducing only U.S. anthropogenic emissions, this approach also does not
3 require the specification of background concentrations as a rollback "floor" (U.S. EPA, 2014,
4 section 4.3.3).
5 The use of this model-based air quality adjustment approach in the second draft HREA
6 has important implications for the patterns of ambient Os concentrations estimated in urban case
7 study areas. Specifically, in locations and time periods when NOx is predominantly contributing
8 to Os formation (e.g., downwind of important NOx sources, where the highest Os concentrations
9 often occur), model-based adjustment to the current and alternative standards decreases
10 estimated ambient Os concentrations compared to recent monitored concentrations (U.S. EPA,
11 2014, section 4.3.3.2). In contrast, in locations and time periods when NOx is predominantly
12 contributing to Os titration (e.g., in urban centers with high concentrations of NOx emissions,
CO
13 where ambient Os concentrations are often suppressed and thus relatively low ), model-based
14 adjustment increases ambient 03 concentrations compared to recent measured levels (U.S. EPA,
15 2014, section 4.3.3.2) (Chapter 2, above).
16 Within urban case study areas, the overall impacts of model-based air quality adjustment
17 are to reduce relatively high ambient Os concentrations (i.e., concentrations at the upper ends of
18 ambient distributions) and to increase relatively low Os concentrations (i.e., concentrations at the
19 lower ends of ambient distributions) (U.S. EPA, 2014, section 4.3.3.2, Figures 4-8 to 4-11).
20 Seasonal means of daily concentrations generally exhibit only modest changes upon model
21 adjustment, reflecting the seasonal balance between daily decreases and increases in ambient
22 concentrations (U.S. EPA, 2014, Figures 4-10 and 4-11). The resulting compression in
23 distributions of ambient Os concentrations is evident in all of the urban case study areas
24 evaluated, though the degree of compression varies considerably across areas (U.S. EPA, 2014,
25 Figures 4-10 and 4-11).
26 Adjusted patterns of Os air quality have important implications for exposure and risk
27 estimates in urban case study areas. Estimates influenced largely by the upper ends of the
28 distribution of ambient concentrations (i.e., exposures of concern and lung function risk
29 estimates, as discussed in sections 3.2.2 and 3.2.3.1 below) will decrease with model-adjustment
30 to the current and alternative standards. In contrast, seasonal risk estimates influenced by the full
31 distribution of ambient Os concentrations (i.e., epidemiology-based risk estimates, as discussed
32 in section 3.2.3.2 below) will either increase or decrease in response to model adjustment,
58Titration is also prominent during time periods when photochemistry is limited, such as at night and on cool,
cloudy days.
3-86
-------
1 depending on the balance between the daily decreases in high O^ concentrations and increases in
2 low 63 concentrations.59
3 We further consider the implications of the spatial and temporal patterns of adjusted air
4 quality within the context of exposure (section 3.2.2) and risk (section 3.2.3) estimates for 63
5 concentrations adjusted to just meet the current standard. As discussed below (section 3.2.3.2),
6 these altered patterns are particularly important to consider when interpreting epidemiology-
7 based risk estimates.
8 3.2.2 Exposure-Based Considerations
9 The exposure assessment presented in the second draft HREA (U.S. EPA, 2014) provides
10 estimates of the number of people exposed to various concentrations of ambient 03, while at
11 specified exertion levels. The second draft HREA estimates exposures in 15 urban case study
12 areas for school-age children (ages 5 to 18), asthmatic school-age children, asthmatic adults, and
13 older adults, reflecting the strong evidence indicating that these populations are potentially at
14 increased risk for (Vattributable effects (EPA, 2013, Chapter 8; section 3.1.2, above). An
15 important purpose of these exposure estimates is to provide perspective on the extent to which
16 air quality adjusted to just meet the current 63 NAAQS could be associated with exposures to 63
17 concentrations reported to result in respiratory effects.60 Estimates of such "exposures of
18 concern" provide perspective on the potential public health impacts of Os-related effects,
19 including for effects that cannot currently be evaluated in a quantitative risk assessment (e.g.,
20 airway inflammation).
21 In the absence of large scale exposure studies that encompass the general population, as
22 well as at-risk populations, modeling is the preferred approach to estimating exposures to 63.
23 Additionally, the use of exposure modeling facilitates the estimation of exposures resulting from
24 ambient air concentrations differing from those in exposure studies (e.g., concentrations just
25 meeting the current standard). In the second draft HREA, population exposures to ambient 63
26 concentrations are estimated using the current version of the Air Pollutants Exposure (APEX)
27 model. The APEX model simulates the movement of individuals through time and space and
28 estimates their exposures to a given pollutant in indoor, outdoor, and in-vehicle
29 microenvironments (U.S. EPA, 2014, section 5.1.3). APEX takes into account the most
59In addition, because epidemiology-based risk estimates use "area-wide" average O3 concentrations, calculated by
averaging concentrations across multiple monitors in urban case study areas (section 3.2.3.2 below), risk estimates
on a given day depend on the daily balance between increasing and decreasing O3 concentrations at individual
monitors.
In addition, the range of modeled personal exposures to ambient O3 provide an essential input to the portion of the
health risk assessment based on exposure-response functions (for lung function decrements) from controlled human
exposure studies. The health risk assessment based on exposure-response information is discussed in section 3.2.3,
below.
3-87
-------
1 significant factors that contribute to total human exposure to ambient Os, including the temporal
2 and spatial distributions of people and Os concentrations throughout an urban area, the variation
3 of Os concentrations within various microenvironments, and the effects of exertion on breathing
4 rate in exposed individuals (U.S. EPA, 2014, section 5.1.3). To the extent spatial and/or temporal
5 patterns of ambient Os concentrations are altered upon model adjustment, as discussed above,
6 exposure estimates reflect population exposures to those altered patterns.
7 The second draft HREA estimates 8-hour exposures at or above benchmark
8 concentrations of 60, 70, and 80 ppb for individuals engaged in moderate or greater exertion.
9 Benchmarks reflect exposure concentrations at which Os-induced respiratory effects are known
10 to occur in some healthy adults engaged in moderate, intermittent exertion, based on evidence
11 from controlled human exposure studies (section 3.1.2.1 above and U.S. EPA, 2013, section 6.2).
12 The amount of weight to place on the estimates of exposures at or above specific benchmark
13 concentrations depends in part on the weight of the scientific evidence concerning health effects
14 associated with 03 exposures at that concentration. It also depends on judgments about the
15 importance, from a public health perspective, of the health effects that are known or can
16 reasonably be inferred to occur as a result of exposures at benchmark concentrations (sections
17 3.1.3, 3.1.5 above).
18 As discussed in more detail above (section 3.1.2.1), the health evidence that supports
19 evaluating exposures of concern at or above benchmark concentrations of 60, 70, and 80 ppb
20 comes from a large body of controlled human exposure studies reporting a variety of respiratory
21 effects in healthy adults. The lowest Os exposure concentration for which controlled human
22 exposure studies have reported respiratory effects in healthy adults is 60 ppb, with more
23 evidence supporting this benchmark concentration in the current review than in the last review.
24 In healthy adults, exposures to 60 ppb Os have been reported to decrease lung function and to
25 increase airway inflammation. Exposures of healthy adults to 70 ppb Os have been reported to
26 result in larger lung function decrements, compared to 60 ppb, as well as in increased respiratory
27 symptoms. Exposures of healthy adults to 80 ppb Os have been reported to result in larger lung
28 function decrements than following exposures to 60 or 70 ppb, increased airway inflammation,
29 increased respiratory symptoms, increased airways responsiveness, and decreased lung host
30 defense (section 3.1.2.1, above). As discussed above (section 3.1.3), respiratory effects reported
31 following exposures to Os concentrations of 60, 70, or 80 ppb meet ATS criteria for adverse
32 effects, result in effects judged important by CAS AC in past reviews, and/or could contribute to
33 the clearly adverse effects reported in epidemiologic studies evaluating broader populations.
34 Compared to the healthy individuals included in the studies that provided the basis for the
35 benchmarks, at-risk populations (e.g., asthmatics, children) are more likely to experience larger
36 and/or more serious effects (e.g., U.S. EPA 2013, p. 6-21).
3-88
-------
1 In considering estimates of Os exposures of concern at or above benchmarks of 60, 70,
2 and 80 ppb, within the context of the adequacy of the current standard, we first address the
3 following specific question:
4 • What are the nature and magnitude of the short-term Os exposures of concern
5 remaining upon adjustment of air quality to just meet the current Os standard?
6 In addressing this question, we focus on modeled exposures for school-age children (ages
7 5-18) and asthmatic school-age children, two of the at-risk populations identified in the ISA
8 (section 3.1.5 above). The percentages of children estimated to experience exposures of concern
9 are considerably larger than the percentages estimated for adult populations (i.e., approximately
10 3-fold larger across cities) (U.S. EPA, 2014, Figures 5-5, 5-6, 5-7). The larger exposure
11 estimates for children are due primarily to the larger percentage of children estimated to spend an
12 extended period of time being physically active outdoors (U.S. EPA, 2014, section 5.4.1, Figure
13 5-16).
14 Key results for children are summarized below for air quality adjusted to simulate just
15 meeting the current Oj, NAAQS (Figures 3-7 to 3-10),61 and we note that estimates for all
16 children and asthmatic children are virtually indistinguishable (U.S. EPA, 2014, section 5.3.2).
17 The estimates presented in Figures 3-7 to 3-10 below reflect consistent reductions in estimated
18 exposures of concern across urban case study areas, relative to recent (i.e., unadjusted) air quality
19 (U.S. EPA, 2014, Appendix to Chapter 5). When averaged over the years evaluated in the
20 UREA, reductions of up to about 70% were estimated, compared to recent air quality. These
21 reductions in estimated exposures of concern, relative to unadjusted air quality, reflect the
22 consistent reductions in the highest ambient 63 concentrations upon model adjustment to just
23 meet the current standard (section 3.2.1 above; U.S. EPA, 2014, Chapter 4).
24 Although exposure estimates differ between children and adults, the patterns of results
25 across the cities and years are similar among all of the populations evaluated (U.S. EPA, 2014,
26 Figures 5-12 to 5-15). Therefore, while we highlight estimates in children, we also note that the
27 patterns of exposures estimated for children represent the patterns estimated for adult asthmatics
28 and older adults.
61Figures 3-7 and 3-8 present estimates of one or more exposures of concern. Figures 3-9 and 3-10 present estimates
of two or more exposures of concern.
3-89
-------
£
o
E
20
18
16
O)
c
"o
c
0>
~i_
0)
Q.
X
LLJ
E
0)
2
IE
O
•s
1
S
o!
E
o>
0
c
o
O
•5
en
o>
3
c/>
o
Q.
X
LU
14
12
10
8
6
4
2
I 60 ppb
I 70 ppb
80 ppb
2 Figure 3-7. Percent of children estimated to experience one or more exposures of concern at or above 60, 70, 80 ppb with air
3 quality adjusted to just meet the current standard (averaged over 2006 to 2010).
3-90
-------
I 60 | r| >l >
I 70 ppb
80 ppb
*? & rf1
-------
160 ppb
70 ppb
80 ppb
K\*
XF
2 Figure 3-9. Percent of children estimated to experience two or more exposures of concern at or above 60, 70, 80 ppb with air
3 quality adjusted to just meet the current standard (Averaged over 2006 to 2010)
3-92
-------
2
O
4
5
^ */ «
• 60 ppb
• 70 ppb
- 80 ppb
Figure 3-10. Percent of children estimated to experience two or more exposures of concern at or above 60, 70, 80 ppb with air
quality adjusted to just meet the current standard (Worst-Case Year, 2006 to 2010)
3-93
-------
1 Based on Figures 3-7 to 3-10 and the associated details described in the second draft
2 HREA (U.S. EPA 2014, Chapter 5), we take note of the following with regard to exposures that
3 are estimated to be allowed by the current standard:
4 1. For exposures of concern at or above 60 ppb:
5 a. On average over the years 2006 to 2010, the current standard is estimated to allow
6 approximately 10 to 18% of children in urban case study areas to experience one or
7 more exposures of concern at or above 60 ppb. Summing across urban case study
8 areas, these percentages correspond to almost 2.5 million children experiencing
9 approximately 4 million exposures of concern at or above 60 ppb during a single 63
10 season. Of these children, almost 250,000 are asthmatics.
11
12 b. On average over the years 2006 to 2010, the current standard is estimated to allow
13 approximately 3 to 8% of children in urban case study areas to experience two or
14 more exposures of concern to Os concentrations at or above 60 ppb. Summing across
15 the urban case study areas, these percentages correspond to almost 900,000 children
16 (including almost 90,000 asthmatic children) estimated to experience at least two Oj
17 exposure concentrations at or above 60 ppb during a single Os season.
18
19 c. In the worst-case years (i.e., those with the largest exposure estimates), the current
20 standard is estimated to allow approximately 10 to 25% of children to experience one
21 or more exposures of concern at or above 60 ppb, and approximately 4 to 14% to
22 experience two or more exposures of concern at or above 60 ppb.
23
24 2. For exposures of concern at or above 70 ppb:
25 a. On average over the years 2006 to 2010, the current standard is estimated to allow up
26 to approximately 3% of children in urban case study areas to experience one or more
27 exposures of concern at or above 70 ppb. Summing across urban case study areas,
28 almost 400,000 children (including almost 40,000 asthmatic children) are estimated to
29 experience Os exposure concentrations at or above 70 ppb during a single Os season.
30
31 b. On average over the years 2006 to 2010, the current standard is estimated to allow
32 less than 1% of children in urban case study areas to experience two or more
33 exposures of concern to Os concentrations at or above 70 ppb.
34
35 c. In the worst-case years, the current standard is estimated to allow approximately 1 to
36 8% of children to experience one or more exposures of concern at or above 70 ppb,
37 and up to approximately 2% to experience two or more exposures of concern, at or
38 above 70 ppb.
39
40 3. For exposures of concern at or above 80 ppb: The current standard is estimated to allow
41 about 1% or fewer children in urban case study areas to experience exposures of concern at
42 or above 80 ppb, even in years with the highest exposure estimates.
3-94
-------
1 In further evaluating estimated exposures of concern from the 2n draft HREA, we next
2 consider the following question:
3 • What are the important sources of uncertainty associated with exposure estimates?
4 Due to variability in responsiveness, only a subset of individuals who experience
5 exposures at or above a benchmark concentration can be expected to experience health effects.
6 Given the lack of sufficient exposure-response information for most of the health effects that
7 informed benchmark concentrations, estimates of the number of people likely to experience
8 exposures at or above benchmark concentrations generally cannot be translated into quantitative
9 estimates of the number of people likely to experience specific health effects.62 We view health-
10 relevant exposures as a continuum with greater confidence and less uncertainty about the
11 existence of health effects at higher Oj, exposure concentrations, and less confidence and greater
12 uncertainty as one considers lower exposure concentrations. This view draws from the overall
13 body of available health evidence, which indicates that as exposure concentrations increase the
14 incidence, magnitude, and severity of effects increases.
15 Though we have less confidence in the likelihood of adverse health effects as Os
16 exposure concentrations decrease, we also note that the controlled human exposure studies that
17 provided the basis for health benchmark concentrations have not evaluated at-risk populations.
18 Compared to the healthy individuals included in controlled human exposure studies, members of
19 at-risk populations (e.g., asthmatics, children) could be more likely to experience adverse effects,
20 could experience larger and/or more serious effects, and/or could experience effects following
21 exposures to lower Os concentrations. In considering estimated exposures of concern within the
22 context of drawing conclusions on the adequacy of the current standard (section 3.4, below), we
23 balance concerns about the potential for adverse health effects, including effects in at-risk
24 populations, with our increasing uncertainty regarding the likelihood of such effects following
25 exposures to lower 63 concentrations.
26 Uncertainties associated with the APEX exposure modeling also have the potential to be
27 important in our consideration of the adequacy of the current standard. For example, the HREA
28 concludes that exposures of concern could be underestimated for some individuals who are
29 frequently and routinely active outdoors during the warm season (U.S. EPA, section 5.5.2). This
30 could include outdoor workers and children who are frequently active outdoors. The HREA
31 specifically notes that long-term diary profiles (i.e., monthly, annual) do not exist for such
32 populations, limiting the extent to which APEX outputs reflect people who follow similar daily
33 routines resulting in high exposures, over extended periods of time. Thus, exposure estimates
The exception to this is lung function decrements, as discussed below (section 3.2.3.1).
3-95
-------
1 generated from the general pool of available diary profiles likely do not reflect the most highly
2 exposed individuals in the population.
3 In order to evaluate the potential implications of this uncertainty for exposure estimates,
4 the second draft HREA reports the results of limited sensitivity analyses using subsets of diaries
5 specifically selected to reflect groups spending a larger proportion of time being active outdoors
6 during the Os season. When diaries were selected to mimic exposures that could be experienced
7 by outdoor workers, the percent of modeled individuals estimated to experience exposures of
8 concern increased compared to other adult populations evaluated. The percent of outdoor
9 workers estimated to experience exposures of concern were generally similar to the percentages
10 estimated for children (i.e., using the full database of diary profiles) in the worst-case cities and
11 years (i.e., cities and years with the highest exposure estimates) (U.S. EPA, 2014, Figure 5-11).
12 In addition, when diaries were restricted to children who did not report any time spent inside a
13 school or performing paid work (i.e., to mimic children spending particularly large portions of
14 their time outdoors during the summer), the number experiencing exposures of concern increased
15 by approximately 30% (U.S. EPA, 2014, section 5.3.3). Though these sensitivity analyses are
16 limited to single urban case study areas, and though there is uncertainty associated with diary
17 selection approaches to mimic highly exposed populations, they suggest the possibility that some
18 portions of the population could experience more frequent exposures of concern than indicated
19 by estimates based on the full database of activity diary profiles.
20 In further considering activity diaries, the HREA also notes that growing evidence
21 indicates that people can change their behavior in response to high Os concentrations, reducing
22 the time spent being active outdoors (U.S. EPA, 2014, section 5.4.3). Commonly termed
23 "averting behaviors," these altered activity patterns could reduce personal exposure
24 concentrations. Therefore, the second draft HREA also performed limited sensitivity analyses to
25 evaluate the potential implications of averting behavior for estimated exposures of concern.
26 These analyses suggest that averting behavior could reduce the percentages of children estimated
27 to experience exposures of concern at or above the 60 or 70 ppb benchmark concentrations by
28 approximately 10 to 30%, with larger reductions possible for the 80 ppb benchmark (U.S. EPA,
29 2014, Figure 5-12). As discussed above for other sensitivity analyses, these analyses are limited
30 to a single urban case study area and are subject to uncertainties associated with assumptions
31 about the prevalence and duration of averting behaviors. However, the results suggest that
32 exposures of concern could be overestimated, particularly in children (Neidell et al., 2009; U.S.
33 EPA, 2013, Figures 4-7 and 4-8), if the possibility for averting behavior is not incorporated into
34 estimates.
3-96
-------
1 3.2.3 Risk-Based Considerations
2 For some health endpoints, there is sufficient scientific evidence and information
3 available to support the development of quantitative estimates of Os-related health risks. In the
4 last review of the Os NAAQS, the quantitative health risk assessment estimated Cb-related lung
5 function decrements, respiratory symptoms, respiratory-related hospital admissions, and non-
6 accidental and cardiorespiratory-related mortality (U.S. EPA, 2007). In those analyses, both
7 controlled human exposure and epidemiologic studies were used for the quantitative assessment
8 of Os-related human health risks.
9 In the current review, for short-term 63 concentrations the second draft HREA estimates
10 lung function decrements; respiratory symptoms in asthmatics; hospital admissions and
11 emergency department visits for respiratory causes; and all-cause mortality (U.S. EPA, 2014).
12 For "long-term" Os concentrations, the second draft UREA estimates respiratory mortality (U.S.
13 EPA, 2014).63 Estimates of Os-induced lung function decrements are based on exposure
14 modeling, as noted above, combined with exposure-response relationships from controlled
15 human exposure studies (U.S. EPA, 2014, Chapter 6). Estimates of Cb-associated respiratory
16 symptoms; hospital admissions and emergency department visits; and mortality are based on
17 concentration-response relationships from epidemiologic studies (U.S. EPA, 2014, Chapter 7).
18 As with the exposure assessment discussed above, Os-associated health risks are estimated for
19 recent air quality and for ambient concentrations model-adjusted to just meet the current 8-hour
20 Os NAAQS, based on 2006-2010 air quality and adjusted precursor emissions.
21 Section 3.2.3.1 below discusses risk results for Os-induced lung function decrements
22 following short-term exposures, based on exposure modeling results and exposure-response
23 relationships from controlled human exposure studies. Section 3.2.3.2 discusses epidemiology-
24 based risk estimates, with a focus on all-cause mortality (short-term Oj, concentrations);
25 respiratory-related morbidity outcomes (short-term 63 concentrations); and respiratory mortality
26 (long-term O^ concentrations).
27 3.2.3.1 Risk of Lung Function Decrements
28 In the last review, EPA conducted a health risk assessment that produced risk estimates
29 for the number and percent of school-aged children, asthmatic school-aged children, and the
30 general population experiencing lung function decrements associated with Os exposures for 12
31 urban areas. These estimates were based on exposure-response relationships developed from
Risk estimates for "long-term" concentrations are based on the concentration-response relationship identified in
the study by Jerrett et al. (2009). As discussed above, study authors used April to September averages of 1-hour
daily maximum O3 concentrations as surrogates for "long-term" exposures.
3-97
-------
1 analysis of data from several controlled human exposure studies, combined with exposure
2 estimates developed for children and adults (U.S. EPA, 2007a).
3 In the current review, the second draft HREA estimates risks of lung function decrements
4 in school-aged children (ages 5 to 18), asthmatic school-aged children, and the general adult
5 population for 15 urban case study areas.64 The results presented in the second draft HREA are
6 based on an updated dose-threshold model that estimates FEVi responses for individuals
7 following short-term exposures to 63 (McDonnell, Stewart, and Smith, 2010), reflecting
8 methodological improvements since the last review (U.S. EPA, 2014, section 6.2.4). The impact
9 of the dose threshold is that Ch-induced FEVi decrements result primarily from exposures on
10 days with ambient 63 concentrations above about 40 ppb (U.S. EPA, 2013, Chapter 6).
11 As discussed above (section 3.1.3), for effects such as lung function decrements, which
12 are transient and reversible, aspects such as the likelihood that these effects would occur
13 repeatedly and would interfere with normal activities are important to consider in making
14 judgments about adversity to individuals. As stated in the 2006 Criteria Document (Table 8-3,
15 p.8-68), for people with lung disease even moderate functional responses (e.g., FEVi decrements
16 > 10% but < 20%) would likely interfere with normal activities for many individuals, and would
17 likely result in more frequent medication use. Moreover, as noted above, in a recent letter to the
18 Administrator, the CASAC 63 Panel stated that '"[c]linically relevant effects are decrements >
19 10%, a decrease in lung function considered clinically relevant by the American Thoracic
20 Society" (Samet, 2011, p.2). The CASAC O3 Panel also stated that:
21 [A] 10% decrement in FEVi can lead to respiratory symptoms, especially in
22 individuals with pre-existing pulmonary or cardiac disease. For example, people
23 with chronic obstructive pulmonary disease have decreased ventilatory reserve
24 (i.e., decreased baseline FEVi) such that a > 10% decrement could lead to
25 moderate to severe respiratory symptoms (Samet, 2011, p.7).
26 In judging the extent to which moderate lung function decrements represent effects that
27 should be regarded as adverse to the health status of individuals, in previous NAAQS reviews we
28 have also considered the extent to which decrements were experienced repeatedly during the
29 course of a year (Staff Paper, U.S. EPA, 2007). Although some experts would judge single
30 occurrences of moderate responses to be a "nuisance," especially for healthy individuals, a
31 more general consensus view of the adversity of such moderate responses emerges as the
32 frequency of occurrence increases. Thus in the past EPA has judged that repeated occurrences of
33 moderate responses, even in otherwise healthy individuals, may be considered to be adverse
64
As noted for the exposure assessment above, the 15 cities assessed are Atlanta, Baltimore, Boston, Chicago,
Cleveland, Dallas, Denver, Detroit, Houston, Los Angeles, New York, Philadelphia, Sacramento, St. Louis, and
Washington, DC.
3-98
-------
1 since they could well set the stage for more serious illness (61 FR 65723). The CAS AC panel in
2 the 1997 NAAQS review expressed a consensus view that these "criteria for the determination of
3 an adverse physiological response were reasonable" (Wolff, 1995). In the review completed in
4 2008, estimates of repeated occurrences continued to be an important public health policy factor
5 in judging the adversity of moderate lung function decrements in healthy and asthmatic
6 populations (72 FR 37850, July 11, 2007).
7 The second draft HREA estimates risks of moderate to large lung function decrements,
8 defined as FEVi decrements > 10%, > 15%, or > 20%. In evaluating these lung function risk
9 estimates within the context of considering the adequacy of the current 63 standard, we first
10 consider the following specific question:
11 • What are the nature and magnitude of lung function risks remaining upon just meeting
12 the current Os standard?
13 In considering risks of Cb-induced FEVi decrements, we focus on the percent of children
14 estimated to experience decrements > 10, 15, and 20%, noting that the percentage of asthmatic
15 children estimated to experience such decrements is virtually the same as the percentage
16 estimated for all children. Compared to children, only a very small percentage of adults were
17 estimated to experience Ch-induced FEVi decrements (U.S. EPA, 2014, Appendix 6B). As for
18 exposures of concern (see above), the patterns of results across urban case study areas and over
19 the years evaluated are similar in children and adults (U.S. EPA, 2014, Chapter 6). Therefore,
20 while we highlight estimates in children, we note that these results are also representative of the
21 patterns estimated for adult populations.
22 Key results for children are summarized below for air quality adjusted to just meet the
23 current 63 NAAQS (Figures 3-11 to 3-14).65 The estimates presented in Figures 3-11 to 3-14
24 below reflect consistent reductions across urban case study areas in the percent of children
25 estimated to experience Os-induced lung function decrements, relative to recent (i.e., unadjusted)
26 air quality (U.S. EPA, 2014, Appendix to Chapter 6). When averaged over the years evaluated
27 in the UREA, reductions of up to about 40% were estimated compared to recent air quality.
28 These reductions reflect the consistent decreases in relatively high ambient 63 concentrations
29 upon adjustment to just meet the current standard (section 3.2.1 above; U.S. EPA, 2014, Chapter
30 4).66
65Figures 3-11 and 3-12 present estimates of one or more decrements. Figures 3-13 and 3-14 present estimates of
two or more decrements.
As noted above, the impact of the dose threshold is that O3-induced FEVi decrements result primarily from days
with ambient O3 concentrations above about 40 ppb (U.S. EPA, 2013, Chapter 6).
3-99
-------
I > 10%
i > 20%
2
O
4
Figure 3-11. Percent of school-aged children (5-18 yrs) estimated to experience one or more days with FEVi decrements > 10,
15, or 20% with air quality adjusted to just meet the current standard (Averaged over 2006 to 2010)
3-100
-------
I > 10%
i > 20%
4
5
Figure 3-12. Percent of school-aged children (5-18 yrs) estimated to experience one or more days with FEVi decrements > 10,
15, or 20% with air quality adjusted to just meet the current standard (Worst-Case Year from 2006 to 2010)
3-101
-------
I > 10%
I > 1596
, > 20%
2 Figure 3-13. Percent of school-aged children (5-18 yrs) estimated to experience two or more days with FEVi decrements > 10,
3 15, or 20% with air quality adjusted to just meet the current standard (Averaged over 2006 to 2010)
3-102
-------
I • 10%
•20%
2 Figure 3-14. Percent of school-aged children (aged 5-18 yrs) estimated to experience two or more days with FEVi decrements >
3 10,15, or 20% with air quality adjusted to just meet the current standard (Worst-Case Year from 2006 to 2010)
3-103
-------
1 Based on Figures 3-11 to 3-14 and the associated details described in the second draft
2 HREA (U.S. EPA 2014, Chapter 6), we take note of the following with regard to lung function
3 decrements estimated to be allowed by the current standard:
4 1. For FEVi decrements > 10%:
5 a. On average over the years 2006 to 2010, the current standard is estimated to allow
6 approximately 14 to 19% of children in urban case study areas to experience one or
7 more lung function decrements > 10%. Summing across urban case study areas, this
8 corresponds to approximately 3 million children experiencing 15 million Os-induced
9 lung function decrements > 10% during a single Os season. Of these children, about
10 300,000 are asthmatics.
11
12 b. On average over the years 2006 to 2010, the current standard is estimated to allow
13 approximately 7 to 12% of children in urban case study areas to experience two or
14 more Os-induced lung function decrements > 10%. Summing across the urban case
15 study areas, this corresponds to almost 2 million children (including almost 200,000
16 asthmatic children) estimated to experience two or more Os-induced lung function
17 decrements greater than 10% during a single Os season.
18
19 c. In the worst-case years, the current standard is estimated to allow approximately 17 to
20 23% of children in urban case study areas to experience one or more lung function
21 decrements > 10%, and approximately 10 to 14% to experience two or more Os-
22 induced lung function decrements > 10%.
23
24 2. For FEVi decrements > 15%:
25 a. On average over the years 2006 to 2010, the current standard is estimated to allow
26 approximately 3 to 5% of children in urban case study areas to experience one or
27 more lung function decrements > 15%. Summing across urban case study areas, this
28 corresponds to approximately 800,000 children (including approximately 80,000
29 asthmatic children) estimated to experience at least one Os-induced lung function
30 decrement > 15% during a single Oj season.
31
32 b. On average over the years 2006 to 2010, the current standard is estimated to allow
33 approximately 2 to 3% of children in urban case study areas to experience two or
34 more Os-induced lung function decrements > 15%.
35
36 c. In the worst-case years, the current standard is estimated to allow approximately 4 to
37 6% of children in urban case study areas to experience one or more lung function
38 decrements > 15%, and approximately 2 to 4% to experience two or more Os-induced
39 lung function decrements > 15%.
40
41 3. For FEVi decrements > 20%:
42 a. On average over the years 2006 to 2010, the current standard is estimated to allow
43 approximately 1 to 2% of children in urban case study areas to experience one or
44 more lung function decrements > 20%. Summing across urban case study areas, this
45 corresponds to approximately 300,000 children (including approximately 30,000
3-104
-------
1 asthmatic children) estimated to experience at least one Cb-induced lung function
2 decrement > 20% during a single Os season.
3
4 b. On average over the years 2006 to 2010, the current standard is estimated to allow
5 less than 1% of children in urban case study areas to experience two or more O^-
6 induced lung function decrements > 20%.
7
8 c. In the worst-case years, the current standard is estimated to allow approximately 2 to
9 3% of children to experience one or more lung function decrements > 20%, and less
10 than 2% to experience two or more Cb-induced lung function decrements > 20%.
11 In further considering estimated lung function risks from the 2n draft HREA, we next
12 consider the following question:
13 • What are the important sources of uncertainty associated with lung function risk
14 estimates?
15 In addition to the uncertainties noted above for exposure estimates, the HREA identifies
16 several key uncertainties associated with estimates of Cb-induced lung function decrements. An
17 uncertainty with particular potential to impact our consideration of risk estimates in this Policy
18 Assessment stems from the lack of exposure-response information in children. In the absence of
19 controlled human exposure data for children, risk estimates are based on the assumption that
20 children exhibit the same lung function response following 63 exposures as healthy 18 year olds
21 (i.e., the youngest age for which controlled human exposure data is available) (U.S. EPA, 2014,
22 section 6.2.4 and 6.5). This assumption was justified in part by the findings of McDonnell et al.
23 (1985), who reported that children 8-11 year old experienced FEVi responses similar to those
24 observed in adults 18-35 years old. In addition, as discussed in the ISA (U.S. EPA, 2013, section
25 6.2.1), summer camp studies of school-aged children reported Os-induced lung function
26 decrements similar in magnitude to those observed in controlled human exposure studies using
27 adults. In extending the risk model to children, the second draft HREA fixes the age term in the
28 model at its highest value, the value for age 18. This approach could result in either over- or
29 underestimates of Os-induced lung function decrements in children, depending on how children
30 compare to the adults used in controlled human exposure studies (U.S. EPA, 2014, section 6.5).
31 A related source of uncertainty is that the risk assessment estimates Os-induced
32 decrements in asthmatics using the exposure-response relationship developed from data collected
33 from healthy individuals. Although the evidence has been mixed (U.S. EPA, 2013, section
34 6.2.1.1), several studies have reported larger Os-induced lung function decrements in asthmatics
35 than in non-asthmatics (Kreit et al., 1989; Horstman et al., 1995; Torres et al., 1996; Alexis et al.,
36 2000). To the extent asthmatics experience larger Os-induced lung function decrements than the
37 healthy adults used to develop exposure-response relationships, the second draft HREA could
3-105
-------
1 underestimate the impacts of O^ exposures on lung function in asthmatics, including asthmatic
2 children. The HREA notes that the magnitude this uncertainty might have on risk estimates
3 remains unknown at this time (U.S. EPA, 2014, Chapter 6).
4 3.2.3.2 Estimated Health Risks Associated with Short- or Long-Term Os Exposures,
5 Based on Epidemiologic Studies
6 Risk estimates based on epidemiologic studies can provide perspective on the most
7 serious (^-associated public health outcomes (e.g., mortality, hospital admissions, emergency
8 department visits) in populations that include at-risk groups. The second draft HREA estimates
9 Os-associated risks in 12 urban case study areas67 using concentration-response relationships
10 drawn from epidemiologic studies. These concentration-response relationships are based on
/-Q
11 "area-wide" average Os concentrations. The HREA estimates risks for the years 2007 and 2009
12 in order to provide estimates of risk for a year with generally higher Os concentrations (2007)
13 and a year with generally lower 63 concentrations (2009) (U.S. EPA, 2014, section 7.2).
14 In the last review, epidemiologic-based risks were estimated for Os concentrations above
15 mean "policy-relevant background concentrations." As discussed above (Chapter 2), policy-
16 relevant background (PRB) concentrations were defined as the distribution of ozone
17 concentrations that would be observed in the U.S. in the absence of anthropogenic (man-made)
18 emissions of ozone precursor emissions (e.g., VOC, CO, NOx) in the U.S., Canada, and Mexico.
19 This approach provided a focus on Os concentrations "that can be controlled by U.S. regulations
20 (or through international agreements with neighboring countries)" (U.S. EPA, 2007, pp. 2-48 to
21 2-54).
22 As in the last review, we recognize that ambient Os concentrations, and therefore Os-
23 associated health risks, result from precursor emissions from various types of sources. Based on
24 the air quality modeling discussed above in chapter 2, approximately 30 to 60% of average
25 daytime O^ during the warm season (i.e., 8-hour daily maximum concentrations averaged from
26 April to October) is attributable to precursor emissions from U.S. anthropogenic sources (section
27 2.4.4). This suggests that, for recent air quality (i.e., not adjusted to meet the current or
28 alternative standards), approximately 30 to 60% of total Os-associated health risk in the urban
The 12 urban areas evaluated are Atlanta, Baltimore, Boston, Cleveland, Denver, Detroit, Houston, Los Angeles,
New York, Philadelphia, Sacramento, and St. Louis.
68In the epidemiologic studies that provide the health basis for HREA risk assessments, concentration-response
relationships are based on daytime O3 concentrations, averaged across multiple monitors within study areas. These
daily averages are used as surrogates for the spatial and temporal patterns of exposures in study populations.
Consistent with this approach, the HREA epidemiologic-based risk estimates also utilize daytime O3 concentrations,
averaged across monitors, as surrogates for population exposures. In this second draft PA, we refer to these averaged
concentrations as "area-wide" O3 concentrations. Area-wide concentrations are discussed in more detail in section
3.1.4, above.
3-106
-------
1 case study areas is attributable to O^ from U.S. anthropogenic emissions. The remainder is
2 attributable to precursor emissions from international anthropogenic sources and natural sources.
3 Because the second draft HREA characterizes health risks from all 03, regardless of source, risk
4 estimates reflect emissions from U.S. anthropogenic, international anthropogenic, and natural
5 sources. Given that HREA risk estimates for adjusted air quality are based on decreasing U.S.
6 anthropogenic precursor emissions, the contributions of U.S. anthropogenic emissions to the risk
7 estimates for the current standard would generally be smaller than the 30 to 60% indicated for
8 recent air quality.
9 In evaluating epidemiology-based risk estimates within the context of the adequacy of the
10 current standard, we first consider the following question:
11 • What are the nature and magnitude of the Os-associated mortality and morbidity risks
12 remaining upon adjustment of air quality to just meet the current Os standard?
13 In addressing this question, as an initial matter we note that the area-wide average 63
14 concentrations associated with health effects in epidemiologic studies, and used to estimate
15 mortality and morbidity risks in the HREA, are surrogates for the ambient O^ exposures expected
16 to have elicited the reported health outcomes (also discussed in section 3.1.4.2, above). The area-
17 wide average concentrations present in epidemiologic study locations represent the spatial and
18 temporal patterns of Os exposures (magnitudes, frequencies, durations of exposures) experienced
19 by study populations. Differences in the patterns of 63 exposures would be expected to result in
20 differences in the health outcome response. Thus, in considering the quantitative risk estimates
21 below we are mindful of uncertainties related to the differences between the spatial and temporal
22 patterns of 63 that existed in the epidemiologic study areas, which contributed to the health
23 outcomes reported in these studies, and the altered patterns associated with adjusted air quality
24 that just meets the current standard (section 3.2.1, above). Among the three main types of
25 exposure/risk analyses generated in the HREA, these altered spatial/temporal patterns have the
26 greatest potential to introduce uncertainty into risk estimates based on epidemiology study
27 concentration-response relationships (see 2nd question below for further discussion).
28 We also note that the second draft HREA estimates mortality and morbidity risks
29 associated with just meeting the current standard by applying concentration-response
30 relationships from epidemiologic studies to the entire distributions of model-adjusted "area-
31 wide" average 03 concentrations present in urban case study areas (U.S. EPA, 2014, Chapter 7).
32 Implicit in this approach to estimating risks is the assumption that concentration-response
33 relationships are linear over those distributions. Therefore, as noted in section 3.2.1, when air
34 quality is adjusted to just meet the current standard, risk estimates are influenced by the
35 decreases in area-wide 63 concentrations at the upper ends of warm season distributions and the
3-107
-------
1 increases in area-wide O^ concentrations at the lower ends of those distributions (U.S. EPA,
2 2014, section 4.3.3.2, Figures 4-8 to 4-11).69 When the decreases and increases are of the same
3 magnitude, they result in the same degree of change in estimated risks, though opposite in
4 direction. Therefore, seasonal estimates of (Vassociated mortality and morbidity risks either
5 increase or decrease in response to air quality adjustment, depending on the seasonal balance
6 between the modeled daily decreases in high area-wide O?, concentrations and increases in low
7 area-wide 63 concentrations. One consequence is that the estimated impacts on mortality and
8 morbidity risks of adjusting air quality to just meet the current standard are more modest, and
9 less directionally consistent across urban case study areas, than on either exposures of concern or
10 Os-induced lung function decrements.
11 In the remainder of this section, we consider estimates of total (non-accidental) mortality
12 and respiratory morbidity associated with short-term 63 concentrations, and respiratory mortality
13 associated with "long-term" Os concentrations.
14 Total Mortality - Short-Term O3
15 Risk estimates for total mortality are based on concentration-response relationships
16 described by Smith et al. (2009). To generate risk estimates, the second draft HREA uses "area-
17 wide" averages of 8-hour daily maximum Oi concentrations over the full monitoring periods in
18 urban case study areas. These monitoring periods vary across areas, in some cases including
19 more of the cooler months that are often characterized by relatively low daytime 63
20 concentrations.70 When air quality was adjusted to the current standard in the 2007 model year
21 (the year with generally "higher" (^-associated risks), 10 of 12 urban case study areas exhibited
22 either small decreases or virtually no change in estimates of (^-associated total mortality (U.S.
23 EPA, 2014, Appendix to Chapter 7).71 Small increases in mortality were estimated in two of the
24 urban case study areas (Houston, Los Angeles) (U.S. EPA, 2014, Appendix to Chapter 7).
25 Figure 3-15 below presents estimates of (Vassociated all-cause mortality in urban case
26 study areas for 2007 and 2009, with air quality adjusted to just meet the current Os standard. The
27 second draft HREA estimates that upon just meeting the current standard, 63 could be associated
28 with from 0.8 to 4.1% of all-cause mortality across the urban case study areas. This corresponds
69On a given day, area-wide O3 concentrations and estimated risks decrease when the sum of the changes at monitors
with decreasing O3 are larger than the sum of the changes at monitors with increasing O3. Area-wide O3
concentrations and estimated risks increase when the opposite occurs.
70Decreases in relatively higher ambient O3 concentrations are more prominent during the warmest months, when
daytime concentrations tend to be highest. In most urban case study areas, increases in relatively low daytime
concentrations have greater influence on risk estimates during cooler months, when O3 concentrations tend to be
lower overall (U.S. EPA, 2014, compare Figures 4-8 and 4-9).
71Decreases were smaller in the 2009 model-adjusted year (i.e., the year with generally lower O3 concentrations).
3-108
-------
1 to approximately 80 to 2,800 Os-associated deaths per season in individual urban case study
2 areas, and approximately 8,000 to 9,000 Os-associated deaths per season summed over the 12
3 urban case study areas (U.S. EPA, 2014, Tables 7-7 and 7-8).
4.5
>•
4
4
5
6
9
10
11
12
13
14
15
16
17
18
19
(T3
•e
o
0)
t/>
3
ro
O
I
=
<
M—
o
0)
o
0)
Q.
£
o
2 1
2.5
I i I n i i i i i i
12007
2009
0°
Figure 3-15. Percent of all-cause mortality associated with Os for air quality adjusted to
just meet the current standard.
In considering the risk estimates presented in Figure 3-15, which are based on applying
linear concentration-response relationships to the full distributions of daily 8-hour "area-wide"
Os concentrations, we note the ISA conclusion that there is less certainty in the shape of
concentration-response functions for area-wide 63 concentrations at the lower ends of warm
season distributions (i.e., below about 20 to 40 ppb depending on the Os metric, health endpoint,
and study population) (U.S. EPA, 2013, section 2.5.4.4). We also recognize that for the range of
health endpoints evaluated, controlled human exposure and animal toxicological studies provide
greater certainty in the increased incidence, magnitude, and severity of effects at higher exposure
concentrations (discussed in sections 3.1.2.2 and 3.1.4.2, above).72 Thus, in addition to
considering estimates of total (Vassociated risks, we also consider the extent to which risks are
associated with days with higher, versus lower, area-wide O^ concentrations.
Figure 3-16 presents risk estimates, summed across urban case study areas, for days with
area-wide concentrations at or above 20, 40, and 60 ppb. Daytime Os concentrations in the upper
72As discussed in section 3.1.4.2, as ambient concentrations increase the potential for exposures to higher O3
concentrations also increases. Thus with increasing ambient concentrations, controlled human exposure and animal
toxicological studies provide greater certainty in the increased incidence, magnitude, and severity of O3-attributable
effects.
3-109
-------
1 portion of the distribution of area-wide concentrations (e.g., at or above 40 or 60 ppb) are
2 estimated to be associated with hundreds to thousands of deaths per year in urban case study
areas.
73
10000
-^ 9000
I Total risk
120+ ppb
140+ ppb
60- ppb
2007
2009
5 Figure 3-16. Estimated Os-associated mortality attributable to days above various area-
6 wide average Os concentrations, with air quality adjusted to just meet current
7 standard (2007 Model Adjustment Year).
8 Respiratory Mortality - "Long-Term" Os
9 The second draft HREA estimates the risk of respiratory mortality associated with long-
10 term O3 exposures, based on the study by Jerrett et al. (2009) (U.S. EPA, 2014, Chapter 7). To
11 generate risk estimates, the second draft HREA uses "area-wide" averages of 1-hour daily
12 maximum Os concentrations during the warm season (April to September). When air quality was
13 adjusted to just meet the current standard 11 of the 12 urban case study areas exhibited modest
14 decreases in estimated (^-associated respiratory mortality (i.e., compared to recent, unadjusted
15 air quality). Risk estimates remained virtually unchanged in the remaining urban case study area
16 (i.e., Los Angeles). Risk estimates for air quality adjusted to just meet the current standard are
17 presented below in Figure 3-17.74
18
73The relatively small proportion of O3-associated deaths attributable to days with area-wide concentrations of 60
ppb or greater reflects the relatively small proportion of days with such elevated area-wide concentrations.
74The second draft HREA does not characterize distributions of respiratory mortality risks over distributions of
ambient O3 concentrations. Therefore, in considering respiratory mortality risks we evaluate only estimates of total
risk.
3-110
-------
25
0
\^ &
^
o^
12007
12009
2 Figure 3-17. Percent of baseline respiratory mortality estimated to be associated with long-
3 term Os.
4 Across urban case study areas, O^ is estimated to be associated with approximately 16 to
5 20% of respiratory mortality during the warm season. This corresponds to approximately 500 to
6 2,800 Os-associated deaths per season across areas, and a total of approximately 12,000 Os-
7 associated deaths in all 12 urban case study areas.
8 Hospital Admissions, Emergency Department Visits, and Asthma Exacerbations
9 Risk estimates for respiratory-related hospital admissions, emergency department visits,
10 and asthma exacerbations associated with air quality adjusted to just meet the current standard
11 are based on several studies, as presented in Table 7-2 of the second draft HREA (U.S. EPA,
12 2014).75 Estimates indicate that (Vassociated respiratory-related hospital admissions account for
13 approximately 2 to 3% of total respiratory-related admissions in urban case study locations.
14 Depending on the city, this corresponds to 10's to 100's of (^-associated hospital admissions per
15 season. Estimates indicate that (^-associated respiratory-related emergency department visits
16 account for approximately 3 to 20% of total respiratory-related emergency department visits in
75As with respiratory mortality above, the second draft HREA does not characterize distributions of respiratory
morbidity risks over distributions of ambient O3 concentrations. Therefore, in considering respiratory morbidity
risks we evaluate estimates of total risk.
3-111
-------
1 Atlanta (approximately 4,000 to 8,000 visits per season), and that (^-associated asthma
2 exacerbations account for approximately 15 to 30% of total exacerbations in Boston (45,000 to
3 130,000 exacerbations per season). Full estimates are presented in Tables 7-9 to 7-11 in the
4 second draft HREA (U.S. EPA, 2014).
5 Based on Figures 3-15 to 3-17 above, and the more detailed information presented in
6 Chapter 7 of the second draft HREA (U.S. EPA, 2014), we note the following key observations:
7 1. In focusing on total risk, the current standard is estimated to allow thousands of 63-
8 associated deaths per year in the urban case study areas. These estimates are based on
9 concentration-response functions from epidemiologic studies that used either 8-hour daily
10 63 concentrations (total mortality associated with short-term 63) or seasonal averages of
11 1-hour daily 63 concentrations (respiratory mortality associated with long-term 63).
12
13 2. In focusing on the risks associated with the upper portions of distributions of ambient
14 concentrations, the current standard is estimated to allow hundreds to thousands of 63-
15 associated deaths per year in the urban case study areas. These estimates are based on
16 concentration-response functions from an epidemiologic study that evaluated associations
17 between 8-hour daily OT, concentrations and total mortality.
18
19 3. In urban case study areas, the current standard is estimated to allow tens to thousands of
20 Os-associated morbidity events per year. Distributions of O^-associated morbidity over
21 distributions of ambient OT, concentrations would likely be similar to mortality, though
22 the second draft HREA did not analyze such distributions for morbidity endpoints.
23 In further considering estimated (^-associated mortality and morbidity risks from the 2nd
24 draft HREA, we next consider the following question:
25 • What are the important sources of uncertainty associated with mortality and morbidity
26 risk estimates?
27 Upon adjusting air quality to the current standard, (^-associated mortality and morbidity
28 risks generally decrease in locations and time periods with relatively high ambient Os
29 concentrations and increase in locations and time periods with relatively low concentrations.
30 Therefore, an important consideration for epidemiology-based risk estimates is the extent to
31 which seasonal risk estimates in urban case study areas represent the U.S. as a whole, in terms of
32 the 63 response to decreasing precursor emissions. To address this, the second draft HREA
33 conducted national air quality analyses evaluating the response of ambient Os concentrations to
34 reductions in NOx emissions. Those analyses indicate that the 12 urban case study areas may not
35 represent the response of 63 in other populated areas of the U.S., including suburban areas,
36 smaller urban areas, and rural areas, and that the majority of the U.S. population lives in
37 locations where reducing NOx emissions would be expected to result in decreases in warm
3-112
-------
1 season averages of daily maximum 8-hour ambient Os concentrations. One implication of this is
2 that HREA estimates for the urban case study areas are likely to understate the average reduction
3 in Os-assocaiated mortality and morbidity risk that would be experienced across the population
4 upon reducing NOx emissions (U.S. EPA, 2014, Chapter 8).
5 Section 7.4 of the second draft HREA also highlights some additional uncertainties
6 associated with epidemiologic-based risk estimates (U.S. EPA, 2014). This section of the HREA
7 identifies and discusses sources of uncertainty and presents a qualitative evaluation of key
8 parameters that can introduce uncertainty into risk estimates (U.S. EPA, 2014, Table 7-4). For
9 several of these parameters the HREA also presents quantitative sensitivity analyses (U.S. EPA,
10 2014, sections 7.4.2 and 7.5.3). Of the uncertainties discussed in Chapter 7 of the HREA, those
11 related to the application of concentration-response functions from epidemiologic studies can
12 have particularly important implications for our consideration of epidemiology-based risk
13 estimates in this second draft PA.
14 As noted above, an important uncertainty is the shape of concentration-response
15 functions at low ambient Oj concentrations (U.S. EPA, 2014, Table 7-4). Consistent with the
16 ISA conclusion that there is no discernible population threshold in Os-associated health effects,
17 the second draft HREA estimates epidemiology-based mortality and morbidity risks for entire
18 distributions of ambient Oj concentrations, with the assumption that concentration-response
19 relationships remain linear over those distributions. In addition, in recognition of the ISA
20 conclusion that certainty in the shape of Os concentration-response functions decreases at low
21 ambient concentrations, the second draft HREA also estimates distributions of total mortality
22 incidence for various portions of the distribution of ambient Os concentrations. In this second
23 draft PA, we consider both types of risk estimates while recognizing that we have greater
24 certainty in the increased incidence and severity of Os-attributable effects at higher ambient Os
25 concentrations (which drive higher exposure concentrations, section 3.2.2 above), as compared
26 to lower concentrations.
27 The second draft HREA also notes important uncertainties associated with using a
28 concentration-response relationship developed for a particular population in a particular location
29 to estimate health risks in different populations and locations (U.S. EPA, 2014, Table 7-4). As
30 discussed above, concentration-response relationships derived from epidemiologic studies reflect
31 the spatial and temporal patterns of population exposures during the study. The second draft
32 HREA applies concentration-response relationships from epidemiologic studies to adjusted air
33 quality in study areas that are different from, and often larger in spatial extent than, the areas
34 used to generate the relationships. This approach ensures the inclusion of the actual non-
35 attainment monitors that often determine the magnitude of emissions reductions for the air
36 quality adjustments throughout the urban case study areas. This approach also allows the HREA
3-113
-------
1 to estimate patterns of health risks more broadly across a larger area, including a broader range
2 of air quality concentrations and a larger population. The second draft HREA notes that it is not
3 possible to quantify the impacts of this uncertainty on risk estimates in most urban case study
4 locations, though the HREA notes that mortality effect estimates for different portions of the
5 New York City CBSA-based assessment area vary by a factor of almost 10 (U.S. EPA, 2014,
6 section 7.5.3).
7 An additional, related uncertainty is that associated with applying concentration-response
8 functions from epidemiologic studies to adjusted air quality. Concentration-response functions
9 from the 63 epidemiologic studies used in the HREA are based on associations between day to
10 day variation in "area-wide" 63 concentrations (i.e., averaged across multiple monitors) and
11 variation in health effects. Epidemiologic studies use these area-wide 03 concentrations, which
12 reflect the particular spatial and temporal patterns of ambient 63 present in study locations, as
13 surrogates for the pattern of Os exposures experienced by study populations. To the extent
14 adjusting O^ concentrations to just meet the current standard results in important alterations in
15 the spatial and/or temporal patterns of ambient 63, there is uncertainty in the appropriateness of
16 applying concentration-response functions from epidemiologic studies to estimate health risks
17 associated with adjusted 63 air quality.76 Although the impact of this uncertainty on risk
18 estimates cannot be quantified (U.S. EPA, 2014, Table 7-4), it has the potential to become more
19 important as model adjustment results in larger changes in spatial and temporal patterns of
20 ambient 63 concentrations across urban case study areas.
21 There is also uncertainty related specifically to the public health importance of the
22 increases in relatively low O?, concentrations following air quality adjustment. This uncertainty
23 relates to the fact that risk estimates are equally influenced by decreasing high concentrations
24 and increasing low concentrations, when the increases and decreases are of equal magnitude.
25 Even on days with increases in relatively low area-wide average concentrations, resulting in
26 increases in estimated risks, some portions of the urban case study areas could experience
76As discussed above (section 3.2.1), decreasing modeled NOX emissions to just meet the current standard can
dramatically alter the spatial and temporal patterns of ambient O3 concentrations across urban case study areas.
Specifically, the relatively high O3 concentrations that often occur downwind of important NOX sources (e.g.,
outside urban centers) generally decrease, while the relatively low O3 concentrations near important NOX sources
(e.g., in urban centers) generally increase (U.S. EPA, 2014, section 4.3.3.2). In addition, decreases or increases in
ambient O3 could occur more broadly across areas in some instances, depending in part on meteorological
conditions. Such decreases in high O3 concentrations and increases in low concentrations can result in compression
of the spatial distributions of ambient O3 used to calculate area-wide average concentrations. Decreases and
increases can also result in compression of the temporal distributions of the area-wide O3 concentrations used to
estimate mortality and morbidity risks over a season, such that area-wide concentrations decrease on "high"-O3 days
and increase on"low"-O3 days. As indicated in the second draft HREA (U.S. EPA, 2014, Figures 4-10 and 4-11),
this compression of seasonal distributions of O3 concentrations is evident in all of the urban case study areas
evaluated, though the degree of compression varies considerably across areas. The most dramatic compression
occurs in Los Angeles (U.S. EPA, 2014, Figures 4-10 and 4-11).
3-114
-------
1 decreases in high O^ concentrations. To the extent Os-attributable effects are more strongly
2 supported for higher ambient concentrations, likely resulting in higher exposure concentrations
3 for some portions of study areas, the impacts on risk estimates of increasing low O^
4 concentrations reflect an important source of uncertainty.
5 Finally, we note the second draft HREA does not quantify any reductions in risk that
6 could be associated with reductions in the ambient concentrations of pollutants other than Os,
7 resulting from control of NOx. For example as discussed in chapter 2 of this second draft PA,
8 NOX emissions contribute to ambient NC>2, and NOX and VOCs can contribute to secondary
9 formation of PIVb.s constituents, including ammonium sulfate (NFLjSC^), ammonium nitrate
10 (NFLjNOs), and organic carbon (OC). Therefore, at some times and in some locations, control
1 1 strategies that would reduce NOx emissions (i.e., to meet an O?, standard) could reduce ambient
12 concentrations of NO2 and PlV^.s, resulting in health benefits beyond those directly associated
13 with reducing ambient Os concentrations.77
14 3.3 CAS AC ADVICE
15 Following the 2008 decision to revise the primary Os standard by setting the level at
16 0.075 ppm (75 ppb), CASAC strongly questioned whether the standard met the requirements of
17 the CAA, further described below. In September 2009, EPA announced its intention to
18 reconsider the 2008 standards, issuing a notice of proposed rulemaking in January 2010 (FR 75
19 2938). Soon after, EPA solicited CASAC review of that proposed rule and in January 201 1
20 solicited additional advice. This proposal was based on the scientific and technical record from
21 the 2008 rulemaking, including public comments and CASAC advice and recommendations. As
22 further described in section 1.2.2 above, EPA in the fall of 201 1 did not revise the standard as
23 part of the reconsideration process but decided to coordinate further proceedings on the
24 reconsideration rulemaking with this ongoing periodic review. Accordingly, in this section we
25 describe CASAC's advice related to the 2008 final decision and the subsequent reconsideration,
26 as well as its advice on the NAAQS review that was initiated in September 2008.
27 In April 2008, the members of the CASAC Ozone Review Panel sent a letter to EPA
28 stating "[I]n our most-recent letters to you on this subject — dated October 2006 and March
29 2007 — the CASAC unanimously recommended selection of an 8-hour average Ozone NAAQS
30 within the range of 0.060 to 0.070 parts per million [60 to 70 ppb] for the primary (human
3 1 health-based) Ozone NAAQS" (Henderson, 2008). The letter continued:
32 The CASAC now wishes to convey, by means of this letter, its additional,
3 3 unsolicited advice with regard to the primary and secondary Ozone NAA QS. In
77We expect little focus by states on controlling NOX for purposes of controlling PM2 5 given the more efficient
control of PM2 5 through reduction of SO2 and direct PM25 emissions in most locations.
3-115
-------
1 doing so, the participating members of the CASAC Ozone Review Panel are
2 unanimous in strongly urging you or your successor as EPA Administrator to
3 ensure that these recommendations be considered during the next review cycle for
4 the Ozone NAAQS that will begin next year ... numerous medical organizations
5 and public health groups have also expressed their support of these CASAC
6 recommendations' ... [The CASAC did] not endorse the new primary ozone
1 standard as being sufficiently protective of public health. The CASAC—as the
8 Agency's statutorily-established science advisory committee for advising you on
9 the national ambient air quality standards—unanimously recommended
10 decreasing the primary standard to within the range of 0.060 0.0 70 ppm [60 to
11 70 ppb]. It is the Committee's consensus scientific opinion that your decision to
12 set the primary ozone standard above this range fails to satisfy the explicit
13 stipulations of the Clean Air Act that you ensure an adequate margin of safety for
14 all individuals, including sensitive populations.
15 In response to EPA's solicitation of their advice on the Agency's proposed rulemaking as
16 part of the reconsideration, CASAC conveyed support (Samet, 2010).
17 CASAC fully supports EPA 's proposed range of 0.060 -0.070 parts per million
18 (ppm) for the 8-hour primary ozone standard. CASAC considers this range to be
19 justified by the scientific evidence as presented in the Air Quality Criteria for
20 Ozone and Related Photochemical Oxidants (March 2006) and Review of the
21 National Ambient Air Quality Standards for Ozone: Policy Assessment of
22 Scientific and Technical Information, OAQPS Staff Paper (July 2007). As stated
23 in our letters of October 24, 2006, March 26, 2007 and April 7, 2008 to former
24 Administrator Stephen L. Johnson, CASAC unanimously recommended selection
25 of an 8-hour average ozone NAAQS within the range proposed by EPA (0.060 to
26 0.070 ppm). In proposing this range, EPA has recognized the large body of data
27 and risk analyses demonstrating that retention of the current standard would
28 leave large numbers of individuals at risk for respiratory effects and/or other
29 significant health impacts including asthma exacerbations, emergency room
3 0 visits, hospital admissions and mortality.
31 In response to EPA's request for additional advice on the reconsideration in 2011,
32 CASAC reaffirmed their conclusion that "the evidence from controlled human and
33 epidemiological studies strongly supports the selection of a new primary ozone standard within
34 the 60 - 70 ppb range for an 8-hour averaging time" (Samet, 2011). As requested by EPA,
35 CASAC's advice and recommendations were based on the scientific and technical record from
36 the 2008 rulemaking. In considering the record for the 2008 rulemaking, CASAC stated the
37 following to summarize the basis for their conclusions (Samet, 2011, pp. ii to iii).
38 • The evidence available on dose-response for effects of ozone shows
39 associations extending to levels within the range of concentrations
40 currently experienced in the United States.
41 • There is scientific certainty that 6.6-hour exposures with exercise of
42 young, healthy, non-smoking adult volunteers to concentrations > 80 ppb
43 cause clinically relevant decrements of lung function.
3-116
-------
1 • Some healthy individuals have been shown to have clinically relevant
2 responses, even at 60 ppb.
3 • Since the majority of clinical studies involve young, healthy adult
4 populations, less is known about health effects in such potentially ozone
5 sensitive populations as the elderly, children and those with
6 cardiopulmonary disease. For these susceptible groups, decrements in
1 lung function may be greater than in healthy volunteers and are likely to
8 have a greater clinical significance.
9 • Children and adults with asthma are at increased risk of acute
10 exacerbations on or shortly after days when elevated ozone concentrations
11 occur, even when exposures do not exceed the NAAQS concentration of 75
12 ppb.
13 • Large segments of the population fall into what EPA terms a "sensitive
14 population group,'' i.e., those at increased risk because they are more
15 intrinsically susceptible (children, the elderly, and individuals with
16 chronic lung disease) and those who are more vulnerable due to increased
17 exposure because they work outside or live in areas that are more polluted
18 than the mean levels in their communities.
19 With respect to evidence from epidemiologic studies, CAS AC stated "[W]hile epidemiological
20 studies are inherently more uncertain as exposures and risk estimates decrease (due to the greater
21 potential for biases to dominate small effect estimates), specific evidence in the literature does
22 not suggest that our confidence on the specific attribution of the estimated effects of ozone on
23 health outcomes differs over the proposed range of 60-70 ppb." (Samet, 2011, p. 10).
24 In advice offered so far in the current review, which is considering an updated scientific
25 and technical record since the 2008 rulemaking, CASAC has not yet conveyed their view on the
26 adequacy of the current standard. In the first draft PA for the current review, staff reached the
27 preliminary conclusion that the currently available evidence supports revising the standard to
28 afford greater public health protection and that it does not support retention of the current
29 standard (USEPA, 2012xx). Staff also concluded that the available evidence provides support
30 for conducting further exposure and risk analyses of alternative standard levels in the range of 60
31 to 70 ppb (USEPA, 2012xx). CASAC commented the draft PA provided "a strong scientific
32 rationale for consideration of ozone levels (8 hour averages of 60 ppb to 70 ppb)" (Frey and
33 Samet, 2012).
34 3.4 PRELIMINARY STAFF CONCLUSIONS ON ADEQUACY OF PRIMARY
35 STANDARD
36 This section presents staffs preliminary conclusions for the Administrator to consider in
37 deciding whether it is appropriate to revise the existing primary Os standard. Our conclusions are
38 based on consideration of the assessment and integrative synthesis of the evidence presented in
3-117
-------
1 the ISA, including consideration of air quality distributions in locations of selected
2 epidemiologic studies; exposure and risk analyses in the second draft HREA; and the comments
3 and advice of CAS AC and public comment on earlier drafts of this document, and on the ISA
4 and HREA.
5 As an initial matter, staff concludes that reducing precursor emissions to achieve Os
6 concentrations that meet the current standard will provide important improvements in public
7 health protection. This initial conclusion is based on (1) the strong body of scientific evidence
8 indicating a wide range of adverse health outcomes attributable to exposures to Os
9 concentrations found in the ambient air and (2) estimates indicating decreased 63 exposures and
10 health risks upon meeting the current standard, compared to recent air quality.
11 Strong support for this initial conclusion is provided by controlled human exposure
12 studies of respiratory effects, and by quantitative estimates of exposures of concern and lung
13 function decrements based on the information in these studies. Analyses in the second draft
14 HREA estimate that the percentages of at-risk populations experiencing exposures of concern or
15 abnormal and potentially adverse lung function decrements are substantially lower for air quality
16 that just meets the current Os standard than for recent air quality.
17 Some support for this initial conclusion is also provided by estimates of Os-associated
18 mortality and morbidity based on application of concentration-response relationships from
19 epidemiologic studies to adjusted air quality. These estimates are more variable than estimates of
20 Oj, exposures and (^-induced lung function risks, and are associated with uncertainties that
21 complicate their interpretation. However, epidemiology-based risk estimates for short- and long-
22 term O?, concentrations, in combination with the HREA's national analysis of Os responsiveness
23 to reductions in NOx emissions and the larger body of health effects evidence, lead us to
24 conclude that Cb-associated mortality and morbidity would be expected to decrease following
25 reductions in 63 precursor emissions to meet the current 63 standard.
26 We next revisit the overarching policy question for this chapter, taking into consideration
27 the responses to specific questions focused on the adequacy of the current primary 03 standard
28 discussed above.
29 • Does the currently available scientific evidence and exposure/risk information, as
30 reflected in the ISA and HREA, support or call into question the adequacy of the
31 protection afforded by the current primary Os standard?
32 In considering the available evidence and information, staff concludes that the Os-
33 attributable health effects estimated to be allowed by air quality that meets the current primary
34 standard for 63 can reasonably be judged important from a public health perspective. Thus, we
35 conclude that the available health evidence and exposure/risk information call into question the
3-118
-------
1 adequacy of the public health protection provided by the current standard. We further conclude
2 that it is appropriate in this review to consider alternative standards that would increase public
3 health protection, compared to the current standard, and that it is not appropriate to consider
4 alternative standards with levels higher than the current standard, which would decrease public
5 health protection (see chapter 4). The basis for these conclusions is discussed below.
6 Studies evaluated since the completion of the 2006 03 AQCD support and expand upon
7 the strong body of evidence that, in the last review, indicated a causal relationship between short-
8 term Os exposures and respiratory health effects. Together, experimental and epidemiologic
9 studies support conclusions regarding a continuum of 63 respiratory effects ranging from small
10 reversible changes in pulmonary function to more serious effects that can result in respiratory-
11 related emergency department visits, hospital admissions, and/or mortality. Recent animal
12 toxicological studies support descriptions of modes of action for these respiratory effects and
13 augment support for biological plausibility for the role of Os in reported effects. With regard to
14 mode of action, evidence indicates that antioxidant capacity may modify the risk of respiratory
15 morbidity associated with 63 exposure. In addition, based on the consistency of findings across
16 studies and evidence for the coherence of results from different scientific disciplines, strong
17 evidence indicates that certain populations are at increased risk of Ch-related effects. These
18 include populations identified in previous reviews (i.e., people with asthma, children, older
19 adults, outdoor workers) and populations identified since the last review (i.e., people with certain
20 genotypes related to anti-oxidant and/or anti-inflammatory status; people with reduced intake of
21 certain nutrients, such as Vitamins C and E).
22 Evidence for adverse respiratory health effects attributable to "long-term" or repeated
23 daily 63 exposures is much stronger than in previous reviews, and the ISA concludes there is
24 likely to be a causal relationship between such OT, exposures and adverse respiratory health
25 effects. Uncertainties related to the extrapolation of data generated by rodent toxicology studies
26 to the understanding of health effects in humans have been reduced by studies in non-human
27 primates and by recent epidemiologic studies. The evidence available in this review includes new
28 epidemiologic studies using a variety of designs and analysis methods, conducted by different
29 research groups in different locations, evaluating the relationships between long-term Os
30 exposures and measures of respiratory morbidity and mortality. New evidence supports
31 associations between long-term or repeated 63 exposures and the development of asthma, with
32 several studies reporting interactions between genetic variants and such OT, exposures. Studies
33 also report associations between long-term or repeated 63 exposure and asthma prevalence,
34 asthma severity and control, respiratory symptoms among asthmatics, and respiratory mortality.
35 In considering the specific exposure concentrations reported to elicit respiratory effects,
36 we note that recent evidence includes controlled human exposure studies reporting lung function
3-119
-------
1 decrements and pulmonary inflammation in healthy adults engaged in intermittent, moderate
2 exertion following 6.6 hour exposures to Oj concentrations as low as 60 ppb, and lung function
3 decrements and respiratory symptoms following exposures to concentrations as low as 70 ppb.
4 Compared to the evidence available in the last review, these studies have strengthened support
5 for the occurrence of abnormal and potentially adverse respiratory effects following short-term
•yo
6 exposures to Oj, concentrations below 80 ppb. It is reasonable to judge exposures to such Oj,
1 concentrations to be potentially important from a public health perspective given the following:
8 1. The respiratory effects reported following exposures to 63 concentrations of 60 and 70
9 ppb, while at moderate exertion, can reasonably be judged adverse based on ATS criteria
10 and past advice from CASAC.
11
12 2. The controlled human exposure studies reporting these respiratory effects were conducted
13 in healthy adults, while at-risk groups (e.g., asthmatics) could experience larger and/or
14 more serious effects.
15
16 3. These respiratory effects are coherent with the serious health outcomes that have been
17 reported in epidemiologic studies (e.g., respiratory-related hospital admissions,
18 emergency department visits, and mortality).
19 Exposure estimates from the second draft HREA for urban case study areas indicate that,
20 in areas just meeting the current 63 standard, approximately 10 to 20% of children would
21 experience one or more exposures of concern to Os concentrations of 60 ppb or above. In the
22 case study areas evaluated in the HREA, this corresponds to over 2 million children experiencing
23 approximately 4 million such exposures, including over 200,000 asthmatic children. Nationally,
24 far more children would be expected to experience such exposures of concern in areas where the
25 current standard is just met. On average over the years evaluated in the HREA, approximately 3
26 to 8% of children are estimated to experience two or more exposures of concern to Os
27 concentrations of 60 ppb or greater. For the worst-case years (i.e., years with air quality patterns
28 resulting in the highest exposure estimates), approximately 10 to 25% of children could
29 experience one or more exposures of concern at or above 60 ppb, and up to 14% could
30 experience two or more.
31 Although the current standard more effectively limits exposures of concern at or above
32 higher O^ concentrations (i.e., 70, 80 ppb), up to about 8% of children are estimated to
33 experience exposures of concern at/above 70 ppb in the worst-case city and year (i.e., city and
34 year with the largest estimates). In the worst-case city and year, about 2% of children were
78 Cf. Misisssippi. 723 F. 3d at 262 ("Perhaps more studies like the Adams studies will yet reveal that the 0.060 ppm
level produces significant adverse decrements that simply cannot be attributed to normal variation in lung
function.")
3-120
-------
1 estimated to experience two or more exposures of concern to O?, concentrations of 70 ppb or
2 greater.
3 Though we focus on children in these analyses of O^ exposures, we also recognize that
4 exposures to 8-hour average 63 concentrations at or above 60, 70, or 80 ppb could be of concern
5 for some adult populations. As discussed above, the patterns of exposure estimates over years
6 and across cities are similar in adult asthmatics, older adults, and children, though smaller
7 percentages of adult populations are estimated to experience exposures of concern. Thus, the
8 results for children are one part of a broader range of potentially at-risk populations that also
9 includes asthmatic adults and older adults.
10 Consistent with estimates of exposures of concern, the second draft HREA also estimates
11 that under air quality conditions just meeting the current O^ NAAQS, hundreds of thousands of
12 asthmatic children would be expected to experience Cb-induced lung function decrements that
13 are large enough to be potentially adverse in people with lung disease. On average over the years
14 evaluated in the HREA, the current standard is estimated to allow about 14% to 19% of children
15 in urban case study areas, including asthmatic children, to experience one or more (^-induced
16 lung function decrements > 10%. This corresponds to about 300,000 asthmatic children.
17 Nationally, far more children would be expected to experience such Os-induced lung function
18 decrements. About 8% to 12% of children are estimated to experience two or more decrements >
19 10%, on average. In the worst-case years, approximately 17% to 22% of children in the urban
20 case study areas are estimated to experience one or more decrements > 10% and about 10% to
21 14% are estimated to experience two or more such decrements. As with exposures of concern,
22 the current standard more effectively limits larger Os-induced lung function decrements (i.e., >
23 15%, 20%). However, up to about 7% of children are estimated to experience one or more 63-
24 induced decrements > 15% in the worst-case city and year analyzed in the HREA (and as high as
25 about 4% for two or more decrements).
26 Recent epidemiologic studies also provide support, beyond that available in the last
27 review, for associations between short-term O?, exposures and a wide range of adverse
28 respiratory outcomes (including respiratory-related hospital admissions, emergency department
29 visits, and mortality) and with total mortality. Associations with morbidity and mortality are
30 stronger during the warm or summer months, and remain robust after adjustment for co-
31 pollutants. In one U.S. and several Canadian studies, associations with respiratory morbidity or
32 mortality were reported in locations that would have met the current Os standard. Even in some
33 study locations where the current standard was not met, considering reported concentration-
34 response functions or cut-point analyses in the context of available air quality data indicate the
35 existence of Os-health effect associations on the subsets of days with ambient O^ concentrations
36 below 75 ppb. Taken together, these studies and associated air quality data indicate a relatively
3-121
-------
1 high degree of confidence in the occurrence of (^-associated hospital admissions, emergency
2 department visits, and mortality at ambient concentrations that meet the current standard.
3 The HREA epidemiology-based risk estimates in 12 urban cases study areas indicate
4 thousands of 63-associated hospital admissions, emergency department visits, and deaths per
5 year for air quality conditions associated with meeting the current standard. Based on area-wide
6 O3 concentrations from the upper portions of seasonal distributions, a focus that we judge
7 appropriate given the greater certainty in O3-attributable effects at higher concentrations,
8 hundreds to thousands of Cb-associated deaths per year are estimated for air quality associated
9 with the current standard in urban case study areas, indicating the potential for substantial public
10 health risk. As recognized above in section 3.2.3.2, we note greater uncertainty in Os-attributable
11 effects at lower concentrations, which are subject to increases upon air quality adjustment.
12 Although there are additional uncertainties in quantifying risks by applying concentration-
13 response functions from epidemiology studies to adjusted Os air quality, the general magnitude
14 of risk estimates suggests the potential for a substantial number of (^-associated deaths and
15 adverse respiratory events nationally when the current standard is met.
16 In addition to the evidence and exposure/risk information discussed above, we also take
17 note of the CAS AC advice provided to the EPA Administrator on the proposed reconsideration
18 of the 2008 decision establishing the current standard and the advice of the CAS AC 63 Panel
19 thus far in the current review. In commenting on the proposed reconsideration, the prior CAS AC
20 63 Panel emphatically recommended revision of the standard to one with a lower level based
21 entirely on the evidence and information in the record for the 2008 standard, which has been
22 substantially strengthened in the current review (Samet, 2011; Samet, 2012). Based on review of
23 the first draft PA in the current review, the current CASAC 63 Panel also described the draft PA
24 as providing strong scientific rationale for consideration of lower standard levels (Frey and
25 Samet, 2012).
26 In consideration of all of the above, staff reaches the preliminary conclusion that the
27 available evidence and exposure and risk information clearly calls into question the adequacy of
28 public health protection provided by the current primary standard. This evidence and information
29 provides strong support for the occurrence of a range of adverse respiratory effects, and
30 mortality, under air quality conditions that would meet the current standard. Based on the
31 analyses in the second draft HREA, we conclude that the exposures and risks projected to remain
32 upon meeting the current standard are indicative of risks that can reasonably be judged to be
33 important from a public health perspective. Thus, staff concludes that the evidence and
34 information provides strong support for giving consideration to revising the current primary
35 standard in order to provide increased public health protection against an array of adverse health
36 effects that range from decreased lung function and respiratory symptoms to more serious
3-122
-------
1 indicators of morbidity (e.g., including emergency department visits and hospital admissions),
2 and mortality. We further conclude that it is not appropriate to consider alternative standards
3 with levels higher than the current standard, which would decrease public health protection. In
4 consideration of all of the above, staff draws the preliminary conclusion that it is appropriate for
5 the Administrator to consider revision of the current primary Os standard to provide increased
6 public health protection.
7
3-123
-------
1 3.5 REFERENCES
2 Adams, W. C. (2006a) Comparison of chamber 6.6 hour exposures to 0.04-0.08 ppm ozone via square-wave and
3 triangular profiles on pulmonary responses. Inhalation Toxicol. 18: 127-136.
4 http://dx.doi.org/10.1080/08958370500306107
5 Adams, WC. (2003a). Comparison of chamber and face mask 6.6-hour exposure to 0.08 ppm ozone via square-wave
6 and triangular profiles on pulmonary responses. Inhal Toxicol 15: 265 -281.
7 Adams, WC. (2003b). Relation of pulmonary responses induced by 66-h exposures to 0.08 ppm ozone and 2-h
8 exposures to 0.30 ppm ozone via chamber and face-mask inhalation. Inhal Toxicol 15: 745-759.
9 Adams, WC. (2002). Comparison of chamber and face-mask 6.6-hour exposures to ozone on pulmonary function
10 and symptoms responses. Inhal Toxicol 14: 745-764. http://dx.doi.org/10.1080/08958370290084610
11 Adams, WC. (1998). Dose-response effect of varied equivalent minute ventilation rates on pulmonary function
12 responses during exposure to ozone. Washington, DC: American Petroleum Institute.
13 Adamson, I; Prieditis, H (1995) Response of mouse lung to carbon deposition during injury and repair.
14 Environmental Health Perspectives. 103: 1: 72-76.
15 Akinbami, LJ; Lynch, CD; Parker, JD; Woodruff, TJ. (2010). The association between childhood asthma prevalence
16 and monitored air pollutants in metropolitan areas, United States, 2001-2004. EnvironRes 110: 294-301.
17 http://dx.doi.0rg/10.1016/j.envres.2010.01.001
18 Alexeeff, SE; Litonjua, AA; Suh, H; Sparrow, D; Vokonas, PS; Schwartz, J. (2007). Ozone exposure and lung
19 function: Effect modified by obesity and airways hyperresponsiveness in the VA Normative Aging Study.
20 Chest 132: 1890-1897. http://dx.doi.org/10.1378/chest.07-1126
21 Alexis, NE; Lay, JC; Hazucha, M; Harris, B; Hernandez, ML; Bromberg, PA; Kehrl, H; Diaz-Sanchez, D; Kim, C;
22 Devlin, RB; Peden, DB. (2010). Low-level ozone exposure induces airways inflammation and modifies cell
23 surface phenotypes in healthy humans. Inhal Toxicol 22: 593-600.
24 ATS (American Thoracic Society). (2000). What constitutes an adverse health effect of air pollution? Am. J. Respir.
25 Crit. Care Med. 161: 665-673.
26 ATS (American Thoracic Society). (1991). Lung function testing: selection of reference values and interpretive
27 strategies. Am. Rev. Respir. Dis. 144: 1202-1218.
28 ATS (American Thoracic Society). (1985) Guidelines as to what constitutes an adverse respiratory health effect,
29 with special reference to epidemiological studies of air pollution. Am. Rev. Respir. Dis. 131: 666-668.
30 Barraza-Villarreal, A; Sunyer, J; Hernandez-Cadena, L; Escamilla-Nunez, MC; Sienra-Monge, JJ; Ramirez-Aguilar,
31 M; Cortez-Lugo, M; Holguin, F; Diaz-Sanchez, D; Olin, AC; Romieu, I. (2008). Air pollution, airway
32 inflammation, and lung function in a cohort study of Mexico City schoolchildren. Environ Health Perspect
33 116: 832-838. http://dx.doi.org/10.1289/ehp.10926
34 Basha, MA; Gross, KB; Gwizdala, CJ; Haidar, AH; Popovich, J, Jr. (1994). Bronchoalveolar lavage neutrophilia in
3 5 asthmatic and healthy volunteers after controlled exposure to ozone and filtered purified air. Chest 106:
36 1757-1765.
37 Bell, ML; Dominici, F. (2008). Effect modification by community characteristics on the short-term effects of ozone
38 exposure and mortality in 98 U.S. communities. Am J Epidemiol 167: 986-997.
39 http://dx.doi.org/10.1093/aje/kwm396
3-124
-------
1 Bell, ML; Kim, JY; Dominici, F. (2007). Potential confounding of paniculate matter on the short-term association
2 between ozone and mortality in multisite time-series studies. Environ Health Perspect 115: 1591-1595.
3 http://dx.doi.org/10.1289/ehp.10108
4 Bell, ML; Peng, RD; Dominici, F. (2006). The exposure-response curve for ozone and risk of mortality and the
5 adequacy of current ozone regulations. Environ Health Perspect 114: 532-536.
6 Bell, M. L.; McDermott, A.; Zeger, S. L.; Samet, J. M.; Dominici, F. (2004) Ozone and short-term mortality in 95
7 U.S. urban communities, 1987-2000. JAMA J. Am. Med. Assoc. 292: 2372-2378.
8 Berhane, K; Zhang, Y; Linn, WS; Rappaport, EB; Bastain, TM; Salam, MT; Islam, T; Lurmann, F; Gilliland, FD.
9 (2011). The effect of ambient air pollution on exhaled nitric oxide in the Children's Health Study. Eur
10 Respir J 37: 1029-1036. http://dx.doi.org/10.1183/09031936.00081410
11 Biggeri, A; Baccini, M; Bellini, P; Terracini, B. (2005). Meta-analysis of the Italian studies of short-term effects of
12 airpollution (MISA), 1990-1999. Int J Occup Environ Health 11: 107-122.
13 Bloom B, Cohen RA, Freeman G. Summary health statistics for U.S. children: National Health Interview Survey,
14 2010. National Center for Health Statistics. Vital Health Stat 10(250). 2011.
15 http://www.cdc.gov/nchs/data/series/sr_10/srlO_250.pdf
16 Bosson, J; Stenfors, N; Bucht, A; Helleday, R; Pourazar, J; Holgate, ST; Kelly, FJ; Sandstrom, T; Wilson, S; Frew,
17 AJ; Blomberg, A. (2003). Ozone-induced bronchial epithelial cytokine expression differs between healthy
18 and asthmatic subjects. Clin Exp Allergy 33: 777-782.
19 Brauer, M; Blair, J; Vedal, S. (1996). Effect of ambient ozone exposure on lung function in farm workers. Am J
20 Respir Crit Care Med 154: 981-987.
21 Brauer, M; Brook, JR. (1997). Ozone personal exposures and health effects for selected groups residing in the Fraser
22 Valley. Atmos Environ 31: 2113-2121.
23 Breton, CV; Salam, MT; Vora, H; Gauderman, WJ; Gilliland, FD. (2011). Genetic variation in the glutathione
24 synthesis pathway, air pollution, and children's lung function growth. Am J Respir Crit Care Med 183: 243 -
25 248. http://dx.doi.org/10.1164/rccm.201006-0849OC
26 Brown, JS; Bateson, TF; McDonnell, WF. (2008). Effects of exposure to 0.06 ppm ozone on FEV1 in humans: A
27 secondary analysis of existing data. Environ Health Perspect 116: 1023-1026.
28 http://dx.doi.org/10.1289/ehp.11396
29 Brunekreef, B; Hoek, G; Breugelmans, O; Leentvaar, M. (1994). Respiratory effects of low-level photochemical air
30 pollution in amateur cyclists. Am J Respir Crit Care Med 150: 962-966.
31 Cakmak, S; Dales, RE; Angelica Rubio, M; Blanco Vidal, C. (2011). The risk of dying on days of higher air
3 2 pollution among the socially disadvantaged elderly. Environ Res 111:388-393.
33 http://dx.doi.0rg/10.1016/j.envres.2011.01.003
34 Cakmak, S; Dales, RE; Judek, S. (2006a). Do gender, education, and income modify the effect of air pollution gases
35 on cardiac disease? J Occup Environ Med 48: 89-94.
36 http://dx.doi.org/10.1097/01.jom.0000184878.11956.4b
37 Cakmak, S; Dales, RE; Judek, S. (2006b). Respiratory health effects of air pollution gases: Modification by
38 education and income. Arch Environ Occup Health 61: 5-10. http://dx.doi.org/10.3200/AEOH.61.L5-10
39 Carey, SA; Minard, KR; Trease, LL; Wagner, JG; Garcia, GJ; Ballinger, CA; Kimbell, JS; Plopper, CG; Corley,
40 RA; Postlethwait, EM; Harkema, JR; Einstein, DR. (2007). Three-dimensional mapping of ozone-induced
3-125
-------
1 injury in the nasal airways of monkeys using magnetic resonance imaging and morphometric techniques.
2 Toxicol Pathol 35: 27-40. http://dx.doi.org/10.1080/01926230601072343
3 Chan, CC; Wu, TH. (2005). Effects of ambient ozone exposure on mail carriers' peak expiratory flow rates. Environ
4 Health Perspect 113: 735-738. http://dx.doi.org/10.1289/ehp.7636
5 Chen, J; Tan, M; Nemmar, A; Song, W; Dong, M; Zhang, G; Li, Y (2006) Quantification of extrapulmonary
6 translocation of intratracheal-instilled particles in vivo in rats: effect of lipopolysaccharide. Toxicology
7 222:195-201.
8 Chhabra, SK; Yasir, A; Chaudhry, K; Shah, B. (2010). Effect of ozone on response to ovalbumin & its modulation
9 by vitamins C & E in sensitized guinea pigs. Indian J Med Res 132: 87-93.
10 Coffin, DL; Blommer, EJ; Gardner, DE; Holzman, R. (1967). Effect of air pollution on alteration of susceptibility to
11 pulmonary infection. Cincinnati, OH: U.S. Department of Health, Education, and Welfare.
12 Coffin, DL; Gardner, DE. (1972). Interaction of biological agents and chemical air pollutants. Ann Occup Hyg 15:
13 219-234.
14 Cohen, MD; Sisco, M; Baker, K; Li, Y; Lawrence, D; Van Loveren, H; Zelikoff, JT; Schlesinger, RB. (2002).
15 Effects of inhaled ozone on pulmonary immune cells critical to antibacterial responses in situ. Inhal
16 Toxicol 14: 599-619. http://dx.doi.org/10.1080/08958370290084520
17 Dales, RE; Cakmak, S; Doiron, MS. (2006). Gaseous air pollutants and hospitalization for respiratory disease in the
18 neonatal period. Environ Health Perspect 114: 1751-1754. http://dx.doi.org/10.1289/ehp.9044
19 Darrow, LA; Klein, M; Sarnat, JA; Mulholland, JA; Strickland, MJ; Sarnat, SE; Russell, AG; Tolbert, PE. (201 la).
20 The use of alternative pollutant metrics in time-series studies of ambient air pollution and respiratory
21 emergency department visits. JExpo Sci Environ Epidemiol 21: 10-19.
22 http://dx.doi.org/10.1038/jes.2009.49
23 DeNavas-Walt, Carmen, Bernadette D. Proctor, and Jessica C. Smith, U.S. Census Bureau, Current Population
24 Reports, P60-239, Income, Poverty, and Health Insurance Coverage in the United States: 2010, U.S.
25 Government Printing Office, Washington, DQ2011. http://www.census.gov/prod/2011pubs/p60-239.pdf
26 Depuydt, P; Joos, GF; Pauwels, RA. (1999). Ambient ozone concentrations induce airway hyperresponsiveness in
27 some rat strains. Eur Respir J 14: 125-131.
28 Devlin, RB; Mcdonnell, WF; Mann, R; Becker, S; House, DE; Schreinemachers, D; Koren, HS. (1991). Exposure of
29 humans to ambient levels of ozone for 6.6 hours causes cellular and biochemical changes in the lung. Am J
30 Respir Cell Mol Biol 4: 72-81.
31 Dimeo, MJ; Glenn, MG; Holtzman, MJ; Sheller, JR; Nadel, JA; Boushey, HA. (1981). Threshold concentration of
32 ozone causing an increase in bronchial reactivity in humans and adaptation with repeated exposures. Am
33 Rev Respir Dis 124: 245-248.
34 Driscoll, KE; Vollmuth, TA; Schlesinger, RB. (1987). Acute and subchronic ozone inhalation in the rabbit:
35 Response of alveolar macrophages. J Toxicol Environ Health 21: 27-43.
36 http://dx.doi.org/10.1080/15287398709531000
37 Dryden, DM; Spooner, CH; Stickland, MK; Vandermeer, B; Tjosvold, L; Bialy, L; Wong, K; Rowe, BH. (2010).
38 Exercise-induced bronchoconstriction and asthma. (AHRQ Publication No. 10-E001). Rockville, MD:
3 9 Agency for Healthcare Research and Quality.
3-126
-------
1 Eiswerth, ME; Shaw, WD; Yen, ST. (2005). Impacts of ozone on the activities of asthmatics: Revisiting the data. J
2 Environ Manage 77: 56-63. http://dx.doi.org/10.1016/j.jenvman.2005.02.010
3 Escamilla-Nunez, MC; Barraza-Villarreal, A; Hernandez-Cadena, L; Moreno-Macias, H; Ramirez-Aguilar, M;
4 Sienra-Monge, JJ; Cortez-Lugo, M; Texcalac, JL; del Rio-Navarro, B; Romieu, I. (2008). Traffic-related
5 air pollution and respiratory symptoms among asthmatic children, resident in Mexico City: The EVA
6 cohort study. RespirRes 9: 74. http://dx.doi.org/10.1186/1465-9921-9-74
7 Fanucchi, MV; Plopper, CG; Evans, MJ; Hyde, DM; Van Winkle, LS; Gershwin, LJ; Schelegle, ES. (2006). Cyclic
8 exposure to ozone alters distal airway development in infant rhesus monkeys. Am J Physiol Lung Cell Mol
9 Physiol 291: L644-L650. http://dx.doi.org/10.1152/ajplung.00027.2006
10 Folinsbee, LJ; Hazucha, MJ. (2000). Time course of response to ozone exposure in healthy adult females. Inhal
11 Toxicol 12: 151-167.
12 Folinsbee, LJ; Hazucha, MJ. (1989). Persistence of ozone-induced changes in lung function and airway
13 responsiveness. In Atmospheric ozone research and its policy implications. Amsterdam, The Netherlands:
14 Elsevier.
15 Folinsbee, LJ; Horstman, DH; Kehrl, HR; Harder, S; Abdul-Salaam, S; Ives, PJ. (1994). Respiratory responses to
16 repeated prolonged exposure to 0.12 ppm ozone. Am J Respir Crit Care Med 149: 98-105.
17 Franklin, M; Schwartz, J. (2008). The impact of secondary particles on the association between ambient ozone and
18 mortality. Environ Health Perspect 116: 453-458. http://dx.doi.org/10.1289/ehp.10777
19 Gauderman, WJ; Avol, E; Gilliland, F; Vora, H; Thomas, D; Berhane, K; McConnell, R; Kuenzli, N; Lurmann, F;
20 Rappaport, E; Margolis, H; Bates, D; Peters, J. (2004). The effect of air pollution on lung development
21 from 10 to 18 years of age. N Engl J Med 351: 1057-1067.
22 Gauderman, WJ; McConnell, R; Gilliland, F; London, S; Thomas, D; Avol, E; Vora, H; Berhane, K; Rappaport, EB;
23 Lurmann, F; Margolis, HG; Peters, J. (2000). Association between air pollution and lung function growth
24 in southern California children. Am J Respir Crit Care Med 162: 1383-1390.
25 Gielen, MH; Van Der Zee, SC; Van Wijnen, JH; Van Steen, CJ; Brunekreef, B. (1997). Acute effects of summer air
26 pollution on respiratory health of asthmatic children. Am J Respir Crit Care Med 155: 2105-2108.
27 Gong, H, Jr; McManus, MS; Linn, WS. (1997). Attenuated response to repeated daily ozone exposures in asthmatic
28 subjects. ArchEnvironOccup Health 52: 34-41.
29 Greer, JR; Abbey, DE; Burchette, RJ. (1993). Asthma related to occupational and ambient air pollutants in
30 nonsmokers. J Occup Environ Med 35: 909-915.
31 Harkema, JR; Plopper, CG; Hyde, DM; St George, JA; Dungworth, DL. (1987a). Effects of an ambient level of
32 ozone on primate nasal epithelial mucosubstances: quantitative histochemistry. Am J Pathol 127: 90-96.
33 Hatch, GE; Slade, R; Harris, LP; Mcdonnell, WF; Devlin, RB; Koren, HS; Costa, DL; Mckee, J. (1994). Ozone dose
34 and effect in humans and rats: A comparison using oxygen-18 labeling and bronchoalveolar lavage. Am J
35 Respir Crit Care Med 150: 676-683.
36 Henderson, R. (2008) Letter from CAS AC Chairman Rogene Henderson to EPA Administrator Stephen Johnson.
37 April 7, 2008, EPA-CASAC-08-001.
3 8 Henderson, R. (2007) Letter from CAS AC Chairman Rogene Henderson to EPA Administrator Stephen Johnson.
39 March 26, 2007, EPA-CASAC-07-002.
3-127
-------
1 Henderson, R. (2006) Letter from CAS AC Chairman Rogene Henderson to EPA Administrator Stephen Johnson.
2 October 24, 2006, EPA-CASAC-07-001.
3 Hernandez, ML; Lay, JC; Harris, B; Esther, CR; Brickey, WJ; Bromberg, PA; Diaz-Sanchez, D; Devlin, RB;
4 Kleeberger, SR; Alexis, NE; Peden, DB. (2010). Atopic asthmatic subjects but not atopic subjects without
5 asthma have enhanced inflammatory response to ozone. J Allergy Clin Immunol 126: 537-544.
6 http://dx.doi.org/10.1016/jjaci.2010.06.043
7 Hernandez-Cadena, L; Holguin, F; Barraza-Villarreal, A; Del Rio-Navarro, BE; Sienra-Monge, JJ; Romieu, I.
8 (2009). Increased levels of outdoor air pollutants are associated with reduced bronchodilation in children
9 with asthma. Chest 136: 1529-1536. http://dx.doi.org/10.1378/chest.08-1463
10 Halonen, JI; Lanki, T; Tiittanen, P; Niemi, JV; Loh, M; Pekkanen, J. (2009). Ozone and cause-specific
11 cardiorespiratory morbidity and mortality. J Epidemiol Community Health 64: 814-820.
12 http://dx.doi.org/10.1136/jech.2009.087106
13 Holz, O; Torres, RA; Timm, P; Mucke, M; Richter, K; Koschyk, S; Magnussen, H. (1999). Ozone-induced airway
14 inflammatory changes differ between individuals and are reproducible. Am J Respir Crit Care Med 159:
15 776-784.
16 Hoppe, P; Peters, A; Rabe, G; Praml, G; Lindner, J; Jakobi, G; Fruhmann, G; Nowak, D. (2003). Environmental
17 ozone effects in different population subgroups. Int J Hyg Environ Health 206: 505-516.
18 http://dx.doi.org/10.1078/1438-4639-00250
19 Hoppe, P; Praml, G; Rabe, G; Lindner, J; Fruhmann, G; Kessel, R. (1995). Environmental ozone field study on
20 pulmonary and subjective responses of assumed risk groups. EnvironRes 71: 109-121.
21 Horstman, DH; Ball, BA; Brown, J; Gerrity, T; Folinsbee, LJ. (1995). Comparison of pulmonary responses of
22 asthmatic and nonasthmatic subjects performing light exercise while exposed to a low level of ozone.
23 Toxicol Ind Health 11: 369-385.
24 Horstman, DH; Folinsbee, LJ; Ives, PJ; Abdul-Salaam, S; McDonnell, WF. (1990). Ozone concentration and
25 pulmonary response relationships for 6.6-hour exposures with five hours of moderate exercise to 0.08, 0.10,
26 and 0.12 ppm. Am J Respir Crit Care Med 142: 1158-1163.
27 Howden, Lindsay M. and Meyer, Julie A., U.S. Census Bureau, 2010 Census Briefs, C2010BR-03, Age and Sex
28 Composition: 2010, U.S. Department of Commerce, Economics and Statistics Administration, U.S. Census
29 Bureau, Washington, DC 20233. http://www.census.gov/prod/cen2010/briefs/c2010br-03.pdf
30 Hurst, DJ; Gardner, DE; Coffin, DL. (1970). Effect of ozone on acid hydrolases of the pulmonary alveolar
31 macrophage. J Reticuloendothel Soc 8: 288-300.
32 Hwang, BF; Lee, YL; Lin, YC; Jaakkola, JJK; Guo, YL. (2005). Traffic related air pollution as a determinant of
33 asthma among Taiwanese school children. Thorax 60: 467-473.
34 Hyde, DM; Miller, LA; Schelegle, ES; Fanucchi, MV; Van Winkle, LS; Tyler, NK; Avdalovic, MV; Evans, MJ;
35 Kajekar, R; Buckpitt, AR; Pinkerton, KE; Joad, JP; Gershwin, LJ; Wu, R; Plopper, CG. (2006). Asthma: a
36 comparison of animal models using stereological methods. Eur Respir Rev 15: 122-135.
37 http://dx.doi.org/10.1183/09059180.00010103
38 Islam, T; Berhane, K; McConnell, R; Gauderman, WJ; Avol, E; Peters, JM; Gilliland, FD. (2009). Glutathione-S-
39 transferase (GST) PI, GSTM1, exercise, ozone and asthma incidence in school children. Thorax 64: 197-
40 202. http://dx.doi.org/10.1136/thx.2008.099366
3-128
-------
1 Islam, T; McConnell, R; Gauderman, WJ; Avol, E; Peters, JM; Gilliland, FD. (2008). Ozone, oxidant defense genes
2 and risk of asthma during adolescence. Am J Respir Crit Care Med 177: 388-395.
3 http://dx.doi.org/10.1164/rccm.200706-863OC
4 Islam, T; Gauderman, WJ; Berhane, K; McConnell, R; Avol, E; Peters, JM; Gilliland, FD. (2007). The relationship
5 between air pollution, lung function and asthma in adolescents. Thorax 62: 957-963.
6 Ito, K; Thurston, GD; Silverman, RA. (2007b). Characterization of PM2.5, gaseous pollutants, and meteorological
7 interactions in the context of time-series health effects models. J Expo Sci Environ Epidemiol 17: S45-S60.
8 Jacquemin, B; Kauffmann, F; Pin, I; Le Moual, N; Bousquet, J; Gormand, F; Just, J; Nadif, R; Pison, C; Vervloet,
9 D; Ktinzli, N; Siroux, V. (In Press) Air pollution and asthma control in the epidemiological study on the
10 genetics and environment of asthma. J Epidemiol Community Health.
11 http://dx.doi.org/10.1136/jech.2010.130229
12 Jerrett, M; Burnett, RT; Pope, CA, III; Ito, K; Thurston, G; Krewski, D; Shi, Y; Calle, E; Thun, M. (2009). Long-
13 term ozone exposure and mortality. N Engl J Med 360: 1085-1095.
14 http://dx.doi.org/10.1056/NEJMoa0803894
15 Johnston, RA; Schwartzman, IN; Flynt, L; Shore, SA. (2005). Role of interleukin-6 in murine airway responses to
16 ozone. Am J Physiol Lung Cell Mol Physiol 288: L390-L397
17 Jorres, R; Nowak, D; Magnussen, H; Speckin, P; Koschyk, S. (1996). The effect of ozone exposure on allergen
18 responsiveness in subjects with asthma or rhinitis. Am J Respir Crit Care Med 153: 56-64.
19 Kan, H; London, SJ; Chen, G; Zhang, Y; Song, G; Zhao, N; Jiang, L; Chen, B. (2008). Season, sex, age, and
20 education as modifiers of the effects of outdoor air pollution on daily mortality in Shanghai, China: The
21 Public Health and Air Pollution in Asia (PAPA) Study. Environ Health Perspect 116: 1183-1188.
22 http://dx.doi.org/10.1289/ehp. 10851
23 Katsouyanni, K; Samet, JM; Anderson, HR; Atkinson, R; Le Tertre, A; Medina, S; Samoli, E; Touloumi, G;
24 Burnett, RT; Krewski, D; Ramsay, T; Dominici, F; Peng, RD; Schwartz, J; Zanobetti, A. (2009). Air
25 pollution and health: A European and North American approach (APHENA). (Research Report 142).
26 Boston, MA: Health Effects Institute. http://pubs.healtheffects.org/view.php?id=327
27 Kehrl, HR; Peden, DB; Ball, BA; Folinsbee, LJ; Horstman, DH. (1999). Increased specific airway reactivity of
28 persons with mild allergic asthma after 7.6 hours of exposure to 0.16 ppm ozone. J Allergy Clin Immunol
29 104: 1198-1204.
30 Khatri, SB; Holguin, FC; Ryan, PB; Mannino, D; Erzurum, SC; Teague, WG. (2009). Association of ambient ozone
31 exposure with airway inflammation and allergy in adults with asthma. J Asthma 46: 777-785.
32 http://dx.doi.org/10.1080/02770900902779284
33 Kim, CS; Alexis, NE; Rappold, AG; Kehrl, H; Hazucha, MJ; Lay, JC; Schmitt, MT; Case, M; Devlin, RB; Peden,
34 DB; Diaz-Sanchez, D. (2011). Lung function and inflammatory responses in healthy young adults exposed
35 to 0.06 ppm ozone for 6.6 hours. Am J Respir Crit Care Med 183: 1215-1221.
36 http://dx.doi.org/10.1164/rccm.201011-1813OC
37 Ko, FWS; Tarn, W; Wong, TW; Lai, CKW. (2007). Effects of air pollution on asthma hospitalization rates in
3 8 different age groups in Hong Kong. Clin Exp Allergy 37: 1312-1319. http://dx.doi.org/10.1111/j. 1365-
39 2222.2007.02791.x
40 Koren, HS, RB Devlin, DE Graham; R Mann; MP Mcgee; DH Horstman; WJ Kozumbo; S Becker; DE House; WF
41 McDonnell; PA Bromberg. (1989). Ozone-induced inflammation in the lower airways of human subjects.
42 Am. Rev. Respir. Dis. 139:407-415.
3-129
-------
1 Korrick, SA; Neas, LM; Dockeiy, DW; Gold, DR; Allen, GA; Hill, LB; Kimball, KD; Rosner, BA; Speizer, FE.
2 (1998). Effects of ozone and other pollutants on the pulmonary function of adult hikers. Environ Health
3 Perspect 106: 93-99. http://dx.doi.org/10.1289/ehp.9810693
4 Kulle, TJ; Sauder, LR; Kerr, HD; Farrell, BP; Bermel, MS; Smith, DM. (1982). Duration of pulmonary function
5 adaptation to ozone in humans. Am Ind Hyg Assoc J 43: 832-837.
6 Larsen, ST; Matsubara, S; Mcconville, G; Poulsen, SS; Gelfand, EW. (2010). Ozone increases airway
7 hyperreactivity and mucus hyperproduction in mice previously exposed to allergen. J Toxicol Environ
8 HealthA73: 738-747. http://dx.doi.org/10.1080/15287391003614034
9 Lee, YL; McConnell, R; Berhane, K; Gilliland, FD. (2009b). Ambient ozone modifies the effect of tumor necrosis
10 factor G-308A on bronchitic symptoms among children with asthma. Allergy 64: 1342-1348.
11 http://dx.doi.0rg/10.llll/j.1398-9995.2009.02014.x
12 Lebowitz, M. D.; Camilli, A. E.; Bronnimann, D.; Quackenboss, J. (1987). The significance and meaningfulness of
13 intraindividual changes in objective test results as responses to air contaminants. Presented at: 80th annual
14 meeting of the Air Pollution Control Association; June; New York, NY. Pittsburgh, PA: Air Pollution
15 Control 30 Association; paper no. 87-32.1.
16 Lewis, TC; Robins, TG; Dvonch, JT; Keeler, GJ; Yip, FY; Mentz, GB; Lin, X; Parker, EA; Israel, BA; Gonzalez, L;
17 Hill, Y. (2005). Air pollution-associated changes in lung function among asthmatic children in Detroit.
18 Environ Health Perspect 113: 1068-1075.
19 Lin, S; Bell, EM; Liu, W; Walker, RJ; Kim, NK; Hwang, SA. (2008a). Ambient ozone concentration and hospital
20 admissions due to childhood respiratory diseases in New York State, 1991-2001. Environ Res 108: 42-47.
21 http://dx.doi.0rg/10.1016/j.envres.2008.06.007
22 Lin, S; Liu, X; Le, LH; Hwang, SA. (2008b). Chronic exposure to ambient ozone and asthma hospital admissions
23 among children. Environ Health Perspect 116: 1725-1730. http://dx.doi.org/10.1289/ehp.11184
24 Lippmann, M. (1988) Health significance of pulmonary function responses to airborne irritants. JAPCA 38: 881-
25 887.
26 Liu, L; Poon, R; Chen, L; Frescura, AM; Montuschi, P; Ciabattoni, G; Wheeler, A; Dales, R. (2009a). Acute effects
27 of air pollution on pulmonary function, airway inflammation, and oxidative stress in asthmatic children.
28 Environ Health Perspect 117: 668-674. http://dx.doi.org/10.1289/ehpl 1
29 Mar, TF; Koenig, JQ. (2009). Relationship between visits to emergency departments for asthma and ozone exposure
30 in greater Seattle, Washington. Ann Allergy Asthma Immunol 103: 474-479.
31 McConnell, R; Berhane, K; Gilliland, F; London, SJ; Islam, T; Gauderman, WJ; Avol, E; Margolis, HG; Peters, JM.
32 (2002). Asthma in exercising children exposed to ozone: A cohort study. Lancet 359: 386-391.
33 http://dx.doi.org/10.1016/S0140-6736(02)07597-9
34 McDonnell, WF; Abbey, DE; Nishino, N; Lebowitz, MD. (1999a). Long-term ambient ozone concentration and the
35 incidence of asthma in nonsmoking adults: the Ahsmog study. Environ Res 80: 110-121.
36 Medina-Ramon, M; Schwartz, J. (2008). Who is more vulnerable to die from ozone air pollution? Epidemiology 19:
37 672-679.
38 Medina-Ramon, M; Zanobetti, A; Schwartz, J. (2006). The effect of ozone and PM10 on hospital admissions for
39 pneumonia and chronic obstructive pulmonary disease: A national multicity study. Am J Epidemiol 163:
40 579-588. http://dx.doi.org/10.1093/aje/kwj078
3-130
-------
1 Meng, YY; Rull, RP; Wilhelm, M; Lombard!, C; Balmes, J; Ritz, B. (2010). Outdoor air pollution and uncontrolled
2 asthma in the San Joaquin Valley, California. J Epidemiol Community Health 64: 142-147.
3 http://dx.doi.org/10.1136/jech.2008.083576
4 Middleton, N; Yiallouros, P; Kleanthous, S; Kolokotroni, O; Schwartz, J; Dockery, DW; Demokritou, P; Koutrakis,
5 P. (2008). A 10-year time-series analysis of respiratory and cardiovascular morbidity in Nicosia, Cyprus:
6 The effect of short-term changes in air pollution and dust storms. Environ Health 7:39.
7 Miller, FJ; Illing, JW; Gardner, DE. (1978). Effect of urban ozone levels on laboratory-induced respiratory
8 infections. Toxicol Lett 2: 163-169.
9 Moore, K; Neugebauer, R; Lurmann, F; Hall, J; Brajer, V; Alcorn, S; Tager, I. (2008). Ambient ozone
10 concentrations cause increased hospitalizations for asthma in children: An 18-year study in Southern
11 California. Environ Health Perspect 116: 1063-1070. http://dx.doi.org/10.1289/ehp.10497
12 Mortimer, KM; Neas, LM; Dockery, DW; Redline, S; Tager, IB. (2002). The effect of air pollution on inner-city
13 children with asthma. EurRespir J 19: 699-705. http://dx.doi.org/10.1183/09031936.02.00247102
14 Mortimer, KM; Tager, IB; Dockery, DW; Neas, LM; Redline, S. (2000). The effect of ozone on inner-city children
15 with asthma: Identification of susceptible subgroups. Am J Respir Crit Care Med 162: 1838-1845.
16 Mudway, IS; Stenfors, N; Blomberg, A; Helleday, R; Dunster, C; Marklund, SL; Frew, AJ; Sandstrom, T; Kelly, FJ.
17 (2001). Differences in basal airway antioxidant concentrations are not predictive of individual
18 responsiveness to ozone: A comparison of healthy and mild asthmatic subjects. Free Radic Biol Med 31:
19 962-974.
20 Neas, LM; Dockery, DW; Koutrakis, P; Speizer, FE. (1999). Fine particles and peak flow in children: Acidity versus
21 mass. Epidemiology 10: 550-553.
22 Neidell, M. (2009). Information, avoidance behavior, and health: The effect of ozone on asthma hospitalizations.
23 Journal of Human Resources 44: 450-478.
24 Neidell, M; Kinney, PL. (2010). Estimates of the association between ozone and asthma hospitalizations that
25 account for behavioral responses to air quality information. Environ Sci Pol 13: 97-103.
26 http://dx.doi.0rg/10.1016/j.envsci.2009.12.006
27 O'Connor, GT; Neas, L; Vaughn, B; Kattan, M; Mitchell, H; Crain, EF; III, ER; Gruchalla, R; Morgan, W; Stout, J;
28 Adams, GK; Lippmann, M. (2008). Acute respiratory health effects of air pollution on children with asthma
29 in U.S. inner cities. J Allergy Clinlmmunol 121: 1133-1139. http://dx.doi.org/10.1016/jjaci.2008.02.020
30 O*Net OnLine (2012). Work Context — Outdoors, Exposed to Weather.
31 http://www.onetonline.Org/find/descriptor/result/4.C.2.a. 1 .c?a= 1
32 Paulu, C; Smith, AE. (2008). Tracking associations between ambient ozone and asthma-related emergency
33 department visits using case-crossover analysis. J Public Health Manag Pract 14: 581-591.
34 Peden, DB; Boehlecke, B; Horstman, D; Devlin, R. (1997). Prolonged acute exposure to 0.16 ppm ozone induces
35 eosinophilic airway inflammation in asthmatic subjects with allergies. J Allergy Clin Immunol 100: 802-
36 808.
37 Peden, DB; Setzer, RW, Jr; Devlin, RB. (1995). Ozone exposure has both a priming effect on allergen-induced
3 8 responses and an intrinsic inflammatory action in the nasal airways of perennially allergic asthmatics. Am J
39 Respir Crit Care Med 151: 1336-1345.
3-131
-------
1 Plopper, CG; Smiley-Jewell, SM; Miller, LA; Fanucchi, MV; Evans, MJ; Buckpitt, AR; Avdalovic, M; Gershwin,
2 LJ; Joad, JP; Kajekar, R; Larson, S; Pinkerton, KE; Van Winkle, LS; Schelegle, ES; Pieczarka, EM; Wu,
3 R; Hyde, DM. (2007). Asthma/allergic airways disease: Does postnatal exposure to environmental
4 toxicants promote airway pathobiology? Toxicol Pathol 35: 97-110.
5 http://dx.doi.org/10.1080/01926230601132030
6 Pope, CA, III; Burnett, RT; Thun, MJ; Calle, EE; Krewski, D; Ito, K; Thurston, GD. (2002). Lung cancer,
7 cardiopulmonary mortality, and long-term exposure to fine paniculate air pollution. JAMA 287: 1132-
8 1141.
9 Qian, Z; Lin, HM; Chinchilli, VM; Lehman, EB; Duan, Y; Craig, TJ; Wilson, WE; Liao, D; Lazarus, SC; Bascom,
10 R. (2009). Interaction of ambient air pollution with asthma medication on exhaled nitric oxide among
11 asthmatics. Arch Environ Occup Health 64: 168-176. http://dx.doi.org/10.1080/19338240903240616
12 Rabinovitch, N; Zhang, LN; Murphy, JR; Vedal, S; Dutton, SJ; Gelfand, EW. (2004). Effects of wintertime ambient
13 air pollutants on asthma exacerbations in urban minority children with moderate to severe disease. J
14 Allergy Clinlmmunol 114: 1131-1137. http://dx.doi.org/10.1016/jjaci.2004.08.026
15 Rage, E; Siroux, V; Kunzli, N; Pin, I; Kauffmann, F. (2009b). Air pollution and asthma severity in adults. Occup
16 Environ Med 66: 182-188. http://dx.doi.org/10.1136/oem.2007.038349
17 Ren, C; Williams, GM; Mengersen, K; Morawska, L; Tong, S. (2008). Does temperature modify short-term effects
18 of ozone on total mortality in 60 large eastern U.S. communities? An assessment using the NMMAPS data.
19 Environ Int 34: 451-458.
20 Romieu, I; Barraza-Villarreal, A; Escamilla-Nunez, C; Texcalac-Sangrador, JL; Hernandez-Cadena, L; Diaz-
21 Sanchez, D; De Batlle, J; Del Rio-Navarro, BE. (2009). Dietary intake, lung function and airway
22 inflammation in Mexico City school children exposed to air pollutants. Respir Res 10: 122.
23 Romieu, I; Barraza-Villarreal, A; Escamilla-Nunez, C; Almstrand, AC; Diaz-Sanchez, D; Sly, PD; Olin, AC. (2008).
24 Exhaled breath malondialdehyde as a marker of effect of exposure to air pollution in children with asthma.
25 J Allergy Clin Immunol 121: 903-909. http://dx.doi.0rg/10.1016/j.jaci.2007.12.004
26 Ross, MA; Persky, VW; Scheff, PA; Chung, J; Curtis, L; Ramakrishnan, V; Wadden, RA; Hryhorczuk, DO. (2002).
27 Effect of ozone and aeroallergens on the respiratory health of asthmatics. Arch Environ Occup Health 57:
28 568-578. http://dx.doi.org/10.1080/00039890209602090
29 Salam, MT; Islam, T; Gauderman, WJ; Gilliland, FD. (2009). Roles of arginase variants, atopy, and ozone in
30 childhood asthma. J Allergy Clinlmmunol 123: 596-602. http://dx.doi.org/10.1016/jjaci.2008.12.020
31 Samet, J.M. Clean Air Scientific Advisory Committee (CASAC) Response to Charge Questions on the
32 Reconsideration of the 2008 Ozone National Ambient Air Quality Standards. EPA-CASAC-11-004.
33 March 30, 2011. Available online at:
34 http://yosemite.epa.gov/sab/sabproduct.nsf/0/F08BEB48C1139E2A8525785E006909AC/$File/EPA-
35 CASAC-ll-004-unsigned+.pdf
36 Samoli, E; Zanobetti, A; Schwartz, J; Atkinson, R; Le Tertre, A; Schindler, C; Perez, L; Cadum, E; Pekkanen, J;
37 Paldy, A; Touloumi, G; Katsouyanni, K. (2009). The temporal pattern of mortality responses to ambient
3 8 ozone in the APHEA project. J Epidemiol Community Health 63: 960-966.
39 http://dx.doi.org/10.1136/jech.2008.084012
40 Scannell, C; Chen, L; Aris, RM; Tager, I; Christian, D; Ferrando, R; Welch, B; Kelly, T; Balmes, JR. (1996).
41 Greater ozone-induced inflammatory responses in subjects with asthma. Am J Respir Crit Care Med 154:
42 24-29.
3-132
-------
1 Schelegle, ES; Morales, CA; Walby, WF; Marion, S; Allen, RP. (2009). 6.6-hour inhalation of ozone concentrations
2 from 60 to 87 parts per billion in healthy humans. Am J Respir Crit Care Med 180: 265-272.
3 http://dx.doi.org/10.1164/rccm.200809-1484OC
4 Schildcrout, JS; Sheppard, L; Lumley, T; Slaughter, JC; Koenig, JQ; Shapiro, GG. (2006). Ambient air pollution
5 and asthma exacerbations in children: An eight-city analysis. Am J Epidemiol 164: 505-517.
6 http://dx.doi.org/10.1093/aje/kwj225
7 Schiller JS, Lucas JW, Ward BW, Peregoy JA. Summary health statistics for U.S. adults: National Health Interview
8 Survey, 2010. National Center for Health Statistics. Vital Health Stat 10(252). 2012.
9 http://www.cdc.gov/nchs/data/series/sr_10/srlO_252.pdf
10 Schwartz, J. (2005a). How sensitive is the association between ozone and daily deaths to control for temperature?
11 Am J Respir Crit Care Med 171: 627-631.
12 Schwartz, J. (2005b). Who is sensitive to extremes of temperature? A case-only analysis. Epidemiology 16: 67-72.
13 http://dx.doi.org/10.1097/01.ede.0000147114.25957.71
14 Sienra-Monge, JJ; Ramirez-Aguilar, M; Moreno-Macias, H; Reyes-Ruiz, NI; Del Rio-Navarro, BE; Ruiz-Navarro,
15 MX; Hatch, G; Crissman, K; Slade, R; Devlin, RB; Romieu, I. (2004). Antioxidant supplementation and
16 nasal inflammatory responses among young asthmatics exposed to high levels of ozone. Clin Exp Immunol
17 138: 317-322. http://dx.doi.0rg/10.llll/j.1365-2249.2004.02606.x
18 Silverman, RA; Ito, K. (2010). Age-related association of fine particles and ozone with severe acute asthma in New
19 York City. J Allergy Clin Immunol 125: 367-373. http://dx.doi.org/10.1016/jjaci.2009.10.061
20 Smith, RL; Xu, B; Switzer, P. (2009b). Reassessing the relationship between ozone and short-term mortality in U.S.
21 urban communities. Inhal Toxicol 21: 37-61. http://dx.doi.org/10.1080/08958370903161612
22 Spektor, DM; Lippmann, M; Lioy, PJ; Thurston, GD; Citak, K; James, DJ; Bock, N; Speizer, FE; Hayes, C.
23 (1988a). Effects of ambient ozone on respiratory function in active, normal children. Am Rev Respir Dis
24 137:313-320.
25 Spektor, DM; Lippmann, M; Thurston, GD; Lioy, PJ; Stecko, J; O'Connor, G; Garshick, E; Speizer, FE; Hayes, C.
26 (1988b). Effects of ambient ozone on respiratory function in healthy adults exercising outdoors. Am Rev
27 Respir Dis 138: 821-828.
28 Stafoggia, M; Forastiere, F; Faustini, A; Biggeri, A; Bisanti, L; Cadum, E; Cernigliaro, A; Mallone, S; Pandolfi, P;
29 Serinelli, M; Tessari, R; Vigotti, MA; Perucci, CA. (2010). Susceptibility factors to ozone-related
30 mortality: A population-based case-crossover analysis. Am J Respir Crit Care Med 182: 376-384.
31 http://dx.doi.org/10.1164/rccm.200908-1269OC
32 Stieb, DM; Szyszkowicz, M; Rowe, BH; Leech, JA. (2009). Air pollution and emergency department visits for
33 cardiac and respiratory conditions: A multi-city time-series analysis. Environ Health Global Access Sci
34 Source 8: 25. http://dx.doi.org/10.1186/1476-069X-8-25
35 Strickland, MJ; Darrow, LA; Klein, M; Flanders, WD; Sarnat, JA; Waller, LA; Sarnat, SE; Mulholland, JA; Tolbert,
36 PE. (2010). Short-term associations between ambient air pollutants and pediatric asthma emergency
37 department visits. Am J Respir Crit Care Med 182: 307-316. http://dx.doi.org/10.1164/rccm.200908-
38 1201OC
39 Thurston, GD; Lippmann, M; Scott, MB; Fine, JM. (1997). Summertime haze air pollution and children with
40 asthma. Am J Respir Crit Care Med 155: 654-660.
3-133
-------
1 Tolbert, PE; Klein, M; Peel, JL; Sarnat, SE; Sarnat, JA. (2007). Multipollutant modeling issues in a study of ambient
2 air quality and emergency department visits in Atlanta. J Expo Sci Environ Epidemiol 17: S29-S35.
3 http://dx.doi.org/10.1038/sj.jes.7500625
4 U.S. EPA (2013). Integrated Science Assessment of Ozone and Related Photochemical Oxidants (Final Report).
5 U.S. Environmental Protection Agency, Washington, DC. EPA/600/R-10/076F. Available at:
6 http://www.epa.gov/ttn/naaqs/standards/ozone/s_o3_2008_isa.html
7 U.S. EPA (2014). Health Risk and Exposure Assessment for Ozone, Second External Review Draft. Office of Air
8 Quality Planning and Standards, Research Triangle Park, NC, 27711. EPA-452/P-14-004a. January 2014.
9 Available at: http://www.epa.gOv/ttn/naaqs/standards/ozone/s o3 2008 rea.html
10 U.S. EPA (2009). Integrated Science Assessment for Paniculate Matter (Final Report). U.S. Environmental
11 Protection Agency, Washington, DC, EPA/600/R-08/139F, 2009.
12 U.S. EPA (2006). Air Quality Criteria for Ozone and Related Photochemical Oxidants (2006 Final). U.S.
13 Environmental Protection Agency, Washington, DC. EPA/600/R-05/004aF-cF. March 2006. Available at:
14 http://www.epa. gov/ttn/naaqs/standards/ozone/s_o3_cr_cd.html
15 U.S. EPA (1996a). Air quality criteria for ozone and related photochemical oxidants [EPA Report]. (EPA/600/P-
16 93/004aF, cF). U.S. Environmental Protection Agency, Research Triangle Park, NC.
17 Vagaggini, B; Taccola, M; Clanchetti, S; Carnevali, S; Bartoli, ML; Bacci, E; Dente, FL; Di Franco, A; Giannini, D;
18 Paggiaro, PL. (2002). Ozone exposure increases eosinophilic airway response induced by previous allergen
19 challenge. Am J Respir Crit Care Med 166: 1073-1077.
20 Villeneuve, PJ; Chen, L; Rowe, BH; Coates, F. (2007). Outdoor air pollution and emergency department visits for
21 asthma among children and adults: A case-crossover study in northern Alberta, Canada. Environ Health
22 Global Access Sci Source 6: 40. http://dx.doi.org/10.1186/1476-069X-6-40
23 Wenten, M; Gauderman, WJ; Berhane, K; Lin, PC; Peters, J; Gilliland, FD. (2009). Functional variants in the
24 catalase and myeloperoxidase genes, ambient air pollution, and respiratory-related school absences: An
25 example of epistasis in gene-environment interactions. Am J Epidemiol 170: 1494-1501.
26 http://dx.doi.org/10.1093/aje/kwp310
27 Wolff, G.T. (1995) Letter to EPA Administrator Carol Browner: "CASAC Closure on the Primary Standard Portion
28 of the Staff Paper for Ozone" EPA-SAB-CASAC-LTR-96-002, November 30, 1995.
29 Wong, CM; Vichit-Vadakan, N; Vajanapoom, N; Ostro, B; Thach, TQ; Chau, PY; Chan, EK; Chung, RY; Ou, CQ;
30 Yang, L; Peiris, JS; Thomas, GN; Lam, TH; Wong, TW; Hedley, AJ; Kan, H; Chen, B; Zhao, N; London,
31 SJ; Song, G; Chen, G; Zhang, Y; Jiang, L; Qian, Z; He, Q; Lin, HM; Kong, L; Zhou, D; Liang, S; Zhu, Z;
32 Liao, D; Liu, W; Bentley, CM; Dan, J; Wang, B; Yang, N; Xu, S; Gong, J; Wei, H; Sun, H; Qin, Z. (2010).
3 3 Part 5. Public health and air pollution in Asia (PAPA): A combined analysis of four studies of air pollution
34 and mortality. In Public Health and Air Pollution in Asia (PAPA): Coordinated Studies of Short-Term
35 Exposure to Air Pollution and Daily Mortality in Four Cities (pp. 377-418). Boston, MA: Health Effects
36 Institute. http://pubs.healtheffects.org/view.php?id=348
37 Wong, CM; Yang, L; Thach, TQ; Chau, PY; Chan, KP; Thomas, GN; Lam, TH; Wong, TW; Hedley, AJ; Peiris, JS.
3 8 (2009). Modification by influenza on health effects of air pollution in Hong Kong. Environ Health Perspect
39 117: 248-253. http://dx.doi.org/10.1289/ehp.11605
40 Woodwell, D. A.; Cherry, D. K. (2004) National Ambulatory Medical Care Survey: 2002 summary. Hyattsville,
41 MD: National Center for Health Statistics; DHHS publication no. (PHS) 2004-1250. (Advance data from
42 vital and health statistics; no. 346). Available: http://www.cdc.gov/nchs/data/ad/ad346.pdf [3 August,
43 2005]
3-134
-------
1 Xia, Y; Tong, H. (2006). Cumulative effects of air pollution on public health. StatMed 25: 3548-3559.
2 http://dx.doi.org/10.1002/sim.2446
3 Yang, Q; Chen, Y; Krewski, D; Burnett, RT; Shi, Y; Mcgrail, KM. (2005). Effect of short-term exposure to low
4 levels of gaseous pollutants on chronic obstructive pulmonary disease hospitalizations. Environ Res 99: 99-
5 105. http://dx.doi.0rg/10.1016/j.envres.2004.09.014
6 Zanobetti, A; Schwartz, J. (2011). Ozone and survival in four cohorts with potentially predisposing diseases. Am J
7 Respir Crit Care Med 184: 836-841. http://dx.doi.org/10.1164/rccm.201102-0227OC
8 Zanobetti, A; Schwartz, J. (2008a). Is there adaptation in the ozone mortality relationship: A multi-city case-
9 crossover analysis. Environ Health 7: 22. http://dx.doi.org/10.1186/1476-069X-7-22
10 Zanobetti, A; Schwartz, J. (2008b). Mortality displacement in the association of ozone with mortality: An analysis
11 of 48 cities in the United States. Am J Respir Crit Care Med 177: 184-189.
12 http://dx.doi.org/10.1164/rccm.200706-823OC
13 Zanobetti, A; Schwartz, J. (2006). Air pollution and emergency admissions in Boston, MA. J Epidemiol Community
14 Health 60: 890-895. http://dx.doi.org/10.1136/iech.2005.039834
3-135
-------
1 4 CONSIDERATION OF ALTERNATIVE PRIMARY STANDARDS
2 Having reached the conclusion that the currently available scientific evidence and
3 exposure/risk information calls into question the adequacy of the current 63 standard, we next
4 consider the following overarching question:
5 • What is the range of potential alternative standards that are supported by the
6 currently available scientific evidence and exposure/risk information, as reflected
7 in the ISA and HREA respectively?
8 To address this overarching question, in the sections below we evaluate a series of more specific
9 questions related to the major elements of the NAAQS: indicator (section 4.1), averaging time
10 (section 4.2), form (section 4.3), and level (section 4.4). In addressing these questions, we
11 consider the currently available scientific evidence and exposure/risk information, including the
12 evidence and information available at the time of the last review and that newly available in the
13 current review, as assessed in the ISA and the second draft HREA. In so doing, we note that the
14 final decision by the Administrator in this review will consider these elements collectively in
15 evaluating the health protection afforded by the primary standard.l
16 4.1 INDICATOR
17 In the last review, EPA focused on 63 as the most appropriate indicator for a standard
18 meant to provide protection against ambient photochemical oxidants. In this review, while the
19 complex atmospheric chemistry in which 03 plays a key role has been highlighted, no
20 alternatives to 63 have been advanced as being a more appropriate indicator for ambient
21 photochemical oxidants. More specifically, the ISA noted that Os is the only photochemical
22 oxidant (other than NO2) that is routinely monitored and for which a comprehensive database
23 exists (ISA section 3.6). Data for other photochemical oxidants (e.g., PAN, H2O2, etc.) typically
24 have been obtained only as part of special field studies. Consequently, no data on nationwide
25 patterns of occurrence are available for these other oxidants; nor are extensive data available on
26 the relationships of concentrations and patterns of these oxidants to those of 63 (ISA section 3.6).
27 We further note that meeting an O^ standard can be expected to provide some degree of
28 protection against potential health effects that may be independently associated with other
29 photochemical oxidants, even though such effects are not discernible from currently available
30 studies indexed by 03 alone. That is, since the precursor emissions that lead to the formation of
31 63 generally also lead to the formation of other photochemical oxidants, measures leading to
:We also take note of the 1997 review (discussed in section 1.3.1.2.3), in which O3 background concentrations were
an additional consideration in selecting a standard. Background O3 is discussed in more detail in chapter 2 of this
second draft PA.
4-1
-------
1 reductions in population exposures to Os can generally be expected to lead to reductions in
2 population exposures to other photochemical oxidants. Taken together, we conclude that 63
3 remains the most appropriate indicator for a standard meant to provide protection against
4 photochemical oxidants.2
5 4.2 AVERAGING TIME
6 The EPA established the current 8-hour averaging time3 for the primary Oj, NAAQS in
7 1997 (62 FR 38856). The decision on averaging time in that review was based on numerous
8 controlled human exposure and epidemiologic studies reporting associations between 6 to 8 hour
9 Os concentrations and adverse respiratory effects (62 FR 38861). It was also noted that a
10 standard with a max 8-hour averaging time is likely to provide substantial protection against
11 respiratory effects associated with 1-hour peak OT, concentrations. Similar conclusions were
12 reached in the last Os NAAQS review and thus, the 8-hour averaging time was retained in 2008.
13 In the current review, we first consider the following question related to averaging time:
14 • To what extent does the available evidence continue to support the
15 appropriateness of a standard with an 8-hour averaging time?
16 In reaching conclusions related to this question, staff considers causality judgments from the
17 ISA, as well as results from the specific controlled human exposure and epidemiologic studies
18 that informed those judgments. These considerations are described below in more detail.
19 As an initial consideration with respect to the most appropriate averaging time for the 63
20 NAAQS, we note that the strongest evidence for Os-associated health effects is for respiratory
21 effects following short-term exposures. More specifically, the ISA concludes that evidence
22 relating short-term 63 exposures to respiratory effects is "sufficient to infer a causal
23 relationship." The ISA also judges that short-term exposures to OT, are "likely to cause" both
24 cardiovascular effects and mortality (U.S. EPA, 2013, section 2.5.2). Therefore, as in past
25 reviews, the strength of the available scientific evidence provides strong support for a standard
26 that protects the public health against short-term exposures to 03.
27 In first considering the level of support available for specific short-term averaging times,
28 we note the evidence available from controlled human exposure studies. As discussed in more
29 detail in chapter 3 of this second draft PA, substantial health effects evidence from controlled
30 human exposure studies demonstrates that a wide range of respiratory effects (e.g., pulmonary
31 function decrements, increases in respiratory symptoms, lung inflammation, lung permeability,
2The D.C. Circuit upheld the use of O3 as the indicator for photochemical oxidants based on these same
considerations. American Petroleum Inst. v. Costle, 665 F. 2d 1176, 1186 (D.C. Cir. 1981).
3This 8-hour averaging time reflects daily max 8-hour average O3 concentrations.
4-2
-------
1 decreased lung host defense, and airway hyperresponsiveness) occur in healthy adults following
2 6.6 hour exposures to 63 (EPA 2013, section 6.2.1.1). Compared to shorter exposure durations
3 (e.g., 1-hour), studies evaluating 6.6 hour exposures in healthy adults have reported respiratory
4 effects at lower Oj, exposure concentrations and at more moderate levels of exertion.
5 We also note the strength of evidence from epidemiologic studies that have evaluated a
6 wide variety of populations (e.g., including at-risk lifestages and populations, such as children
7 and people with asthma, respectively). A number of different averaging times are used in 63
8 epidemiologic studies, with the most common being the max 1-hour concentration within a 24-
9 hour period (1-hour max), the max 8-hour average concentration within a 24-hour period (8-hr
10 max), and the 24-hour average. These studies are discussed in chapter 3 of this second draft PA,
11 and are assessed in detail in chapter 6 of the ISA (U.S. EPA, 2013). Limited evidence from time-
12 series and panel epidemiologic studies comparing risk estimates across averaging times does not
13 indicate that one exposure metric is more consistently or strongly associated with respiratory
14 health effects or mortality, though the ISA notes some evidence for "smaller O^ risk estimates
15 when using a 24-hour average exposure metric" (EPA 2013, section 2.5.4.2; p. 2-31). For single-
16 and multi-day average Os concentrations, lung function decrements were associated with 1-hour
17 max, 8-hour max, and 24-hour average ambient 63 concentrations, with no strong difference in
18 the consistency or magnitude of association among the averaging times (EPA 2013, p. 6-71).
19 Similarly, in studies of short-term exposure to Os and mortality, Smith et al. (2009) and Darrow
20 et al. (2011) have reported high correlations between risk estimates calculated using 24-hour
21 average, 8-hour max, and 1-hour max averaging times (EPA 2013, p. 6-253). Thus, the
22 epidemiologic evidence alone does not provide a strong basis for distinguishing between the
23 appropriateness of 1-hour, 8-hour, and 24-hour averaging times.
24 Considering the health information discussed above, we conclude that an 8-hour
25 averaging time remains appropriate for addressing health effects associated with short-term
26 exposures to ambient 63. An 8-hour averaging time is similar to the exposure periods evaluated
27 in controlled human exposure studies, including recent studies that provide evidence for
28 respiratory effects following exposures to 63 concentrations below the level of the current
29 standard. In addition, epidemiologic studies provide evidence for health effect associations with
30 8-hour 03 concentrations, as well as with 1-hour and 24-hour concentrations. As in previous
31 reviews, we note that a standard with an 8-hour averaging time (combined with an appropriate
32 standard form and level) would also be expected to provide substantial protection against health
33 effects attributable to 1-hour and 24-hour exposures (e.g., 62 FR 38861, July 18, 1997).
34 The ISA also concludes that long-term 63 exposures are "likely to cause" respiratory
35 effects (US EPA, 2013, chapter 7). Thus, in this review we also consider the extent to which
36 currently available evidence and exposure/risk information suggests that a standard with an 8-
4-3
-------
1 hour averaging time can provide protection against respiratory effects associated with longer
2 term exposures to ambient O3. In doing so, staff considers the following question:
3 • To what extent does the available evidence and exposure/risk information indicate
4 that a standard with the current 8-hour averaging time could provide protection
5 against long-term exposures to ambient Os?
6 In considering this issue in the last review of the O3 NAAQS, staff noted that "because long-term
7 air quality patterns would be improved in areas coming into attainment with an 8-hr standard, the
8 potential risk of health effects associated with long-term exposures would be reduced in any area
9 meeting an 8-hr standard" (U.S. EPA, 2007, p. 6-57).
10 In the current review, we further evaluate this issue, with a focus on the "long-term" O3
11 metrics reported to be associated with mortality or morbidity in recent epidemiologic studies. As
12 discussed in section 3.1.3, much of the recent evidence for such associations is based on studies
13 that defined long-term O3 in terms of seasonal averages of daily max concentrations (e.g.,
14 seasonal averages of 1-hour or 8-hour daily max concentrations).
15 As an initial consideration, we note the risk results from the second draft HREA for
16 respiratory mortality associated with long-term O3 concentrations. As discussed in section
17 3.2.3.2, HREA analyses indicate that as air quality is adjusted to just meet the current 8-hour
18 standard, most urban case study areas are estimated to experience reductions in respiratory
19 mortality associated with long-term O3 concentrations based on the seasonal averages of 1-hour
20 daily max O3 concentrations evaluated in the study by Jerrett et al. (2009) (U.S. EPA, 2014,
21 chapter 7). As air quality is adjusted to meet lower potential alternative standard levels, for
22 standards based on 3-year averages of the annual fourth-highest daily max 8-hour O3
23 concentrations, respiratory mortality risks are estimated to be reduced further in urban case study
24 areas (section 4.4.2.3,below). This analysis indicates that an O3 standard with an 8-hour
25 averaging time, when coupled with an appropriate form and level, can reduce respiratory
26 mortality reported to be associated with "long-term" O3 concentrations.
27 In further considering the study by Jerrett et al. (2009), we compare long-term O3
28 concentrations following model adjustment in urban case study areas (i.e., adjusted to meet the
29 current and potential alternative 8-hour standards) to the concentrations present in study cities
30 that provided the basis for the positive and statistically significant association with respiratory
31 mortality. As indicated below (Table 4-3), this comparison suggests that a standard with an 8-
32 hour averaging time can decrease seasonal averages of 1-hour daily max O3 concentrations, and
33 can maintain those O3 concentrations below the seasonal average where we have the most
34 confidence in the reported concentration-response relationship with respiratory mortality (see
35 section 4.4.1 for further discussion).
4-4
-------
1 The second draft HREA also conducted analyses evaluating the impacts of reducing
2 regional NOx emissions on the seasonal averages of 8-hour daily max 63 concentrations.4
3 Seasonal averages of 8-hour daily max 03 concentrations reflect long-term metrics that have
4 been reported to be associated with respiratory morbidity effects in several recent 63
5 epidemiologic studies (e.g., Islam et al., 2008; Lin et al., 2008; Salam et al., 2009). The HREA
6 analyses indicate that the large majority of the U.S. population lives in locations where reducing
7 NOx emissions would be expected to result in decreases in seasonal averages of daily max 8-
8 hour ambient Os concentrations (U.S. EPA, 2014, chapter 8). Thus, consistent with the
9 respiratory mortality risk estimates noted above, this analysis suggests that reductions in 63
10 precursor emissions in order to meet a standard with an 8-hour averaging time would also be
11 expected to reduce the types of long-term 03 concentrations that have been reported in recent
12 epidemiologic studies to be associated with respiratory morbidity.
13 Taken together, we conclude that a standard with an 8-hour averaging time, coupled with
14 the current 4* high form and an appropriate level, would be expected to provide appropriate
15 protection against the long-term 63 concentrations that have been reported to be associated with
16 respiratory morbidity and mortality. This issue is considered further, within the context of
17 specific potential alternative standard levels, in section 4.4 below.
18 4.3 FORM
19 The "form" of a standard defines the air quality statistic that is to be compared to the
20 level of the standard in determining whether an area attains the standard. The foremost
21 consideration in selecting a form for potential alternative primary standards is the adequacy of
22 the public health protection provided by the combination of the form and the other elements of
23 the standard. As such, in reaching staff conclusions regarding the appropriate form(s) to consider
24 for a potential alternative primary OT, standard, we consider the following question:
25 • To what extent do the available evidence and/or information continue to support
26 the appropriateness of a standard with a form defined by the 3-year average of
27 annual 4th-highest 8-hour daily max Os concentrations?
28 The EPA established the current form of the primary O3 NAAQS in 1997 (62 FR 38856).
29 Prior to that time, the standard had a "1-expected-exceedance" form.5 An advantage of the
30 current concentration-based form recognized in the 1997 review is that such a form better
31 reflects the continuum of health effects associated with increasing ambient 63 concentrations.
4Analyses are based on regional NOX reductions, which are effective in bringing down peak ambient O3
concentrations, but can have variable impacts on seasonal mean concentrations.
5For a standard with a 1-expected-exceedance form to be met at an air quality monitoring site, the fourth-
highest air quality value in 3 years, given adjustments for missing data, must be less than or equal to the level of the
standard.
4-5
-------
1 Unlike an expected exceedance form, a concentration-based form gives proportionally more
2 weight to years when 8-hour 63 concentrations are well above the level of the standard than to
3 years when 8-hour O?, concentrations are just above the level of the standard. It was judged
4 appropriate to give more weight to higher 63 concentrations, given that available health evidence
5 indicated a continuum of effects associated with exposures to varying concentrations of Os, and
6 given that the extent to which public health is affected by exposure to ambient O^ is related to the
7 actual magnitude of the 63 concentration, not just whether the concentration is above a specified
8 level.
9 During the 1997 review, EPA considered a range of alternative "concentration-based"
10 forms, including the second-, third-, fourth- and fifth-highest daily max 8-hour concentrations in
11 an 03 season. The fourth-highest daily max was selected, recognizing that a less restrictive form
12 (e.g., fifth highest) would allow a relatively large percentage of sites to experience 63 peaks well
13 above the level of the standard, and would allow more days on which the level of the standard
14 may be exceeded when attaining the standard (62 FR 38856). Consideration was also given to
15 setting a standard with a form that would provide a margin of safety against possible but
16 uncertain chronic effects, and would provide greater stability to ongoing control programs.6 A
17 more restrictive form was not selected, recognizing that the differences in the degree of
18 protection afforded by the alternatives were not well enough understood to use any such
19 differences as a basis for choosing the most restrictive forms (62 FR 38856).
20 In the 2008 review, EPA additionally considered the potential value of a percentile-based
21 form. In doing so, EPA recognized that such a statistic is useful for comparing datasets of
22 varying length because it samples approximately the same place in the distribution of air quality
23 values, whether the dataset is several months or several years long. However, EPA concluded
24 that a percentile-based statistic would not be effective in ensuring the same degree of public
25 health protection across the country. Specifically, a percentile-based form would allow more
26 days with higher air quality values in locations with longer 63 seasons relative to places with
27 shorter 03 seasons.
28 Thus, in the 2008 review EPA concluded that a form based on the nth-highest max Os
29 concentration would more effectively ensure that people who live in areas with different length
30 03 seasons receive the same degree of public health protection. Based on analyses for forms
31 specified in terms of an nth-highest concentration (n ranged from 3 to 5), advice from CAS AC,
6 See American Trucking Assn's v. EPA, 283 F. 3d 355, 374-75 (D.C. Cir. 2002) (less stable implementation
programs may be less effective, and therefore EPA can consider programmatic stability in determining the form of a
NAAQS).
4-6
-------
1 and public comment,7 the Administrator concluded that a 4th-highest daily max should be
2 retained (73 FR 16465). In reaching this decision, the Administrator recognized that "the
3 adequacy of the public health protection provided by the combination of the level and form is a
4 foremost consideration" (73 FR 16475).
5 The Administrator also recognized that it is important to have a form that provides
6 stability with regard to implementation of the standard. In the case of Os, for example, he noted
7 the importance of a form insulated from the impacts of the meteorological events that are
8 conducive to Os formation. Such events could have the effect of reducing public health
9 protection, to the extent they result in frequent shifts between meeting and violating the standard
10 due to meteorological conditions. The Administrator noted that such frequent shifting could
11 disrupt an area's ongoing implementation plans and associated control programs (73 FR 16474).
12 In its notice of proposed rulemaking to reconsider the 2008 standard, the EPA did not propose to
13 reconsider the form of the standard.
14 In the current review, we consider the extent to which newly available information
15 provides support for consideration of alternative forms. In so doing, we take note of the
16 conclusions of prior reviews summarized above. We recognize the value of an nth-high statistic
17 over that of an expected exceedance or percentile-based form in the case of the 63 standard, for
18 the reasons summarized above. We additionally take note of the importance of stability in
19 implementation to achieving the level of protection specified by the NAAQS. Specifically, we
20 note that to the extent that areas engaged in implementing the 63 NAAQS frequently shift from
21 meeting to violating the standard, it is possible that ongoing implementation plans and associated
22 control programs could be disrupted, thereby reducing public health protection.
23 In light of this, while giving foremost consideration to the adequacy of public health
24 protection provided by the combination of all elements of the standard, including the form, we
25 consider particularly findings from prior reviews with regard to the use of the nth-high metric.
26 As noted above, the 4th-highest daily max was selected in 1997 in recognition of the public
27 health protection provided by this form, when coupled with an appropriate averaging time and
28 level, and recognizing that such a form can provide stability for implementation programs. The
29 currently available evidence and information does not call into question these conclusions from
30 previous reviews. Therefore, we conclude that it would be appropriate to retain the current 4* -
7 In the 2008 review, one group of commenters expressed the view that the standard was not adequate and supported
a more health-protective form (e.g., a second- or third-highest daily max form). Another group of commenters
expressed the view that the standard was adequate and did not provide any views on alternative forms that would be
appropriate should the Administrator consider revisions to the standard. The Administrator considered the protection
afforded by the combination of level and form in revising the standard in 2008 to 75 ppb, as a 3 -year average of the
annual fourth-highest daily max 8-hour concentrations (73 FR 16475).
4-7
-------
1 highest daily max form for an O^ standard with an 8-hour averaging time and a revised level, as
2 discussed below.
3 4.4 LEVEL
4 In considering potential alternative standards levels to provide greater protection than that
5 afforded by the current standard against (Vrelated adverse health effects, we address the
6 following overarching question.
7 • For an Os standard defined in terms of the current indicator, averaging time, and
8 form, what alternative levels are appropriate to consider in order to provide
9 adequate public health protection against short- and long- term exposures to Os
10 in ambient air?
11 In considering this question, we take into account the experimental and epidemiologic evidence
12 as presented in the ISA, as well as the uncertainties and limitations associated with this evidence
13 (section 4.4.1). In addition, we consider the quantitative estimates of exposure and risk provided
14 by the HREA, as well as the uncertainties and limitations associated with these risk estimates
15 (section 4.4.2).
16 4.4.1 Evidence-based Considerations
17 In this section, we consider the available evidence from controlled human exposure and
18 epidemiologic studies, including the uncertainties and limitations associated with that evidence,
19 within the context of potential alternative standard levels. We consider both the exposure
20 concentrations at which controlled human exposure studies provide evidence for health effects,
21 and the ambient 63 concentrations present in locations where epidemiologic studies have
22 reported health effect associations (see also section 3.1).
23 Controlled human exposure studies
24 We consider the following question related to controlled human exposure studies:
25 • To what extent does the available evidence from controlled human exposure
26 studies provide support for consideration of potential alternative standard levels
27 lower than 75 ppb?
28 To inform our conclusions regarding this question, we consider the lowest 63 concentrations at
29 which various effects have been evaluated and statistically significant effects reported. We also
30 consider the potential for reported effects to be adverse, including in at-risk populations.
31 As discussed in section 3.1.2.1, in healthy adults group mean Os-induced lung function
32 decrements exhibit a smooth dose-response relationship without evidence of a threshold from 40
33 to 120 ppb Os (US EPA, 2013, Figure 6-1). The lowest Os exposure concentration for which
34 statistically significant decrements have been reported is 60 ppb (Brown, 2006; Kim et al., 2011).
4-8
-------
1 The ISA concludes that mean FEVi is clearly decreased by 6.6-hour exposures to O^
2 concentrations of 60 ppb and higher in young, healthy adults during moderate exertion (US EPA,
3 2013, p. 6-9). As discussed in section 3.1.3, such a decrease in mean lung function meets the
4 ATS criteria for an adverse response given that a downward shift in the distribution of FEVi
5 would result in diminished reserve function, and therefore would increase risk from further
6 environmental insult. In addition, the ISA notes that following exposures to 60 ppb O^ 10% of
7 healthy individuals experience FEVi decrements > 10% (U.S. EPA, 2013, page 6-19). A 10%
8 decrement in FEVi is accepted by ATS as an abnormal response, and based on advice received
9 from CASAC in previous reviews, such decrements could be adverse in people with lung disease
10 (section 3.1.3).
11 As discussed in section 3.1.2.1, one recent controlled human exposure study has reported
12 Os-induced pulmonary inflammation (PMN influx to the ELF) following exposures of young,
13 healthy adults to Os concentrations of 60 ppb (Kim et al., 2011), the lowest concentration at
14 which inflammatory responses have been evaluated in human studies. Induction of pulmonary
15 inflammation is evidence that injury has occurred. The possibility of chronic effects due to
16 repeated inflammatory events has been evaluated in animal studies. Repeated events of acute
17 inflammation can have several potentially adverse outcomes including: induction of a chronic
18 inflammatory state; altered pulmonary structure and function, leading to diseases such as asthma;
19 altered lung host defense response to inhaled microorganisms, particularly in potentially at-risk
20 populations such as the very young and old; and, altered lung response to other agents such as
21 allergens or toxins (U.S. EPA, 2013, Section 6.2.3). Thus, lung injury and the resulting
22 inflammation, particularly if experienced repeatedly, provide a mechanism by which Os may
23 cause other more serious respiratory effects (e.g., asthma exacerbations) and possibly
24 extrapulmonary effects.
25 With respect to respiratory symptoms, a recent study by Schelegle et al. (2009) reported a
26 statistically significant increase in respiratory symptoms in young, healthy adults following 6.6
27 hour exposures to an average Os concentration of 70 ppb. This study also reported a statistically
28 significant decrease in FEVi following such exposures. As discussed in section 3.1.3, the
29 occurrence of both lung function decrements and respiratory symptoms meets criteria established
30 by the ATS defining an "adverse" respiratory response. Although some studies have reported
31 that respiratory symptoms develop during exposures at 60 ppb, the increases in symptoms in
32 these studies have not reached statistical significance by the end of the 6.6 hr exposures (Adams
33 2006; Schelegle et al., 2009).8
8Adams (2006) reported an increase in respiratory symptoms in healthy adults during a 6.6 hour exposure protocol
with an average O3 exposure concentration of 60 ppb. This increase was significantly different from initial
respiratory symptoms, but not from filtered air controls.
4-9
-------
1 Based on the results discussed above and in section 3.1.2.1, we conclude that controlled
2 human exposure studies provide evidence of potentially adverse lung function decrements and
3 airway inflammation in healthy individuals following exposures to 60 ppb Os, and evidence of
4 respiratory symptoms combined with lung function decrements (an "adverse" response based on
5 ATS criteria) following exposures to 70 ppb. In reaching these conclusions, we recognize that
6 most studies have not evaluated exposure concentrations below 60 ppb, and that 60 ppb does not
7 necessarily reflect an exposure concentration below which effects no longer occur. Specifically,
8 given the occurrence of airway inflammation following exposures to 60 ppb and higher, it may
9 be reasonable to expect that inflammation would also occur following exposures to 63
10 concentrations somewhat below 60 ppb. Although some studies show that respiratory symptoms
11 develop during exposures at 60 ppb, they have not reached statistical significance by the end of
12 the 6.6 hr exposures (Adams 2006; Schelegle et al. 2009). Thus, respiratory symptoms combined
13 with lung function decrements are likely to occur to some degree in healthy individuals with 6.6-
14 hr exposures to concentrations below 70 ppb. Further, we note that these controlled human
15 exposure studies were conducted in healthy adults and that people with asthma, including
16 asthmatic children, are likely to be more sensitive to Os-induced respiratory effects. Therefore,
17 these exposure concentrations are more likely to cause adverse respiratory effects in children and
18 adults with asthma, and more generally in people with respiratory disease.
19 In further considering effects following exposures to O^ concentrations below 75 ppb, in
20 section 3.1.4.1 we discuss panel studies highlighted in the ISA for the extent to which monitored
21 ambient Os concentrations reflect exposure concentrations in their study populations (U.S. EPA,
22 2013, section 6.2.1.2). These panel studies used on-site monitoring to evaluate Os-attributable
23 lung function decrements in individuals engaged in outdoor recreation, exercise, or work. Table
24 3-2 includes Os panel studies that report analyses of Cb-attributable lung function decrements for
25 Oj, concentrations at or below 75 ppb, and that measure 63 concentrations with monitors located
26 in the areas where study subjects were active (e.g., on site at summer camps or in locations
27 where exercise took place). Consistent with the results of controlled human exposure studies
28 discussed above, these panel studies report associations with lung function decrements for
29 subjects exposed to on-site monitored Os concentrations below 75 ppb. Associations in panel
30 studies have been reported for a wider range of populations than has been evaluated in controlled
31 human exposure studies, including children.
32 In considering controlled human exposure studies of other Os-induced effects, we note
33 that airway hyper-responsiveness and impaired lung host defense capabilities have been reported
34 in healthy adults engaged in moderate exertion following exposures to 63 concentrations as low
35 as 80 ppb, the lowest concentration evaluated for these effects. As discussed in section 3.1.2.1,
36 these physiological effects have been linked to aggravation of asthma and increased
4-10
-------
1 susceptibility to respiratory infection, potentially leading to increased medication use, increased
2 school and work absences, increased visits to doctors' offices and emergency departments, and
3 increased hospital admissions. These are all indicators of adverse Os-related morbidity effects,
4 which are consistent with, and provide plausibility for, the adverse morbidity effects and
5 mortality effects observed in epidemiologic studies.
6 In revisiting the question above, we conclude that the available controlled human
7 exposure evidence supports an upper end of the range of potential alternative standard levels for
8 consideration no higher than 70 ppb. For 6.6 hour exposures at 70 ppb, lung function decrements
9 and respiratory symptoms, a combination of effects that meet ATS criteria for an adverse
10 response (as discussed in section 3.1.3), have been demonstrated in healthy adults in controlled
11 human exposure studies.9 In addition, potentially adverse respiratory effects, including lung
12 function decrements and airway inflammation, have been demonstrated following 6.6 hour
13 exposures to Os concentrations below 70 ppb (i.e., at 60 ppb, as discussed below). A level of 70
14 ppb would also be below the lowest-observed-effects level for effects such as airway hyper-
15 responsiveness and impaired host-defense capabilities in healthy adults while at prolonged
16 moderate exertion. As discussed in section 3.1.2.1 of this second draft PA, such physiological
17 effects have been linked to aggravation of asthma and increased susceptibility to respiratory
18 infection, potentially leading to increased medication use, increased school and work absences,
19 increased visits to doctors' offices and emergency departments, and increased hospital
20 admissions.
21 Based on the above considerations, we also conclude that the evidence from controlled
22 human exposure studies supports setting the lower end to the range of alternative O?, standards at
23 60 ppb. Potentially adverse lung function decrements and pulmonary inflammation have been
24 demonstrated to occur at in healthy adults at 60 ppb. This is a short-term exposure concentration
25 that may be reasonably concluded to elicit adverse effects in at-risk groups. Pulmonary
26 inflammation, particularly if experienced repeatedly, provides a mechanism by which 63 may
27 cause other more serious respiratory morbidity effects (e.g., asthma exacerbations) and possibly
28 extrapulmonary effects.
29 Epidemiologic evidence
30 We also consider what the information from epidemiologic studies indicates with regard
31 to potential alternative standard levels appropriate for consideration. Based on the information in
32 section 3.1.4.2 of this second draft PA (see Table 3-3), we first note that several epidemiologic
'Although some studies report that respiratory symptoms develop during exposures to 60 ppb O3, these effects have
not reached statistical significance by the end of the 6.6 hour exposures (Adams, 2006; Schelegle et al., 2009). Thus
respiratory symptoms, in combination with lung function decrements, are likely to occur to some degree in healthy
individuals following exposures to O3 concentrations somewhat below 70 ppb.
4-11
-------
1 studies have reported positive and statistically significant associations with hospital admissions,
2 emergency department visits, and/or mortality in study areas where ambient 63 concentrations
3 would have met the current standard (i.e., with its level of 75 ppb). This includes Canadian
4 multicity studies in which the majority of study cities would have met the current standard over
5 entire study periods (Cakmak et al., 2006; Dales et al., 2006; Katsouyanni et al., 2009; Stieb et
6 al., 2009), and a U.S. single-city study conducted in a location likely to have met the current
7 standard over the entire study period (Mar and Koenig, 2009).
8 In further evaluating these studies, and building upon our conclusions based on controlled
9 human exposures studies, as discussed above, we consider the following question related to the
10 epidemiologic evidence:
11 • To what extent have U.S. and Canadian epidemiologic studies reported associations
12 with mortality or morbidity in locations likely to have met potential alternative Os
13 standards with levels from 70 to 60 ppb?
14 Addressing this question can provide important insights into the extent to which O^-
15 health effect associations are present for distributions of ambient 63 concentrations that would be
16 allowed by various potential alternative standards. To the extent Os health effect associations are
17 reported in study areas that would have met potential alternative standards, we have greater
18 confidence that exposures to ambient 63 concentrations allowed by such alternatives could result
19 in the types of clearly adverse effects evaluated in these studies.10 Therefore, our focus in this
20 section is to consider what these studies convey regarding the extent to which health effects may
21 be occurring (i.e., as indicated by associations) under air quality conditions allowed by potential
22 alternative standards. Specifically, we consider the numbers of study locations likely to have met
23 potential alternative standards with levels of 70, 65, and 60 ppb during study periods (Table 4-1).
24
See ATA III, 283 F.3d at 370 (EPA justified in revising NAAQS when health effect associations are observed at
levels allowed by the NAAQS).
4-12
-------
1 Table 4-1. Numbers of epidemiologic study locations likely to have met potential
2 alternative standards with levels of 70, 65, and 60 ppb
Study
Cakmak et al.
(2006)
Dales et al.
(2006)
Katsouyanni et
al. (2009)
Katsouyanni et
al. (2009)
Mar and Koenig
(2009)
Stieb et al.
(2009)
Result
Positive and statistically
significant association with
respiratory hospital
admissions
Positive and statistically
significant association with
respiratory hospital
admissions
Positive and statistically
significant associations with
respiratory hospital
admissions
Positive and statistically
significant associations with
total and cardiovascular
mortality
Positive and statistically
significant associations with
asthma emergency
department visits
Positive and statistically
significant association with
respiratory emergency
department visits
Cities
10 Canadian
cities
11 Canadian
cities
12 Canadian
cities
12 Canadian
cities
Single city:
Seattle
7 Canadian
cities
Number of study cities meeting potential
alternative standards during entire study
period
70 ppb
7
5
9
7
0
5
65 ppb
6
4
9
5
0
4
60 ppb
2
0
5
1
0
3
3 No U.S. or Canadian studies reported positive and statistically significant health effect
4 associations when all study locations would have met a standard with a level from 70 to 60 ppb
5 over the entire study period. However, for the studies by Cakmak et al. (2006), Katsouyanni et
6 al. (2009), and Stieb et al. (2009), the majority of study locations would likely have met a
7 standard with a level of either 70 or 65 ppb (Cakmak et al., 2006; Katsouyanni et al., 2009; Stieb
8 et al., 2009). In contrast, the majority of locations in these studies would likely have violated a
9 standard with a level of 60 ppb. While there is uncertainty in ascribing the multicity effect
10 estimates reported in these Canadian studies to ambient concentrations that would have met
11 standards with levels of 70 or 65 ppb (i.e., given that some study locations would have violated
12 such standards over at least part of the study period), reported multicity effect estimates are
13 largely influenced by locations meeting these potential alternative standards.
14 As with our consideration of the current standard (section 3.1.4.2), we next consider the
15 extent to which epidemiologic studies have characterized 63 health effect associations, including
16 confidence in those associations, for various portions of distributions of ambient Os
17 concentrations. In considering such analyses within the context of potential alternative standards,
4-13
-------
1 we focus on the extent to which epidemiologic studies report health effect associations for air
2 quality distributions restricted to ambient pollutant concentrations below one or more
3 predetermined cut-points. As discussed in section 3.1.4.2, such "cut-point" analyses can provide
4 information on the magnitude and statistical precision of effect estimates for defined
5 distributions of ambient concentrations, which may in some cases include distributions that
6 would be allowed by potential alternative standards. Specifically, we consider the following
7 question:
8 • To what extent do cut-point analyses from epidemiologic studies report health effect
9 associations at ambient Os concentrations that are likely to be allowed by potential
10 alternative standards with levels from 70 to 60 ppb?
11 As with our consideration of the current standard in section 3.1.4.2 of this second draft
12 PA, we evaluate the cut-point analyses presented in the U.S. multicity study by Bell et al. (2006).
13 These cut-point analyses can provide insights into the magnitude and statistical precision of
14 health effect associations for different portions of the distribution of ambient concentrations,
15 including insights into the ambient concentrations below which uncertainty in reported
16 associations becomes notably greater. Our analysis of air quality data associated with the cut-
17 points evaluated by Bell et al., and uncertainties associated with that analysis, is described
18 elsewhere in this document (section 3.1.4.2). In this section, we consider what these cut-point
19 analyses indicate with regard to the potential for health effect associations to extend to ambient
20 Os concentrations likely to be allowed by a revised Os NAAQS with a level below 75 ppb.
21 We particularly focus on the lowest cut-point for which the association between 63 and
22 mortality was reported to be statistically significant (i.e., 30 ppb, as discussed in section 3.1.4.2).
23 Based on the 63 air quality concentrations that met the criteria for inclusion in the 30 ppb cut-
24 point analysis, 84% of study areas had 3-year averages of annual 4* highest 8-hour daily max Os
25 concentrations at or below 70 ppb over the entire study period (Table 4-2). In addition, 64% of
26 study areas had 3-year averages of annual 4th highest 8-hour daily max 63 concentrations at or
27 below 65 ppb (Table 4-2). In contrast, the majority of study areas had 4* highest concentrations
28 above 60 ppb. While there are uncertainties in interpreting these cut-point analyses within the
29 context of potential alternative standard levels, they suggest that the majority of the air quality
30 distributions that provided the basis for a positive and statistically significant association with
31 mortality would have been allowed by a standard with a level of 70 or 65 ppb, but would have
32 violated a standard with a level of 60 ppb. For higher cut-points, all of which also resulted in
33 statistically significant associations with mortality, the majority of study cities had 3-year
34 averages of annual 4th highest 8-hour daily max concentrations greater than 70 ppb.
35
4-14
-------
1 Table 4-2.
2
Number of study cities with 3-year averages of 4 highest 8-hour daily max
concentrations greater than 70, 65, or 60 ppb, for various cut-point analyses
presented in Bell et al. (2006)
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
Number (%) of
Cities with 4th
highest >70 (any
3-yr period; 1987-
2000)
Number (%) of
Cities with 4th
highest >65 (any
3-yr period; 1987-
2000)
Number (%) of
Cities with 4th
highest >60 (any
3-yr period; 1987-
2000)
Cut-point for 2-day moving average across monitors and cities (24-h avg)
25
0 (0%)
3 (3%)
16
(16%)
30
16
(16%)
35
(36%)
61
(62%)
35
55
(56%)
77
(79%)
86
(88%)
40
82
(84%)
89
(91%)
94
(96%)
45
89
(91%)
94
(96%)
95
(97%)
50
92
(94%)
95
(97%)
96
(8%)
55
94
(96%)
95
(97%)
96
(8%)
60
95
(97%)
95
(97%)
96
(8%)
All
95
(97%)
95
(97%)
96
(8%)
In considering the implications of these analyses for potential alternative standard levels,
we also note the important uncertainties described in section 3.1.4. Several of these uncertainties
become increasingly important as health effect associations are evaluated for lower ambient Os
concentrations, such as when considering associations reported at the lower ends of the
distributions of ambient 63. These include uncertainties that could obscure presence of potential
thresholds, affecting our characterization of confidence in O^ health effect associations over
distributions of ambient concentrations; uncertainty in the extent to which the relatively low
ambient Os concentrations present in some study areas cause or contribute to reported effects;
and uncertainty in the extent to which we were able to identify the air quality data associated
with health effects in some published analyses (particularly for the subset analyses by Bell et al.,
2006) (section 3.1.4.2).
We next consider the extent to which epidemiologic studies employing longer-term
ambient 63 concentration metrics can inform our consideration of potential alternative standard
levels. In doing so, we consider the following question:
• To what extent does the available evidence indicate that a standard with a level
from 70 to 60 ppb, combined with the current 8-hour averaging time and 4th high
form, could provide protection from long-term exposures to ambient Os
concentrations for which there is evidence of health effects?
We first note that, as discussed in section 3.1.4.3 of this second draft PA, virtually all of
the study cities that provided the basis for the positive and statistically significant association
4-15
-------
1 between long-term Os and respiratory mortality (Jerrett et al., 2009) would have violated the
2 current standard, and therefore potential alternative standards with lower levels. Thus, as with
3 our consideration of the current standard in section 3.1.4.3, while the study by Jerrett et al.
4 (2009) contributes to our understanding of health effects associated with ambient 63
5 (summarized in section 3.1.2), it is less informative regarding the extent to which those health
6 effects may be occurring under air quality conditions that would meet potential alternative
7 standards.
8 To further evaluate this issue, we use the adjusted air quality in urban case study areas, as
9 described in the second draft HREA, to consider the extent to which just meeting alternative 63
10 standards with levels of 70, 65, and 60 ppb could maintain long-term 63 concentrations below
11 those in the cities that provided the basis for the positive and statistically significant association
12 with respiratory mortality reported by Jerrett et al. (2009).u Upon adjustment of air quality in
13 U.S. urban case study areas to meet the current and potential alternative 8-hour standards,
14 seasonal average 1-hour daily max concentrations were calculated and compared to the
15 concentrations in study cities.
16 As discussed in section 3.1.4.3, Jerrett et al. (2009) reported that when seasonal averages
17 of 1-hour daily max 63 concentrations12 ranged from 33 to 104 ppb, there was no statistical
18 deviation from a linear concentration-response relationship between 63 and respiratory mortality
19 across 96 U.S. cities (U.S. EPA, 2013, section 7.7). However, as discussed in section 3.1.4.3, we
20 have the greatest confidence in the reported linear concentration-response function for "long-
21 term" Os concentrations above the first quartile (i.e., 53.2 ppb), given the notable widening in
22 confidence intervals for lower concentrations (based on visual inspection of Figure 3-6 in section
23 3.1.4.3); the limited evidence noted by study authors for a threshold at 56 ppb;13 and the fact that
24 most study cities contributing to the linear function had OT, concentrations in the highest three
25 quartiles (accounting for approximately 72% of the respiratory deaths in the cohort, based on
26 Table 2 in the published study).
27 Given the above, we note the extent to which long-term 03 concentrations (i.e., seasonal
28 average of 1-hour daily max) in urban case study areas are estimated to be at or below 53 ppb
1: Air quality in U. S. urban case study areas was adjusted to just meet the current 8-hour standard at 75 ppb, as well
as potential potential alternative 8-hour standards at 70 ppb, 65 ppb, and 60 ppb, as described in the second draft
HREA (chapter 4). After a given adjustment, seasonal average 1-hour daily max concentrations were calculated.
12
Jerrett et al. (2009) evaluated the April to September averages of 1 -hour daily max O3 concentrations across 96
U.S. metropolitan areas from 1977- 2000. In urban areas with multiple monitors, April to September 1-hour daily
max concentrations from each individual monitor were averaged. This step was repeated for each year in the study
period. Finally, each yearly averaged O3 concentrations was then averaged again to yield the single averaged 1-hour
daily max O3 concentration depicted on the x-axis of Figure 3-6 below.
13The ISA does not reach conclusions regarding the potential for a threshold in the association between "long-term"
O3 concentrations and respiratory mortality.
4-16
-------
1 following model adjustment to meet potential alternative standards with levels of 70, 65, and 60
2 ppb. To the extent air quality adjustment to just meet potential alternative short-term standards
3 results in long-term concentrations near or below 53 ppb, we have greater confidence in the
4 degree to which those short-term standards could protect against the health effects associated
5 with longer term Os exposures. Though there is uncertainty associated with these comparisons
6 (e.g., due to uncertainty in the potential for a threshold to exist; uncertainty in the long-term
7 concentration below which confidence intervals widen notably, based on visual inspection of
8 concentration-response function in the published study; and the limited number of urban case
9 study areas for which adjusted air quality is available), this analysis can provide insight into the
10 extent to which various alternative short-term standards would be expected to maintain long-term
11 Os concentrations below those where we have the most confidence in the reported concentration-
12 response relationship with respiratory mortality.
13 Table 4-3 indicates that when considering recent (i.e., unadjusted) air quality, 2 of 12
14 urban case study areas had seasonal average 1-hour daily max O^ concentrations at or below 53
15 ppb in all of the years examined. When air quality was adjusted to just meet the current 8-hour
16 standard (75 ppb in Table 4-3), 6 of 12 urban case study areas had seasonal average 1-hour daily
17 max Oj, concentrations at or below 53 ppb in all of the years examined. When air quality is
18 further adjusted to just meet potential alternative standards with lower levels, seasonal averages
19 of 1-hour daily max O^ concentrations are estimated to be at or below 53 ppb in 9 of 12 urban
20 case study areas (70 ppb level), 10 of 12 urban case study areas (65 ppb level), and 11 of 11
21 urban case study areas (60 ppb level).14 Though as noted above there are important uncertainties
22 associated with interpreting these comparisons, they suggest that in many locations across the
23 U.S. a standard with an 8-hour averaging time, when combined with the current 4th high form
24 and an appropriate standard level, would be expected to maintain seasonal averages of 1-hour
25 daily max Os concentrations below those where analyses indicate the most confidence in the
26 concentration-response relationship with respiratory mortality reported by Jerrett et al. (2009).
14As described in the second draft HREA, a standard level of 60 ppb was not evaluated in New York City (U.S.
EPA, 2014, chapter 4).
4-17
-------
1 Table 4-3. Seasonal averages of 1-hour daily max Os concentrations in U.S. urban case
2 study areas for recent air quality and air quality adjusted to just meet the
3 current and potential alternative standards.
Atlanta
Baltimore
Boston
Cleveland
Denver
Detroit
Houston
Los Angeles
New York City
Philadelphia
Sacramento
Saint Louis
Air Quality
Adjusted to:
Recent
75
70
65
60
Recent
75
70
65
60
Recent
75
70
65
60
Recent
75
70
65
60
Recent
75
70
65
60
Recent
75
70
65
60
Recent
75
70
65
60
Recent
75
70
65
60
Recent
75
70
65
60
Recent
75
70
65
60
Recent
75
70
65
60
Recent
75
70
65
60
2006
(Adj Yrs 2006-2008)
65
53
50
47
45
60
54
52
49
46
49
48
46
44
43
51
49
47
45
41
63
62
60
58
53
50
50
48
47
45
53
48
47
46
45
65
58
55
52
50
53
47
44
36
NA
56
51
49
47
45
66
55
52
50
47
58
53
50
47
44
2007
(Adj Yrs 2006-2008)
63
52
49
46
44
59
54
51
49
46
50
49
47
45
43
52
50
48
45
41
63
61
59
58
53
54
52
50
49
46
48
46
45
44
43
61
59
56
53
51
54
47
45
36
NA
59
52
50
48
46
59
50
48
46
44
58
53
51
48
45
2008
(Adj Yrs 2008-2010)
57
53
49
46
44
57
53
51
48
46
46
49
48
46
44
53
51
48
45
41
63
63
62
59
53
51
NA
51
49
46
47
47
46
45
43
64
60
57
54
52
55
51
48
39
NA
57
54
51
49
47
65
54
51
49
46
52
51
50
48
45
2009
(Adj Yrs 2008-2010)
50
47
44
42
40
52
49
48
46
44
45
45
44
43
41
49
47
45
43
40
58
58
58
56
51
48
NA
49
47
45
47
48
47
46
44
62
60
58
54
52
48
47
45
38
NA
51
49
47
45
43
61
51
49
47
44
51
50
48
46
43
2010
(Adj Yrs 2008-2010)
56
52
49
46
44
60
55
53
50
48
49
48
48
46
44
54
51
48
45
42
60
60
58
55
50
52
NA
52
50
47
46
46
46
45
44
57
58
56
53
50
55
51
48
39
NA
58
54
52
49
47
55
48
46
44
42
55
54
52
49
46
4-18
-------
1 Based on the above analyses, we conclude that the available epidemiologic evidence is
2 consistent with the available evidence from controlled human exposure studies in providing
3 support for consideration of a standard level in the range of 70 to 60 ppb. Compared to the
4 current standard, a standard level from within this range would expected to be more effective at
5 maintaining short-term and long-term ambient Os concentrations below those where the evidence
6 indicates (^-associated mortality and/or morbidity.
7 In reaching overall staff conclusions about an appropriate range of standard levels for
8 consideration, we further evaluate the results of the exposure and risk assessments that are based
9 on modeling changes in the entire distribution of ambient 63 concentrations to simulate just
10 meeting potential alternative standards. These results are discussed below in section 4.4.2.
11 4.4.2 Air Quality-, Exposure-, and Risk-Based Considerations
12 Beyond considering the available evidence, we also consider the extent to which specific
13 potential alternative standard levels, in conjunction with the current averaging time and form (3-
14 year average of annual 4* highest 8-hour daily max), could reduce estimated Os exposures and
15 health risks. In the first draft PA (U.S. EPA, 2012b), we concluded that the available evidence
16 supports conducting further exposure and risk analyses of potential alternative standard levels in
17 the range of 70 down to 60 ppb. Based on these conclusions, the second draft HREA evaluates
18 exposures and risks estimated to be associated with potential alternative standard levels from the
19 upper (70 ppb), middle (65 ppb), and lower (60 ppb) portions of this range. In considering these
20 analyses in this second draft PA, we consider the following question:
21 • To what extent does the available exposure and risk information provide support
22 for considering potential alternative standard levels from 70 to 60 ppb, when
23 combined with the current 8-hour averaging time and 4th high form?
24 In considering exposure and risk analyses, we emphasize the nature and magnitude of the Os
25 exposures and health risks estimated to remain upon just meeting each alternative standard level,
26 and the changes in exposures and risks estimated for each alternative level when compared to the
27 current standard. Section 4.4.2.1 below discusses our exposure-based considerations. Sections
28 4.4.2.2 and 4.4.2.3 discuss our consideration of estimates of lung function risks and estimates of
29 epidemiology-based mortality/morbidity risks, respectively.
30 4.4.2.1 Exposure-Based Considerations
31 As discussed in more detail in section 3.2.2 of this second draft PA, the exposure
32 assessment presented in the second draft HREA (U.S. EPA, 2014) provides estimates of the
33 number and percent of people exposed to Os concentrations at or above benchmark
34 concentrations of 60, 70, and 80 ppb, while at moderate or greater exertion. Estimates of such
4-19
-------
1 "exposures of concern" provide perspective on the potential public health impacts of Os-related
2 effects, including for effects that cannot currently be evaluated in a quantitative risk assessment.
3 The approach taken in the second draft HREA to estimating exposures of concern, and the key
4 uncertainties associated with exposure estimates, are summarized in section 3.2.2 for air quality
5 adjusted to just meet the current standard and are discussed in more detail in chapter 5 of the
6 second draft HREA (U.S. EPA, 2014). As discussed in section 3.2.2, when evaluating potential
7 alternative standard levels we focus on modeled exposures for school-age children (ages 5-18),
8 noting that percentages of asthmatic school-age children estimated to experience exposures of
9 concern are virtually indistinguishable from those for all children, and that patterns of exposure
10 in children represent a broader range of at-risk populations, which includes adult asthmatics and
11 older adults.
12 In this section, we consider the following question:
13 • To what extent are potential alternative standards with revised levels estimated to
14 reduce the occurrence of Os exposures of concern, compared to the current
15 standard, and what are the nature and magnitude of the exposures remaining
16 for each alternative standard level evaluated?
17 Key results related to this question are summarized below (Figures 4-1 to 4-4). Figures 4-1 and
18 4-2 present estimates of one or more exposures of concern, and Figures 4-3 and 4-4 present
19 estimates of two or more exposures of concern.
4-20
-------
1 Figure 4-1. Percent of children estimated to experience one or more exposures of concern at or above 60, 70, or 80 ppb for air
2 quality adjusted to just meet the current and potential alternative standards (averaged over 2006 to 2010)
60 ppb benchmark
70 ppb benchmark
80 ppb benchmark
75 ppb 70 ppb 65 ppb 60ppb 75 ppb 70 ppb 65 ppb 60ppb 75 ppb 70 ppb 65 ppb 60ppb
-Atlanta
-Baltimore
-Boston
-Chicago
-Cleveland
-Dallas
-Denver
-Detroit
Houston
-Los Angeles
-New York
Philadelphia
Sacramento
St. Louis
Washington
4-21
-------
1
2
3
Figure 4-2. Percent of children estimated to experience one or more exposures of concern at or above 60, 70, or 80 ppb for air
adjusted to just meet the current and potential alternative standards (worst-case year from 2006 to
20101*)
60 ppb benchmark
70 ppb benchmark
80 ppb benchmark
75 ppb 70 ppb 65 ppb 60ppb 75 ppb 70 ppb
6 5 ppb
7 5 ppb
70 ppb 65 ppb 60ppb
Standard Level
> Atlanta
—•—Baltimore
—*—Boston
X Chicago
'!•: Cleveland
• Dallas
—i—Denver
Detroit
Houston
+• Los Angeles
—•—New York
—*— Philadelphia
Sacramento
—t—St. Louis
Washington
"Worst-case" year refers to the year in each urban case study area with the largest percentage of children estimated to experience exposures of concern.
4-22
-------
1 Figure 4-3. Percent of children estimated to experience two or more exposures of concern at or above 60, 70, or 80 ppb for air
2 quality adjusted to just meet the current and potential alternative standards (averaged over 2006 to 2010)
60 ppb benchmark
70 ppb benchmark
80 ppb benchmark
75ppl) 70 ppb 65 ppb 60 ppb 75 ppb 70 ppb 65 ppb 60 ppb 75 ppb 70 ppb 65 ppb 60 ppb
-Atlanta
-Baltimore
-Boston
-Chicago
-Cleveland
-Dallas
-Denver
-Detroit
Houston
-Los Angeles
-NewYork
Philadelphia
Sacramento
St. Louis
Washington
4-23
-------
1
2
Figure 4-4. Percent of children estimated to experience two or more exposures of concern at or above 60, 70, or 80 ppb for air
quality adjusted to just meet the current and potential alternative standards (worst-case year from 2006 to 2010)
60 ppb benchmark
70 ppb benchmark
80 ppb benchmark
75 ppb 70ppb 65ppb 60ppb 75ppb 70ppb 65ppb 60ppb 75ppb
Standard Level
70 ppb 65 ppb 60 ppb
4
—»—Atlanta
—•—Baltimore
—*—Boston
Chicago
i|i Cleveland
-•-Dallas
—>—Denver
Detroit
Houston
> Los Angeles
-•-New York
Philadelphia
Sacramento
—St. Louis
Washington
4-24
-------
1 As illustrated above in Figures 4-1 to 4-4, adjusting air quality to just meet progressively
2 lower potential alternative standard levels reduces estimated exposures of concern consistently
3 across urban case study areas. These results reflect the consistent reductions in the highest
4 ambient 63 concentrations upon model adjustment, as summarized in section 3.2.1 and as
5 discussed in more detail in the second draft HREA (U.S. EPA, 2014, chapter 4). Based on
6 Figures 4-1 to 4-4 and the associated details described in the second draft UREA (U.S. EPA
7 2014, chapter 5), we take note of the following with regard to exposures of concern for specific
8 potential alternative standard levels:
9 1. For a standard level of 70 ppb:
10 a. On average over the years 2006 to 2010, a standard with a level of 70 ppb is
11 estimated to allow approximately 3 to 11% of children in urban case study areas to
12 experience one or more exposures of concern at or above 60 ppb (approximately 30 to
13 70% reduction, relative to current standard). Summing across urban case study areas,
14 these percentages correspond to over 1 million children experiencing over 1.5 million
15 exposures of concern at or above 60 ppb during a single 63 season. Of these children,
16 over 100,000 are asthmatics.
17
18 b. On average over the years 2006 to 2010, a standard with a level of 70 ppb is
19 estimated to allow approximately 0.5 to 3.5% of children in urban case study areas to
20 experience two or more exposures of concern at or above 60 ppb (approximately 50
21 to 85% reduction, relative to current standard).
22
23 c. In the worst-case years (i.e., those with the largest exposure estimates), a standard
24 with a level of 70 ppb is estimated to allow approximately 5 to 19% of children in
25 urban case study areas to experience one or more exposures of concern at or above 60
26 ppb, and approximately 2 to 9% to experience two or more.
27
28 d. On average over the years 2006 to 2010, a standard with a level of 70 ppb is
29 estimated to allow approximately 1% or less of children to experience one or more
30 exposures of concern at or above 70 ppb (approximately 55 to 90% reduction, relative
31 to current standard), and far less than 1% to experience two or more such exposures
32 (approximately 65 to 100% reduction, relative to current standard).
33
34 e. In the worst-case years, approximately 3% or less of children are estimated to
35 experience one or more exposures of concern at or above 70 ppb, and less than 1%
36 are estimated to experience two or more such exposures.
37
38 f. A standard with a level of 70 ppb is estimated to allow less than 1% of children to
39 experience one or more exposures of concern at or above 80 ppb, even in the worst-
40 case years. No children are estimated to experience two or more such exposures.
41
42 2. For a standard level of 65 ppb:
4-25
-------
1 a. On average over the years 2006 to 2010, a standard with a level of 65 ppb is
2 estimated to allow approximately 4% or less of children in urban case study areas to
3 experience one or more exposures of concern at or above 60 ppb (approximately 70 to
4 100% reduction, relative to current standard). Summing across urban case study
5 areas, these percentages correspond to almost 500,000 children experiencing
6 approximately 500,000 exposures of concern at or above 60 ppb during a single Os
7 season. Of these children, almost 50,000 are asthmatics.
8
9 b. On average over the years 2006 to 2010, a standard with a level of 65 ppb is
10 estimated to allow less than 1% of children to experience two or more exposures of
11 concern at or above 60 ppb (approximately 85 to 100% reduction, relative to current
12 standard).
13
14 c. In the worst-case years, a standard with a level of 65 ppb is estimated to allow
15 approximately 9% or less of children to experience one or more exposures of concern
16 at or above 60 ppb, and approximately 3% or less to experience two or more such
17 exposures.
18
19 d. On average over the years 2006 to 2010, a standard with a level of 65 ppb is
20 estimated to allow less than 1% of children to experience one or more exposures of
21 concern at or above 70 ppb (approximately 90 to 100% reduction, relative to current
22 standard), and no children to experience two or more such exposures (100%
23 reduction, relative to current standard). Even in the worst-case years, a level of 65
24 ppb is estimated to allow less than 1% of children to experience exposures of concern
25 at or above 70 ppb.
26
27 e. A standard with a level of 65 ppb is estimated to allow virtually no children to
28 experience exposures of concern at or above 80 ppb, even in the worst-case years.
29
30 3. For a standard level of 60 ppb:
31 a. On average over the years 2006 to 2010, a standard with a level of 60 ppb is
32 estimated to allow approximately 1% or less of children to experience one or more
33 exposures of concern at or above 60 ppb (approximately 90 to 100% reduction,
34 relative to current standard), and virtually no children to experience multiple such
35 exposures.
36
37 b. In the worst-case years, a standard with a level of 60 ppb is estimated to allow
38 approximately 2% or less of children to experience one or more exposures of concern
39 at or above 60 ppb, and virtually no children to experience multiple such exposures.
40
41 c. On average over the years 2006 to 2010, a standard with a level of 60 ppb is
42 estimated to eliminate exposures of concern at or above 70 ppb (100% reduction,
43 relative to current standard) or 80 ppb. Even in years with the highest exposure
44 estimates, virtually no children are estimated to experience such exposures.
4-26
-------
1 In further considering these exposure estimates, we take note of the associated
2 uncertainties, as discussed in more detail in section 3.2.2 of this second draft PA. These include
3 (1) individual variability in responsiveness to O^ exposures; (2) potential to underestimate
4 exposures in most highly exposed populations; and (3) potential to overestimate exposures in
5 populations who alter behavior in response to high Os days (i.e., spend less time being active
6 outdoors).
7 4.4.2.2 Risk-Based Considerations: Lung Function
8 As discussed above in more detail in section 3.2.3.1 of this second draft PA, the
9 assessment of lung function risks presented in the second draft HREA (U.S. EPA, 2014)
10 provides estimates of the number and percent of people experiencing (Vinduced lung function
11 decrements greater than or equal to 10, 15, and 20%. In the last review, CAS AC advised EPA to
12 focus on decrements of 10% or greater when considering people with pre-existing lung disease
13 (Samet, 2011).
14 Lung function risk estimates are based on an updated dose-threshold model that estimates
15 FEVi responses for individuals following short-term exposures to O^ (McDonnell, Stewart, and
16 Smith, 2010), reflecting methodological improvements since the last review (U.S. EPA, 2014,
17 section 6.2.4). The approach taken in the second draft HREA to estimating Os-induced lung
18 function decrements, and the key uncertainties associated with these estimates, are summarized
19 in section 3.2.3.1 for air quality adjusted to just meet the current standard and are discussed in
20 more detail in chapter 6 of the HREA (U.S. EPA, 2014).
21 As discussed in section 3.2.3.1, in evaluating potential alternative standard levels we
22 focus on modeled exposures for school-age children, with an emphasis on asthmatic children. As
23 with exposures of concern, the percentages of all school age children and asthmatic school age
24 children estimated to experience particular (Vinduced lung function decrements are virtually
25 indistinguishable.
26 In this section, we consider the following question:
27 • To what extent are potential alternative standards with revised levels estimated to
28 decrease the occurrence of Os-induced lung function decrements, compared to
29 the current standard, and what are the nature and magnitude of the decrements
30 remaining for each alternative standard level evaluated?
31 Key results related to this question are summarized below (Figures 4-5 to 4-8). Figures 4-5 and
32 4-6 present estimates of one or more Os-induced lung function decrements, and Figures 4-7 and
33 4-8 present estimates of two or more decrements.
4-27
-------
1 Figure 4-5. Percent of children estimated to experience one or more Os-induced lung function decrements greater than 10,15,
2 or 20% for air quality adjusted to just meet the current and potential alternative standards (averaged over 2006
3 to 2010)
4
5
Decrements > 10%
Decrements > 15%
Decrements > 20%
75ppb
70ppb
6S ppb 60 ppb
75 ppb 70ppb 65ppb
Standard Level
60 ppb .". |i|il. 70 ppb 65 ppb
60 pph
-Atlanta
-Baltimore
-Boston
-Chicago
-tlevelancl
-Dallas
-Denver
-Detroit
Houston
-Los Angeles
-NewYork
Philadelphia
Sacramento
St Louis
Washington
4-28
-------
1 Figure 4-6. Percent of children estimated to experience one or more Os-induced lung function decrements greater than 10,15,
2 or 20% for air quality adjusted to just meet the current and potential alternative standards (worst-case year from
3 2006 to 2010)
4
25%
0'..
Decrements > 10%
Decrements > 15%
Decrements > 20%
75ppl> 70ppl> 65pph OOppb 75ppb 70ppb 65ppb 60ppb 75ppb 70ppb 65ppb
Standard Level
60 ppb
-Atlanta
-Baltimore
-Boston
-Chicago
-Cleveland
-Dallas
-Denver
-Detroit
Houston
-Los Angeles
-New York
Philadelphia
Sacramento
St Louis
Washington
4-29
-------
1 Figure 4-7. Percent of children estimated to experience two or more Os-induced lung function decrements greater than 10,
2 15, or 20% for air quality adjusted to just meet the current and potential alternative standards (averaged over
3 2006 to 2010)
4
? c
C o
81
T-
£
o» 5
•o ^
O o,
'c o
14
Decrements > 10%
Decrements > 15%
Decrements > 20%
12".
10'.;.
8%
(,"..
4%
2'\.
0"o
75ppb
70|)|>b 65ppb
60ppb 75ppb 70ppb GSppb
Standard Level
60ppb 75ppb 70ppb 65ppb
60 ppb
-Atlanta
-Baltimore
-Boston
-Chicago
-Cleveland
-Dallas
-Denver
-Detroit
Houston
-Los Angeles
-New York
Philadelphia
Sacramento
St Louis
Washington
4-30
-------
1 Figure 4-8. Percent of children estimated to experience two or more Os-induced lung function decrements greater than 10,
2 15, or 20% for air quality adjusted to just meet the current and potential alternative standards (worst-case year
3 from 2006 to 2010)
Decrements > 10%
Decrements > 15%
Decrements > 20%
75 ppb 70pph 65ppb 60 ppb 75 ppb
70ppb 65ppb
Standard Level
60 ppb 75ppb 70 ppb 6Sppb GOppb
-Atlanta
-Baltimore
-Boston
-Chicago
Cle. eland
-Dallas
-Denver
-Detroit
Houston
-Los Angeles
-New York
Philadelphia
Sacramento
St Louis
Washington
4-31
-------
1 As illustrated above in Figures 4-5 to 4-8, adjusting air quality to just meet progressively
2 lower potential alternative standard levels consistently reduces the percent of children estimated
3 to experience potentially adverse lung function decrements. These results reflect the consistent
4 reductions in the highest ambient 63 concentrations upon model adjustment (section 3.2.1; U.S.
5 EPA, 2014, chapter 4).16 Based on Figures 4-5 to 4-8 and the associated details described in the
6 second draft UREA (U.S. EPA 2014, chapter 6), we take note of the following with regard to
7 specific potential alternative standard levels:
8 1. For a standard level of 70 ppb:
9 a. On average over the years 2006 to 2010, a standard with a level of 70 ppb is
10 estimated to allow approximately 11 to 17% of children in urban case study areas,
11 including asthmatic children, to experience one or more Os-induced lung function
12 decrements > 10% (approximately 6 to 27% reduction, relative to current
13 standard) per season. Summing across case study areas, these percentages
14 correspond to approximately 260,000 asthmatic children experiencing
15 approximately 1 million total occurrences of Cb-induced lung function
16 decrements greater than or equal to 10%.
17
18 b. On average over the years 2006 to 2010, a standard with a level of 70 ppb is
19 estimated to allow approximately 6 to 11% of children, including asthmatic
20 children, to experience two or more Os-induced lung function decrements > 10%
21 (approximately 8 to 30% reduction, relative to current standard).
22
23 c. In the worst-case years, a standard with a level of 70 ppb is estimated to allow
24 approximately 14 to 20% of children, including asthmatic children, to experience
25 one or more Os-induced lung function decrements >10%, and approximately 7 to
26 13% to experience two or more such decrements.
27
28 d. On average over the years 2006 to 2010, a standard with a level of 70 ppb is
29 estimated to allow approximately 2 to 4% of children, including asthmatic
30 children, to experience one or more Os-induced lung function decrements > 15%,
31 and approximately 1 to 2.5% of children to experience two or more such Os-
32 induced decrements. In the worst-case years, approximately 3 to 5% of children
33 are estimated to experience one or more Os-induced lung function decrements
34 >15%, and approximately 1 to 3% are estimated to experience two or more such
35 decrements.
36
37 e. A standard with a level of 70 ppb is estimated to allow 2% or fewer children to
38 experience any Os-induced lung function decrements > 20%, even in the worst-
39 case years. Approximately 1% or fewer children are estimated to experience two
16The impact of the dose threshold in the lung function risk model is that O3-induced FEV1 decrements result
primarily from exposures to O3 concentrations above about 40 ppb (US EPA, 2013, chapter 6).
4-32
-------
1 or more Os-induced lung function decrements > 20%, even in the worst-case
2 years.
3
4 2. For a standard level of 65 ppb:
5 a. On average over the years 2006 to 2010, a standard with a level of 65 ppb is
6 estimated to allow approximately 3 to 15% of children, including asthmatic
7 children, to experience one or more Os-induced lung function decrements > 10%
8 (approximately 20 to 77% reduction, relative to current standard). Summing
9 across urban case study areas, these percentages correspond to approximately
10 190,000 asthmatic children experiencing almost 750,000 total occurrences of Os-
11 induced lung function decrements > 10%.
12
13 b. On average over the years 2006 to 2010, a standard with a level of 65 ppb is
14 estimated to allow approximately 1 to 9% of children, including asthmatic
15 children, to experience two or more Os-induced lung function decrements > 10%
16 (approximately 20 to 80% reduction, relative to current standard).
17
18 c. In the worst-case years, a standard with a level of 65 ppb is estimated to allow
19 approximately 4 to 18% of children to experience one or more Os-induced lung
20 function decrements > 10%, and approximately 2 to 11% to experience two or
21 more such decrements.
22
23 d. On average over the years 2006 to 2010, a standard with a level of 65 ppb is
24 estimated to allow approximately 3% or less of children to experience one or
25 more Os-induced lung function decrements > 15%, and approximately 2% or less
26 of children to experience two or more such Os-induced decrements. In the worst-
27 case years, approximately 4% or less of children are estimated to experience one
28 or more Os-induced lung function decrements > 15%, and up to approximately
29 2.5% are estimated to experience two or more such decrements.
30
31 e. A standard with a level of 65 ppb is estimated to allow less than 1.5% of children
32 to experience any Os-induced lung function decrements > 20%, even in the worst-
33 case years. A standard with a level of 65 ppb is estimated to allow less than 1% of
34 children to experience two or more Os-induced lung function decrements > 20%,
35 even in the worst-case years.
36
37 3. For a standard level of 60 ppb:
38 a. On average over the years 2006 to 2010, a standard with a level of 60 ppb is
39 estimated to allow approximately 5 to 11% of children, including asthmatic
40 children, to experience one or more Os-induced lung function decrements > 10%
41 (approximately 35 to 77% reduction, relative to current standard). Summing
42 across urban case study areas, these percentages correspond to approximately
43 140,000 asthmatic children experiencing approximately 500,000 total occurrences
44 of Os-induced lung function decrements > 10%.
45
4-33
-------
1 b. On average over the years 2006 to 2010, a standard with a level of 60 ppb is
2 estimated to allow approximately 2 to 6% of children to experience two or more
3 Os-induced lung function decrements > 10% (approximately 40 to 70% reduction,
4 relative to current standard).
5
6 c. In the worst-case years, a standard with a level of 60 ppb is estimated to allow
7 approximately 5 to 13% of children to experience one or more Os-induced lung
8 function decrements > 10%, and approximately 2 to 7% to experience two or
9 more such decrements.
10
11 d. A standard with a level of 60 ppb is estimated to allow less than about 3% of
12 children to experience any Os-induced lung function decrements > 15% and less
13 than 1% to experience decrements greater than 20%, even in years with the
14 highest exposure estimates. A standard with a level of 60 ppb is estimated to
15 allow less than 1.5% of children to experience two or more Os-induced lung
16 function decrements > 15% and less than 0.5% to experience two or more
17 decrements > 20%, even in years with the highest exposure estimates.
18 In further considering these exposure estimates, we take note of the associated
19 uncertainties, as discussed in more detail in section 3.2.2 of this second draft PA. In addition to
20 the uncertainties in exposure estimates noted above, these include the relative lack of exposure-
21 response information for key at-risk populations (i.e., children and asthmatics), since most
22 controlled human exposures studies are conducted in healthy adults.
23 4.4.2.3 Risk-Based Considerations: Epidemiology-Based Mortality and Morbidity
24 The epidemiology-based risk assessments presented in the second draft HREA (U.S.
25 EPA, 2014, chapter 7) provide estimates of total mortality, respiratory hospital admissions and
26 emergency department visits, and asthma exacerbations associated with short-term Os
27 concentrations. The HREA also presents estimates of respiratory mortality associated with long-
28 term17 concentrations. In evaluating these risk estimates, we consider the following question:
29 • To what extent are potential alternative standards with revised levels estimated to
30 decrease Os health risks, compared to the current standard, and what are the
31 nature and magnitude of the health risks remaining for each alternative standard
32 level evaluated?
33 As discussed in more detail in section 3.2.3.2 of this second draft PA, in considering this
34 question we are mindful that the model-based approach used to adjust air quality in the second
Estimates of respiratory mortality associated with long-term O3 concentrations are based on the study by Jerrett et
al. (2009). Consistent with the O3 metric used in the study, risk estimates are based on seasonal averages of 1-hour
daily max O3 concentrations.
4-34
-------
1 draft HREA has important implications for risk estimates developed by applying concentration-
2 response relationships from epidemiologic studies (section 3.2.1). In particular, we note the
3 uncertainty associated with using such concentration-response relationships to estimate risks for
4 model-adjusted air quality with spatial and temporal patterns of ambient 63 that are different
5 from those present in the epidemiologic study locations. In addition, given the use of linear
6 concentration-response relationships, risk estimates are equally influenced by decreasing high O^
1 concentrations and increasing low 63 concentrations following model adjustment, when the
8 increases and decreases are of equal magnitude. These and other uncertainties associated with
9 risk estimates are discussed in section 3.2.3.2.
10 Key results from the second draft HREA (U.S. EPA, 2014, chapter 7) are summarized
11 below for estimates of total mortality associated with short-term 03 concentrations (Figures 4-9
12 and 4-10), respiratory hospital admissions associated with short-term 63 concentrations (Figure
13 4-11), and respiratory mortality associated with long-term Os concentrations (Figure 4-12). The
14 other morbidity effects evaluated in the second draft HREA (i.e., respiratory emergency
15 department visits and asthma symptoms associated with short-term concentrations) exhibit
16 patterns across standard levels that are similar to those reported for total mortality and respiratory
17 hospital admissions (U.S. EPA, 2014, chapter 7).
18 As discussed in section 3.2.3.2, for total mortality associated with short-term 63
19 concentrations we consider estimates of risk based on the full distributions of area-wide 03
20 concentrations (Figure 4-9) and estimates of risk associated with various portions of those
1 &
21 distributions (Figure 4-10). In doing so, we recognize the reduced certainty in a linear
22 concentration-response relationship at the lower ends of air quality distributions, and the greater
23 certainty in increased incidence and severity of effects at higher exposure concentrations
24 (discussed in more detail in section 3.2.3.2).19
18The second draft HREA does not present distributions of risk over distributions of area-wide concentrations for
other epidemiology-based risk endpoints (U.S. EPA, 2014, chapter 7).
19As discussed in section 3.1.2.2, as ambient concentrations increase the potential for exposures to higher O3
concentrations also increases. Thus with increasing ambient concentrations, controlled human exposure and animal
lexicological studies provide greater certainty in the increased incidence, magnitude, and severity of O3-attributable
effects.
4-35
-------
1 Figure 4-9. Estimates of Total Mortality Associated with Short-Term Os Concentrations in Urban Case Study Areas (Air
2 Quality Adjusted to Current and Potential alternative standard levels) - Total Risk
3
4
2007 Model Adjustment
2009 Model Adjustment
75epb
70ppb
65ppb
SQppb
75ppb
70pcb
-Atlanta. GA
-Baltimore. MO
-Boston. MA
-Cleveland. OH
-Oenvtr.CO
-Detroit Ml
- Houston. TX
•Los Angeles. CA
-New York. NY
-Philadelphia. PA
-Sacramento. CA
-St Louis. MO
4-36
-------
1 Figure 4-10. Estimates of Total Mortality Attributable to Days with 8-Hour Area-Wide Os Concentrations at or above 20, 40,
2 or 60 ppb, Summed Across Urban Case Study Areas (Air Quality Adjusted to Current and Potential alternative
3 standard levels)
IOIMII)
2007 Model Adjustment
2009 Model Adjustment
4
5
6
All days 20+ppb 40+ppb 60+ppb
All days 20+ppb 40+ppb 60+ppb
4-37
-------
1 Figure 4-11. Estimates of Respiratory Hospital Admissions Associated with Short-Term Os Concentrations in Urban Case
2 Study Areas (Air Quality Adjusted to Current and Potential alternative standard levels) - Total Risk
2007 Model Adjustment
2009 Model Adjustment
3
4
-Attent*. CA
-atfemore. MO
-Sotteo.MA
-Ctt.«l»«->So. CA
-H loun. MO
4-38
-------
1 Figure 4-12. Estimates of Respiratory Mortality Associated with long-term Os Concentrations in Urban Case Study Areas (Air
2 Quality Adjusted to Current and Potential alternative standard levels) - Total Risk
2007 Model Adjustment
2009 Model Adjustment
-Atlanta, GA
•Baltimore. MO
-Boston. MA
-Cleveland. OH
-Denver. CO
-Detroit. Ml
- Houston. TX
-Los Angeles, CA
-NevvYork.NY
-Philadelphia. PA
- Sacramento, C A
-St. Louis.MO
75ppb
70ppb
65PPb
60ppb
75ppb
70ppb
6Sppb
60ppb
4-39
-------
1 Based on Figures 4-9 to 4-12 and the associated details described in the second draft
2 HREA (U. S. EPA 2014, chapter 7), we take note of the following for a standard level of 70 ppb:
3 1. Total mortality associated with short-term Os concentrations:
4 a. A standard level of 70 ppb results in modest changes in total risk, compared to the
5 current standard. Across urban case study areas, risks are estimated to decrease by
6 up to approximately 5%. These risk reductions are estimated most consistently for
7 the model year with generally higher Os-associated risks (2007). In the year with
8 generally lower risks (2009), a standard level of 70 ppb results in either no change
9 or small reductions in estimated risks in most urban case study areas. In one area
10 (Detroit) for the 2009 model year, (^-associated mortality is estimated to increase
11 by approximately 4%, compared to the current standard (see section 3.2.3.2 for
12 further discussion of increased risk estimates following model adjustment20).
13
14 b. When summed across urban case study areas, a standard level of 70 ppb is
15 estimated to reduce Os-associated total mortality by approximately 4% (2007
16 model year) and 2% (2009 model year), compared to the current standard. For
17 days with area-wide concentrations at or above 40 ppb, a standard level of 70 ppb
18 is estimated to reduce (^-associated total mortality by approximately 9% (2007
19 model year) and 8% (2009 model year). For days with area-wide concentrations at
20 or above 60 ppb, a standard level of 70 ppb is estimated to reduce (^-associated
21 total mortality by approximately 50% (2007 model year) and 70% (2009 model
22 year).21
23
24 2. Respiratory hospital admissions associated with short-term 63 concentrations: Compared to
25 the current standard, changes in total risk estimated for a standard level of 70 ppb are similar
26 to the changes in total risks estimated for total mortality (U.S. EPA, 2014, chapter 7).
27
28 3. Respiratory mortality associated with long-term 63 concentrations: A standard level of 70
29 ppb reduces total risk, compared to the current standard. Across urban case study areas, risks
30 are estimated to decrease by up to approximately 8%. These risk reductions are estimated
31 most consistently for the model year with generally higher Os-associated risks (2007). In the
32 year with generally lower 63 concentrations (2009), a standard level of 70 ppb results in
33 either no change or small reductions in estimated risks in most urban case study areas. In one
34 area (Detroit) for the 2009 model year, (^-associated mortality is estimated to increase by
35 approximately 1%, compared to the current standard.
20 As discussed in more detail above (section 3.2.3.2), because of the influence of the entire distribution of ambient
O3 concentrations on total risk estimates, the impacts of adjusting air quality to just meet potential alternative
standards are more modest, and are less directionally consistent across urban case study areas, than observed for
exposures of concern or O3-induced lung function decrements.
21These results reflect the fact that increases in area-wide O3 concentrations upon model adjustment occur primarily
at relatively low concentrations (i.e., at or below area-wide concentrations of approximately 45 ppb) (U.S. EPA,
2014, section 4.3.3.2 and appendix 7B).
4-40
-------
1 Based on Figures 4-9 to 4-12 and the associated details described in the second draft
2 HREA (U. S. EPA 2014, chapter 7), we take note of the following for a standard level of 65 ppb:
3 1. Total mortality associated with short-term Os concentrations:
4 a. A standard level of 65 ppb results in small changes in total risk, compared to the
5 current standard. Across most urban case study areas, risks are estimated to
6 decrease by up to approximately 9%. In one area (New York City), risks are
7 estimated to decrease by up to approximately 22%.22 These risk reductions are
8 estimated most consistently for the model year with generally higher 63-
9 associated risks (2007). In the year with generally lower risks (2009), a standard
10 level of 65 ppb results in smaller reductions in estimated risks in most urban case
11 study areas. In one area (Detroit) for the 2009 model year, Os-associated mortality
12 is estimated to increase by approximately 1% compared to the current standard.
13
14 b. When summed across urban case study areas, a standard level of 65 ppb is
15 estimated to reduce Cb-associated total mortality by approximately 11% (2007
16 model year) and 8% (2009 model year), compared to the current standard. For
17 days with area-wide concentrations at or above 40 ppb, a standard level of 65 ppb
18 is estimated to reduce (^-associated total mortality by almost 40% (2007 and
19 2009 model years). For days with area-wide concentrations at or above 60 ppb, a
20 standard level of 65 ppb is estimated to reduce (^-associated total mortality by
21 over 80% (2007 and 2009 model years).
22
23 2. Respiratory hospital admissions associated with short-term Os concentrations: Compared to
24 the current standard, changes in total risk estimated for a standard level of 65 ppb are similar
25 to the changes in total risk estimated for total mortality (U.S. EPA, 2014, chapter 7).
26
27 3. Respiratory mortality associated with long-term Os concentrations: A standard level of 65
28 ppb reduces total risk, compared to the current standard. Across most urban case study areas,
29 risks are estimated to decrease by up to approximately 10%. In one area (New York City),
30 risks are estimated to decrease by up to approximately 22%. Risk reductions are estimated
31 across all urban case study areas and in both model years evaluated, with larger reductions
32 estimated for 2007 (i.e., the model year with generally higher (^-associated risks).
33 Based on Figures 4-9 to 4-12 and the associated details described in the second draft
34 HREA (U.S. EPA 2014, chapter 7), we take note of the following for a standard level of 60 ppb:
35 1. Total mortality associated with short-term Os concentrations:
36 a. A standard level of 60 ppb is estimated to reduce total risk, compared to the
37 current standard, in all urban case study areas. Across urban case study areas,
22As discussed in the second draft HREA (U.S. EPA, 2014, section 4.5), the New York and Los Angeles urban case
study areas required the largest reductions in NOX in order to meet the existing and potential alternative standards.
The HDDM-based O3 estimates become more uncertain for larger changes in precursor emissions, and the HREA
notes less overall confidence in results for New York and Los Angeles.
4-41
-------
1 risks are estimated to decrease by up to approximately 15%. Estimated risk
2 reductions are larger for the model year with generally higher Os-associated risks
3 (2007).
4
5 b. When summed across urban case study areas, a standard level of 60 ppb is
6 estimated to reduce Os-associated total mortality by approximately 14% (2007
7 model year) and 10% (2009 model year), compared to the current standard. For
8 days with area-wide concentrations at or above 40 ppb, a standard level of 60 ppb
9 is estimated to reduce (^-associated total mortality by approximately 50% (2007
10 and 2009 model years). For days with area- wide concentrations at or above 60
1 1 ppb, a standard level of 60 ppb is estimated to reduce (^-associated mortality by
12 over 95% (2007 and 2009 model years).
13
14 2. Respiratory hospital admissions associated with short-term Os concentrations: Compared to
15 the current standard, changes in total risk estimated for a standard level of 60 ppb are similar
16 to the changes in total risk estimated for total mortality (U.S. EPA, 2014, chapter 7).
17
18 3. Respiratory mortality associated with long-term Os concentrations: A standard level of 60
19 ppb reduces total risk, compared to the current standard. Across urban case study areas, risks
20 are estimated to decrease by up to approximately 17%. Risk reductions are estimated across
21 all urban case study areas and in both model years evaluated, with larger reductions
22 estimated for 2007 (i.e., the model year with generally higher Os-associated risks).
23 In further considering these risk estimates, we take note of the associated uncertainties, as
24 discussed in more detail in section 3.2.3.2 of this second draft PA. These include (1) the national
25 representativeness of urban case study areas in terms of the 63 response to reductions in NOx
26 emissions; (2) the shape of the concentration-response function at lower ambient concentrations;
27 (3) the use of concentration-response relationships developed for particular populations in
28 particular locations to estimate health risks in different populations and locations; (4) the
29 applications of concentration-response relationships to model-adjusted air quality, given the
30 altered spatial/temporal patterns of ambient 63 and the potential for increases in relatively low
31 Os concentrations to increase risk estimates; and (5) the possibility for reductions in risk
32 associated with reductions in PM and/or NC>2 resulting from control
33 4.5 CASAC ADVICE
34 In the fall of 201 1, rather than revising the Os NAAQS as part of the reconsideration
35 process, EPA coordinated further proceedings on the reconsideration rulemaking with the current
36 ongoing periodic review. Accordingly, in this section we are briefly describing CASAC advice
37 from the reconsideration of the 2008 final decision as well as CASAC advice received during the
38 current review as it pertains to potential alternative standards.
4-42
-------
1 Consistent with their advice in 2008, CASAC reiterated during the reconsideration its
2 support for an 8-hour primary 63 standard with a level ranging from 60 to 70 ppb, combined
3 with the current form. Specifically, in response to EPA's solicitation of their advice during the
4 reconsideration, the CASAC letter (Samet 2010) to the Administrator stated:
5 CASA Cfully supports EPA 's proposed range of0.060 - 0.070 parts per million
6 (ppm)for the 8-hour primary ozone standard. CASAC considers this range to be
1 justified by the scientific evidence as presented in the Air Quality Criteria for
8 Ozone and Related Photochemical Oxidants (March 2006) and Review of the
9 National Ambient Air Quality Standards for Ozone: Policy Assessment of
10 Scientific and Technical Information, OAQPS Staff Paper (July 2007).
11
12 Similarly, in response to EPA's request for additional advice on the reconsideration in
13 2011, CASAC reaffirmed their conclusion that "the evidence from controlled human and
14 epidemiological studies strongly supports the selection of a new primary ozone standard within
15 the 60 - 70 ppb range for an 8-hour averaging time" (Samet, 2011). CASAC further concluded
16 that this range "would provide little margin of safety at its upper end" (Samet, 2011, p. 2).
17 In the first draft PA, staff concluded that the available evidence provides support for
18 conducting further exposure and risk analyses of potential alternative standard levels in the range
19 of 60 to 70 ppb (USEPA, 2012b). In response, CASAC noted that the draft PA provided "a
20 strong scientific rationale for consideration of ozone levels (8 hour averages) of 60 ppb to 70
21 ppb" (Frey and Samet, 2012).
22 4.6 PRELIMINARY STAFF CONCLUSIONS ON ALTERNATIVE PRIMARY
23 STANDARDS FOR CONSIDERATION
24 Staffs consideration of alternative primary Os standards builds upon our conclusion,
25 discussed in section 3.4, that the overall body of evidence and exposure/risk information calls
26 into question the adequacy of public health protection afforded by the current standard,
27 particularly for at-risk populations. In section 3.4, we further conclude that it is appropriate in
28 this review to consider alternative standards that would increase public health protection,
29 compared to the current standard, and that it is not appropriate to consider alternative standards
30 with levels higher than the current standard, which would decrease public health protection.
31 As an initial matter, for the reasons discussed in section 4.1 above, we conclude it is
32 appropriate to continue using Os as the indicator for the standard that protects against exposures
33 to ambient Os and other photochemical oxidants. For the reasons discussed in sections 4.2 and
34 4.3 above, we also conclude that it is appropriate for the Administrator to consider retaining the
35 current averaging time and form. In the remainder of this section, we present our more focused
36 discussion on the range of alternative levels that, in our judgment, it is appropriate for the
37 Administrator to consider.
4-43
-------
1 For a standard that is defined in terms of the current indicator, averaging time, and form,
2 we reach the conclusion that, depending on the public health policy judgments made by the
3 Administrator, the evidence and information available in this review supports consideration of
4 alternative levels from 70 down to 60 ppb (section 4.4, above).23 Compared to the current
5 standard, a revised standard with a level from 70 to 60 ppb would be expected to increase public
6 health protection against both short- and long-term Os exposures, including for members of at-
7 risk populations. The scientific evidence and exposure/risk estimates that could support revised
8 standards with levels from the upper, middle, and lower portions of this range are summarized
9 below, with a specific focus on levels of 70 ppb, 65 ppb, and 60 ppb. Key exposure/risk
10 information is summarized in Tables 4-4 and 4-5, and Figure 4-13.
11
23 As discussed in sections 3.1.2, 3.2, and 3.4 of this second draft PA, we further conclude that it would not be
appropriate to consider a standard level higher than 75 ppb, which would decrease public health protection
compared to the current standard.
4-44
-------
1
2
Table 4-4. Summary of Estimated Exposures of Concern for Potential alternative
standard levels of 70, 65, and 60 ppb in Urban Case Study Areas
24
Benchmark
Level
Alternative
Standard
Level (ppb)
Average %
Children
Exposed25
Number of Children (5
to 18 years)
[Number of Asthmatic
Children]26
Average %
Reduction from
Current
Standard
% Children -
Worst Year and
Worst Area
One or more exposures of concern per season
> 70 ppb
> 60 ppb
70
65
60
70
65
60
0.1-1.2
0-0.2
0
3.3-10.2
0-4.2
0-1.2
95,000 [10,000]
16,000 [1,600]
0 [0]
1,177,000 [122,000]
392,000 [40,000]
69,000 [7,000]
73
95
100
46
80
96
3.2
0.5
0.1
18.9
9.5
2.2
Two or more exposures of concern per season
> 70 ppb
> 60 ppb
70
65
60
70
65
60
0-0.1
0
0
0.5-3.5
0-0.8
0-0.2
3,000 [360]
0 [0]
0 [0]
319,000 [33,000]
65,000 [7,000]
3,800 [400]
95
100
100
61
92
100
0.4
0
0
9.2
2.8
0.3
3
4
As illustrated above in Figures 4-1 to 4-4, all alternative standard level s evaluated in the HREA were effective at
limiting exposures of concern at or above 80 ppb. Therefore, Table 4-4 focuses on exposures of concern at or above
the 70 and 60 ppb benchmark concentrations.
25Estimates for each urban case study area were averaged for the years evaluated in the second draft HREA (2006 to
2010). Ranges reflect the ranges across urban case study areas.
26Numbers of children exposed in each urban case study area were averaged over the years 2006 to 2010. These
averages were then summed across urban case study areas. Numbers are rounded to nearest thousand unless
otherwise indicated.
4-45
-------
1 Table 4-5.
2
Summary of Estimated Lung Function Decrements for Potential alternative
standard levels of 70, 65, and 60 ppb in Urban Case Study Areas
Lung
Function
Decrement
Alternative
Standard
Level
Average %
Children27
Number of Children (5
to 18 years) [Number of
Asthmatic Children]28
Average %
Reduction from
Current Standard
% Children
Worst Year and
Area
One or more decrements per season
> 10%
> 15%
> 20%
70
65
60
70
65
60
70
65
60
11-17
3-15
5-11
2-4
0-3
1-2
1-2
0-1
0-1
2,547,000 [263,000]
1,931,000 [195,000]
1,392,000 [139,000]
564,000 [58,000]
355,000 [36,000]
224,000 [22,000]
189,000 [20,000]
107,000 [11,000]
57,000 [6,000]
15
31
45
26
50
67
32
59
77
20
18
13
5
4
o
J
2.1
1.4
0.7
Two or more decrements per season
> 10%
> 15%
> 20%
70
65
60
70
65
60
70
65
60
5.5-11
1.3-8.8
2.1-6.4
0.9-2.4
0.1-1.8
0.2-1.0
0.3-0.8
0-0.5
0-0.2
1,418,000 [146,000]
1,020,000 [102,000]
743,000 [74,000]
274,000 [28,000]
169,000 [17,000]
100,000 [10,000]
84,000 [9,000]
40,000 [4,000]
22,000 [2,000]
17
37
51
29
54
71
34
66
83
13
11
7.3
3.1
2.3
1.4
1.1
0.8
0.4
27Estimates in each urban case study area were averaged for the years evaluated in the second draft HREA (2006 to
2010). Ranges reflect the ranges across urban case study areas.
28Numbers of children estimated to experience decrements in each study urban case study area were averaged over
2006 to 2010. These averages were then summed across urban case study areas. Numbers are rounded to nearest
thousand unless otherwise indicated.
4-46
-------
1 Figure 4-13. Estimates of Total Mortality Attributable to Days with 8-Hour Area-Wide O
2 Concentrations at or above 20, 40, or 60 ppb - Risks Summed Across Urban
3 Case Study Areas and Expressed Relative to 75 ppb
4
5
I 75ppb
i 70 ppb
i 65 ppb
. 60 ppb
Total risk
20* ppb
40* ppb
60* ppb
4-47
-------
1 As an initial observation, we note that controlled human exposure studies provide the
2 most certain evidence indicating the occurrence of health effects in humans following exposures
3 to specific Os concentrations. As discussed above and in section 3.1.2.1, controlled human
4 exposure studies have reported a variety of respiratory effects in healthy adults following
5 exposures to Os concentrations of 60, 70, or 80 ppb, and higher. The largest respiratory effects,
6 and the broadest range of effects, have been reported following exposures to 80 ppb Os or higher,
7 in part because most exposure studies have been conducted at these higher concentrations.
8 Exposures to Os concentrations of 80 ppb or higher have been reported to decrease lung function,
9 increase airway inflammation, increase respiratory symptoms, result in airway
10 hyperresponsiveness, and decrease lung host defenses in healthy adults. Most of these effects
9Q
11 have also been reported in healthy adults following exposures to lower Os concentrations.
12 Exposures to Os concentrations of 70 ppb have been reported to decrease lung function and
13 increase respiratory symptoms, a combination that meets the ATS criteria for an "adverse"
14 response (section 3.1.3). Exposures to Os concentrations of 60 ppb have been demonstrated to
15 decrease lung function, with decrements in some individuals large enough to be judged an
16 abnormal response by ATS, and which could be adverse in individuals with lung disease.
17 Exposures to Os concentrations of 60 ppb have also been reported in one study to increase
18 airway inflammation, which provides a mechanism by which Os may cause other more serious
19 respiratory effects (e.g., asthma exacerbations).
20 Compared to the current standard, the second draft HREA estimates that a revised
21 standard with a level of 70 ppb would reduce exposures of concern to Os concentrations of 60,
22 70, and 80 ppb, with such a standard level estimated to be most effective at limiting exposures at
23 or above the higher health benchmark concentrations. On average over the years 2006 to 2010, a
24 standard with a level of 70 ppb is estimated to allow only up to about 1% of children (i.e., ages 5
25 to 18) to experience exposures of concern at or above 70 ppb (73% reduction, compared to
26 current standard), and far less than 1% to experience two or more such exposures (95%
27 reduction, compared to current standard). In the worst-case location and year (i.e., location and
28 year with the largest exposure estimate), about 3% of children are estimated to experience one or
29 more exposures of concern at or above 70 ppb, and less than 1% are estimated to experience two
30 or more. A standard with a level of 70 ppb is estimated to allow far less than 1% of children to
31 experience exposures of concern at or above the 80 ppb benchmark concentration, even in the
32 worst-case year (Table 4-4).
29Though airway hyperresponsiveness and lung host defense have not been evaluated following exposures to O3
concentrations below 80 ppb. The extent to which these respiratory effects occur following lower exposure
concentrations is not clear from the available evidence, though we have no basis for concluding that an exposure
concentration of 80 ppb reflects an effects threshold.
4-48
-------
1 A standard with a level of 70 ppb is estimated to allow about 3 to 10% of children to
2 experience one or more exposures of concern at or above 60 ppb in a single 63 season.
3 Compared to the current standard, this reflects about a 46% reduction, on average across urban
4 case study areas. A standard with a level of 70 ppb is estimated to allow about 1% to 4% of
5 children to experience two or more exposures of concern at or above 60 ppb. In the worst-case
6 location and year, a standard set at 70 ppb is estimated to allow about 19% of children to
7 experience one or more exposures of concern at or above 60 ppb, and 9% to experience two or
8 more such exposures (Table 4-4).
9 Compared to the current standard, the second draft HREA estimates that a revised
10 standard with a level of 70 ppb would also reduce (Vinduced lung function decrements in
11 children. A level of 70 ppb is estimated to be most effective at limiting the occurrences of
12 moderate and large lung function decrements (i.e., FEVi decrements > 15% and > 20%,
13 respectively). On average over the years 2006 to 2010, a standard with a level of 70 ppb is
14 estimated to allow about 2 to 4% of children to experience one or more moderate Os-induced
15 lung function decrements (i.e., > 15%), which would be of concern for healthy people, and about
16 1 to 2.5% of children to experience two or more such decrements (approximately 30% reduction,
17 compared to the current standard). In the worst-case location and year, up to 5% of children are
18 estimated to experience one or more (Vinduced lung function decrements > 15%, and up to 3%
19 are estimated to experience two or more such decrements. A standard set at 70 ppb is estimated
20 to allow about 2% or fewer children to experience large (Vinduced lung function decrements
21 (i.e., > 20%), and to allow about 1% or fewer children to experience two or more such
22 decrements, even in the years and locations with the largest exposure estimates (Table 4-5).
23 On average over the years 2006 to 2010, a standard set at 70 ppb is estimated to allow
24 about 11 to 17% of children to experience one or more moderate Os-induced lung function
25 decrements (i.e., > 10%), which could be adverse for people with lung disease. This reflects an
26 average reduction of about 15%, compared to current standard. A standard with a level of 70 ppb
27 is also estimated to allow about 6 to 11% of children to experience two or more such decrements
28 (17% reduction, compared to current standard). In the worst-case location and year, a standard
29 set at 70 ppb is estimated to allow about 20% of children, including asthmatic children, to
30 experience one or more Os-induced lung function decrements > 10%, and 13% to experience two
31 or more such decrements (Table 4-5).
32 With regard to our analyses of epidemiologic studies we note that, compared to the
33 current standard, a revised standard with a level of 70 ppb would be more effective in
34 maintaining short-term ambient Oj concentrations below those present in locations that provided
35 the basis for positive and statistically significant health effect associations. In particular, the
36 study by Mar and Koenig (2009) reported positive and statistically significant associations with
4-49
-------
1 respiratory emergency department visits in Seattle, a location that met the current standard over
2 the entire study period but that would have violated a revised standard with a level of 70 ppb. In
3 addition, most of the Canadian study cities evaluated by Dales et al. (2006), who reported
4 positive and statistically significant associations with respiratory hospital admissions, would
5 have met the current standard over the entire study period but would have violated a revised
6 standard with a level of 70 ppb over at least part of the study period (Table 4-1). Finally,
7 compared to the current standard, fewer of the study cities evaluated by Katsouyanni et al.
8 (2009) (for mortality and hospital admissions) and Bell et al. (for the 30 ppb cut-point) would
9 have met a revised standard with a level of 70 ppb (Tables 4-1, 4-2).30
10 With regard to long-term 63 concentrations, air quality analyses indicate that in 9 out of
11 the 12 urban case study areas a revised standard with a level of 70 ppb would be expected to
12 maintain long-term 63 concentrations below those where the study by Jerrett et al. (2009)
13 indicates the most confidence in the reported association with respiratory mortality. This is
14 compared to 6 out of 12 areas for the current standard.
15 In further considering the potential implications of epidemiology studies for potential
16 alternative standard levels, we note that the emphasis given to estimates of Cb-associated
17 mortality or morbidity risks will depend in large part on the extent to which key uncertainties in
18 these estimates are emphasized (e.g., uncertainties in applying concentration-response
19 relationships from epidemiologic studies to adjusted air quality with different spatial/temporal
20 distributions of ambient 63). To the extent emphasis is placed on estimates of (^-associated
21 mortality and morbidity risks, it could be judged appropriate to focus on estimates of total risk
22 (i.e., based on the full distributions of ambient Os concentrations) and/or on estimates of risk
23 associated with 63 concentrations in the upper portions of ambient distributions.
24 A focus on total risks could be judged appropriate to the extent greater emphasis is placed
25 on the possibility that concentration-response relationships remain linear over the entire
26 distributions of ambient 63 concentrations, and thus to the extent greater emphasis is placed on
27 the potential for mortality and/or morbidity to be affected by changes in relatively low O?,
28 concentrations (as discussed above and in section 3.2.3.2). When summed across urban case
29 study areas, a standard with a level of 70 ppb is estimated to reduce total mortality associated
30 with short-term 03 exposures by about 2 to 3%, compared to the current standard (with
31 reductions up to about 5% for individual urban case study areas).31 Based on a national modeling
32 analysis, the majority of the U.S. population would be expected to experience modest reductions
33 in such risks upon reducing precursor emissions. A standard with a level of 70 ppb is estimated
30Though in the analyses presented in both of these studies, the majority of cities evaluated would have met a
standard with a level of 70 ppb over the entire study periods.
31 Similar changes are estimated for respiratory morbidity associated with short-term O3 concentrations.
4-50
-------
1 to reduce respiratory mortality associated with long-term Os concentrations by up to about 8% in
2 urban case study areas. For both short- and long- term metrics, risk reductions are larger, and are
3 estimated more consistently across urban case study areas, in a year with relatively higher 03
4 concentrations.
5 A focus on risks associated with Os concentrations in the upper portions of ambient
6 distributions may reasonably be judged appropriate in light of the overall evidence indicating
7 increasing magnitude, severity, and incidence of (Vattributable effects as exposure
8 concentrations increase, as well as the greater uncertainty associated with the shapes of
9 concentration-response curves for Oj, concentrations in the lower portions of ambient
10 distributions (section 3.2.3.2 and 3.4). There is no single ambient concentration below which
11 uncertainty in Os-attributable effects increases notably in all locations, for all health endpoints,
12 and in all populations. Therefore, we consider the distribution of mortality associated with
13 various portions of the distribution of area-wide Os concentrations (Figure 4-13, above).
14 For days with area-wide concentrations at or above 20 ppb, a standard with a level of 70
15 ppb is estimated to reduce mortality associated with short-term 63 exposures by about 2 to 3%
16 compared to the current standard, when Os-associated deaths are summed across urban case
17 study areas. For days with area-wide concentrations at or above 40 ppb, a standard with a level
18 of 70 ppb is estimated to reduce mortality associated with short-term 63 exposures by about 8 to
19 9% compared to the current standard. For days with area-wide concentrations at or above 60 ppb,
20 a standard with a level of 70 ppb is estimated to reduce (Vassociated mortality by about 50% to
21 70% (Figure 4-13).32
22 Based on all of the above considerations, we conclude that a standard with a level of 70
23 ppb would be expected to provide additional incremental protection over that provided by the
24 current Os standard. A level of 70 ppb could be supported to the extent more emphasis is placed
25 on the public health importance of higher 63 exposure concentrations (e.g., > 70, 80 ppb), larger
26 (Vinduced lung function decrements (e.g., > 15, 20%), and estimates of multiple occurrences of
27 exposures of concern and Os-induced lung function decrements. In addition, a revised standard
28 with a level of 70 ppb would be expected to be more effective than the current standard at
29 maintaining short- and long-term ambient Os concentrations below those where we have the
30 most confidence in associations with mortality and morbidity. Overall across the U.S., such a
32Fewer than 10% of total O3-associated deaths are attributable to days with such high area-wide concentrations, due
to the relatively small number of days with area-wide concentrations of 60 ppb or above.
4-51
-------
1 standard would also be expected to reduce risks of (^-associated mortality and morbidity,
2 particularly the portions of the risk associated with relatively high ambient concentrations.33
3 Next, we consider a standard with a level of 65 ppb. A level of 65 ppb is well-below the
4 lowest 63 exposure concentration (i.e., 80 ppb) that has been reported to elicit a wide range of
5 effects, including: lung function decrements, airway inflammation, respiratory symptoms, airway
6 hyperresponsiveness, and decreased lung host defense in healthy adults, as noted above. A
7 standard level of 65 ppb is also somewhat below the lowest exposure concentration at which the
8 combined occurrence of respiratory symptoms and lung function decrements has been reported
9 (i.e., 70 ppb), a combination judged adverse by the ATS (section 3.1.3). A level of 65 ppb is
10 above the lowest exposure concentration demonstrated to result in lung function decrements
11 large enough to be judged an abnormal response by ATS, and that could be adverse in
12 individuals with lung disease. A level of 65 ppb is also above the lowest exposure concentration
13 at which pulmonary inflammation has been reported (i.e., 60 ppb).
14 Compared to the current standard and a revised standard with a level of 70 ppb, the
15 second draft HREA estimates that a standard with a level of 65 ppb would reduce exposures of
16 concern to the range of Os benchmark concentrations analyzed (i.e., 60, 70, and 80 ppb). The
17 HREA estimates that meeting a standard with a level of 65 ppb would eliminate exposures of
18 concern at or above 80 ppb in urban case study areas. Such a standard is estimated to allow far
19 less than 1% of children to experience one or more exposures of concern at or above the 70 ppb
20 benchmark level, even in the years and locations with the largest exposure estimates, and is
21 estimated to eliminate the occurrence of two or more exposures at or above 70 ppb (Table 4-4).
22 In addition, on average over the years 2006 to 2010, a standard with a level of 65 ppb is
23 estimated to allow between 0 and about 4% of children in urban case study areas to experience
24 exposures of concern at or above 60 ppb. This reflects an 80% reduction (on average across
25 areas), relative to the current standard. A standard with a level of 65 ppb is estimated to allow
26 less than 1% of children to experience two or more exposures of concern at or above 60 ppb (>
27 90% reduction, compared to current standard). In the worst-case location and year, about 10% of
28 children are estimated to experience one or more exposures of concern at or above 60 ppb, with
29 about 3% estimated to experience two or more such exposures (Table 4-4).
30 Compared to the current standard and a revised standard with a level of 70 ppb, the
31 second draft HREA estimates that a standard with a level of 65 ppb would also reduce the
33In reaching this conclusion we recognize that, as discussed in detail in chapter 3 (sections 3.2.1 and 3.2.3.2),
reducing NOX emissions to meet alternative O3 standards could result in increases in relatively low ambient
concentrations. When we consider the epidemiologic-based estimates in light of all of the health effects evidence,
and the considerations discussed more fully in section 3.2.3.2 above, we have greater certainty in the health benefits
of reducing high ozone concentrations, and appreciable uncertainty regarding estimates of risk at lower
concentrations. Accordingly, we judge that the range of levels discussed here is appropriate.
4-52
-------
1 occurrence of Os-induced lung function decrements. A level of 65 ppb is estimated to allow
2 about 4% or less of children to experience (Vinduced lung function decrements > 15% (50%
3 reduction, compared to current standard), even considering the worst-case location and year.
4 Such a standard is estimated to allow about 2% or less of children to experience two or more
5 such decrements. A standard set at 65 ppb is estimated to allow about 1% or less of children to
6 experience large Os-induced lung function decrements (i.e., > 20%), even in the worst-case year
7 and location (Table 4-5).
8 On average over the years 2006 to 2010, a standard with a level of 65 ppb is estimated to
9 allow about 3 to 15% of children to experience one or more moderate Cb-induced lung function
10 decrements (i.e., > 10%), which could be adverse for people with lung disease. This reflects an
11 average reduction of about 30%, relative to current standard. A standard with a level of 65 ppb is
12 also estimated to allow about 1 to 9% of children to experience two or more such decrements
13 (37% reduction, compared to current standard). In the worst-case location and year, a standard
14 set at 65 ppb is estimated to allow up to about 18% of children to experience one or more 03-
15 induced lung function decrements > 10%, and up to 11% to experience two or more such
16 decrements (Table 4-5).
17 With regard to 63 epidemiologic studies we note that, compared to the current standard
18 and a standard with a level of 70 ppb, a revised standard with a level of 65 ppb would be more
19 effective in maintaining short-term ambient O?, concentrations below those present in locations
20 that provided the basis for positive and statistically significant health effect associations. In
21 addition to the studies by Mar and Koenig (2009) and Dales et al. (2006) (discussed above for a
22 level of 70 ppb), a revised standard with a level of 65 ppb would not allow the ambient 03
23 concentrations that provided the basis for mortality associations in most of the Canadian study
24 cities evaluated by Katsouyanni et al. (2009) (Table 4-1).34 In addition, fewer of the study cities
25 evaluated by Bell et al. (for the 30 ppb cut-point) would have met a revised standard with a level
26 of 65 ppb (Table 4-2).35
27 With regard to long-term O^ concentrations, air quality analyses indicate that in 10 out of
28 the 12 urban case study areas a revised standard with a level of 65 ppb would be expected to
29 maintain long-term Os concentrations below those where the study by Jerrett et al. (2009)
30 indicates the most confidence in the reported association with respiratory mortality. This is
31 compared to 6 out of 12 areas for the current standard and 9 out of 12 for a standard with a level
32 of 70 ppb (Table 4-3).
34As discussed above, most of the study cities evaluated by Katsouyanni et al. (2009) for mortality would have met
the current standard and a revised standard with a level of 70 ppb.
35Though the majority of cities evaluated, based on the 30 ppb cut point analysis, would have met a standard with a
level of 65 ppb over the entire study period (Table 4-2).
4-53
-------
1 Compared to the current standard and a revised standard with a level of 70 ppb, the
2 second draft HREA estimates that a revised standard with a level of 65 ppb would generally
3 reduce (^-associated mortality and morbidity risks based on estimates that use the full
4 distributions of ambient 63 concentrations. When summed across urban case study areas, a
5 standard with a level of 65 ppb is estimated to reduce total mortality associated with short-term
6 Os exposures by about 10%, compared to the current standard (Figure 4-13). Risks of respiratory
7 mortality associated with long-term exposures were reduced by up to about 10%, compared to
8 the current standard. As with a level of 70 ppb discussed above, reductions are estimated to be
9 larger and more consistent across urban case study areas for the year with relatively higher
10 ambient 63 concentrations.
11 Similar to a standard with a level of 70 ppb, risk estimates indicate that a standard with a
12 level of 65 ppb more effectively reduces mortality associated with the upper portions of
13 distributions of ambient OT, concentrations. For days with area-wide concentrations at or above
14 40 ppb, a standard level of 65 ppb is estimated to reduce (^-associated total mortality by about
15 40%, when summed across cities. For days with area-wide concentrations at or above 60 ppb, a
16 standard level of 65 ppb is estimated to reduce Cb-associated total mortality by more than 80%
17 (Figure 4-13).
18 In summary, the rationale to support a level of 65 ppb would emphasize the potential
19 public health significance of exposures of concern to 60 and 70 ppb 03, as well as 80 ppb, and of
20 Os-attributable health effects that could occur across the range of ambient 63 concentrations.
21 Compared to a level of 70 ppb, a level of 65 ppb would place more emphasis on evidence from
22 controlled human exposure studies showing lung function decrements and pulmonary
23 inflammation in healthy adults at 60 ppb 63. A standard with a level of 65 ppb would also place
24 more emphasis on the potential for mortality and morbidity to be caused by the relatively low
25 ambient 63 concentrations present in locations that would have met a standard with a level of 70
26 ppb, and on the potential for further reductions in health risks, beyond those estimated for a level
27 of 70 ppb. In addition, compared to a standard with a level of 70 ppb, a level of 65 ppb would
28 place relatively less emphasis on the uncertainties associated with the evidence for 63-
29 attributable health effects at lower exposure and ambient concentrations, and less emphasis on
30 uncertainties in the public health significance of O^ exposure and risk estimates for lower
31 ambient concentrations.
32 We next consider a standard with a level of 60 ppb. A level of 60 ppb is well-below the
33 lowest 63 exposure concentration that has been reported to elicit a wide range of potentially
34 adverse respiratory effects in healthy adults (i.e., 80 ppb), as noted above. A level of 60 ppb is
35 also well-below the concentration where the combined occurrence of respiratory symptoms and
36 lung function decrements was observed, a combination judged adverse by the ATS (discussed in
4-54
-------
1 section 3.1.3). A level of 60 ppb corresponds to the lowest exposure concentration demonstrated
2 to result in lung function decrements large enough to be judged an abnormal response by ATS,
3 and that could be adverse in individuals with lung disease. A level of 60 ppb also corresponds to
4 the lowest exposure concentration at which pulmonary inflammation has been reported.
5 Based on the HREA analyses of Os exposures of concern, a standard with a level of 60
6 ppb is estimated to eliminate exposures of concern at or above the 70 and 80 ppb benchmark
7 concentrations and to be more effective than higher standard levels at limiting exposures of
8 concern at or above 60 ppb. On average over the years 2006 to 2010, a standard with a level of
9 60 ppb is estimated to allow between 0 and about 1% of children in urban case study areas to
10 experience exposures of concern at or above 60 ppb. This reflects a 96% reduction (on average
11 across areas), compared to the current standard. A standard with a level of 60 ppb is estimated to
12 allow virtually no children to experience two or more exposures of concern at or above 60 ppb.
13 In the location and year with the highest exposure estimate, about 2% of children are estimated
14 to experience exposures of concern at or above 60 ppb, with far less than 1% estimated to
15 experience two or more such exposures (Table 4-4).
16 Based on the HREA analyses of Cb-induced lung function decrements, a standard with a
17 level of 60 ppb would be expected to be more effective than a level of 70 or 65 ppb at limiting
18 the occurrence of (Vinduced lung function decrements. A standard with a level of 60 ppb is
19 estimated to allow about 2% or less of children to experience one or more Os-induced lung
20 function decrements > 15% (almost 70% reduction, compared to current standard), and about 1%
21 or less to experience two or more such decrements (3% in the location and year with the largest
22 estimates). A standard set at 60 ppb is estimated to allow about 1% or less of children to
23 experience large (Vinduced lung function decrements (i.e., > 20%), even in the worst-case
24 locations and year (Table 4-5).
25 On average over the years 2006 to 2010, a standard with a level of 60 ppb is estimated to
26 allow about 5 to 11% of children, including asthmatic children, to experience one or more Oj-
27 induced lung function decrements > 10%. This reflects an average reduction of about 45%,
28 compared to current standard. A standard with a level of 60 ppb is also estimated to allow about
29 2 to 6% of children to experience two or more such decrements (51% reduction, compared to
30 current standard). In the worst-case location and year, a standard set at 60 ppb is estimated to
31 allow up to about 13% of children to experience one or more Os-induced lung function
32 decrements > 10%, and 7% to experience two or more such decrements (Table 4-5).
33 A revised standard with a level of 60 ppb would be more effective than a level of 70 or
34 65 ppb at maintaining short-term ambient Oj concentrations below those present in U.S. and
35 Canadian epidemiologic study locations during study periods. Specifically, in all of the U.S. and
36 Canadian epidemiologic studies evaluated, the majority of study cities had ambient Os
4-55
-------
1 concentrations that would likely have violated a standard with a level of 60 ppb. This includes
2 the studies noted above, for which the majority of study cities would have met the current
3 standard or a revised standard with a level of either 70 or 65 ppb (Tables 4-1 and 4-2).
4 With regard to long-term Oj, concentrations, air quality analyses indicate that in all of the
5 urban case study areas evaluated a revised standard with a level of 60 ppb would be expected to
6 maintain long-term Os concentrations below those where the study by Jerrett et al. (2009)
7 indicates the most confidence in the reported association with respiratory mortality. This is
8 compared to 6 out of 12 areas for the current standard, 9 out of 12 for a standard with a level of
9 70 ppb, and 10 out of 12 for a standard with a level of 65 ppb (Table 4-3).
10 A standard with a level of 60 ppb would also be estimated to reduce epidemiology-based
11 mortality and morbidity risks, compared to the current standard and potential alternative
12 standards with levels of 70 or 65 ppb. Across urban case study areas, a standard set at 60 ppb is
13 estimated to reduce total mortality associated with short-term Os concentrations by up to 15%,
14 and respiratory mortality associated with long-term Os concentrations by up to 17%. Estimated
15 risk reductions are larger for the model year with generally higher (^-associated risks (2007).
16 When summed across urban case study areas, a standard set at 60 ppb is estimated to reduce total
17 (^-associated mortality by about 10% (2009) and 14% (2007), compared to the current standard
18 (Figure 4-13).
19 As for standard levels of 70 and 65 ppb, risk estimates indicate that a standard with a
20 level of 60 ppb more effectively reduces mortality associated with the upper portions of
21 distributions of ambient Os concentrations. For days with area-wide concentrations at or above
22 40 ppb, a standard set at 60 ppb is estimated to reduce (^-associated total mortality by
23 approximately 50%. For days with area-wide concentrations at or above 60 ppb, a standard level
24 of 60 ppb is estimated to reduce Cb-associated total mortality by over 95% (Figure 4-13).
25 Overall we note that, compared to a standard with a level of 70 or 65 ppb, a level of 60
26 ppb would place relatively more emphasis on the potential public health significance of the Oj
27 exposures and Os-attributable health effects that could occur at lower ambient concentrations. A
28 standard with a level of 60 ppb would reflect placing a relatively large emphasis on the evidence
29 from controlled human exposure studies reporting lung function decrements and pulmonary
30 inflammation in some healthy adults following exposures to 60 ppb Os, and relatively little
31 emphasis on the uncertainties associated with the public health significance of these effects. A
32 standard with a level of 60 ppb would also emphasize the importance of the small number of
33 studies that reported health effect associations in locations that would have met a revised
34 standard with a level of 65 ppb (Table 4-1), and on the potential for further reductions in health
35 risks, beyond those estimated for a level of 65 ppb. In addition, compared to a standard with a
36 level of 65 ppb, a level of 60 ppb would place relatively little emphasis on the uncertainties
4-56
-------
1 associated with the evidence for Os-attributable health effects at lower exposure and ambient
2 concentrations, and little emphasis on uncertainties in the public health significance of 63
3 exposure and risk estimates for lower ambient concentrations.
4 Based on our consideration of the available scientific evidence and exposure/risk
5 information, we also conclude that standard levels below 60 ppb are not appropriate for
6 consideration in this review. In reaching this conclusion, we take particular note of the following:
7 • The HREA estimates that meeting a standard with a level from 70 to 60 ppb would
8 effectively reduce exposures of concern and lung function decrements. In particular, a
9 level from the low end of this range could virtually eliminate exposures of concern, even
10 for the lowest health benchmark concentration supported by the available controlled
11 human exposure evidence (i.e., 60 ppb). To the extent lower exposure concentrations may
12 result in adverse respiratory effects in some individuals, a standard level from 70 to 60
13 ppb (particularly at or near 60 ppb) would be expected to also reduce exposures to 63
14 concentrations somewhat below 60 ppb.
15
16 • A revised standard with a level from 70 to 60 ppb would be more effective than the
17 current standard at maintaining short- and long-term ambient 63 concentrations below
18 those in locations that provided the basis for positive and statistically significant health
19 effect associations in epidemiologic studies. In particular, in all of the U.S. and Canadian
20 epidemiologic studies evaluated, the majority of study cities had ambient 63
21 concentrations that would likely have violated a standard with a level of 60 ppb.
22
23 • To the extent emphasis is placed on epidemiology-based mortality and/or morbidity risk
24 estimates, meeting a standard with a level from 70 to 60 ppb would generally reduce such
25 risks across the majority of the U.S. This is particularly the case for risks associated with
26 ambient Os concentrations from the upper portions of seasonal distributions. Given that
27 uncertainty in quantitative estimates of (^-associated mortality and morbidity increases
28 with the magnitude of NOx reductions required to simulate potential alternative
29 standards, risk estimates associated with standard levels below 60 ppb would be
30 associated with increasing uncertainty.
31 Given the above, we conclude that, compared to standard levels from 70 to 60 ppb, the
32 extent to which standard levels below 60 ppb could result in further public health improvements
33 becomes notably less certain. Therefore, we conclude that it is not appropriate to consider
34 standard levels below 60 ppb in this review.
35 4.7 KEY UNCERTAINTIES AND AREAS FOR FUTURE RESEARCH AND
36 DATA COLLECTION
37 It is important to highlight the uncertainties associated with establishing standards for 63
38 during and after completion of the NAAQS review process. Research needs go beyond what is
39 necessary to understand health effects, population exposures, and risks of exposure for purposes
4-57
-------
1 of setting standards. Research can also support the development of more efficient and effective
2 control strategies. In this section, we highlight areas for future health-related research, model
3 development, and data collection activities to address these uncertainties and limitations in the
4 current body of scientific evidence.
5 As has been presented and discussed in the ISA, particularly chapters 4 through 7, the
6 scientific body of evidence informing our understanding of health effects associated with long-
7 and short-term exposures to 63 has been broadened and strengthened since the 63 NAAQS
8 review completed in 2008. Still, we have concluded that OT, health research needs and priorities
9 have not changed substantially since the 2007 63 Staff Paper (EPA 2007). Key uncertainties and
10 research needs that continue to be high priority for future reviews of the health-based standards
11 are identified below:
12 (1) An important aspect of risk characterization and decision making for air quality
13 standard levels for the OT, NAAQS is the characterization of the shape of exposure-response
14 functions for Os, including the identification of potential population threshold levels. Recent
15 controlled human exposure studies of measurable lung function effects provide evidence for a
16 smooth dose-response curve without evidence of a threshold for exposures between 40 and 120
17 ppb 63 (US EPA, 2013, Figure 6-1). Considering the importance of estimating health risks in the
18 range below 80 ppb 63, additional research is needed to evaluate responses in healthy and
19 especially asthmatic individuals in the range of 40 to 70 ppb for 6-8 hr exposures while engaged
20 in moderate exertion.
21 (2) Similarly, for health endpoints reported in epidemiologic studies such as hospital
22 admissions, ED visits, and premature mortality, an important aspect of characterizing risk is the
23 shape of concentration-response functions for 63, including identification of potential population
24 threshold levels. Most of the recent studies and analyses continue to show no evidence for a clear
25 threshold in the relationships between Os concentrations commonly observed in the U.S. during
26 the 63 season and these health endpoints, though evidence indicates less certainty in the shape of
27 the concentration-response curve at the lower end of the distribution of O^ concentrations.
28 However, there continues to be heterogeneity in the Os-mortality relationship across cities (or
29 regions), including effect modifiers that are also expected to vary regionally, which are sources
30 of uncertainty. Additionally, whether or not exposure errors, misclassification of exposure, or
31 potential impacts of other copollutants may be obscuring potential population thresholds is still
32 unknown.
33 (3) The extent to which the broad mix of photochemical oxidants and more generally
34 other copollutants in the ambient air (e.g., PM, NC>2, 862, etc.) may play a role in modifying or
35 contributing to the observed associations between ambient 03 and various morbidity effects and
36 mortality continues to be an important research question. Ozone has long been known as an
4-58
-------
1 indicator of health effects of the entire photochemical oxidant mix in the ambient air and has
2 served as a surrogate for control purposes. A better understanding of sources of the broader
3 pollutant mix, of human exposures, and of how other pollutants may modify or contribute to the
4 health effects of 63 in the ambient air, and vice versa, is needed to better inform future NAAQS
5 reviews.
6 (4) As epidemiologic research has continued to be an important factor in assessing the
7 public health impacts of 63, methodological issues in epidemiologic studies have received
8 greater visibility and scrutiny. There remains a need to further examine alternative modeling
9 specifications and control of time-varying factors, and to better understand the role of
10 copollutants in the ambient air. Additionally, there remains uncertainty around the role of
11 temperature as a potential confounder or effect modifier in epidemiologic models.
12 (5) Recent animal toxicological evidence, combined with limited evidence from
13 controlled human exposure studies of cardiovascular morbidity and epidemiologic studies of
14 cardiovascular mortality, have provided evidence of both direct and indirect effects on the
15 cardiovascular system. However, additional work will need to examine biologically plausible
16 mechanisms of cardiovascular effects, expand upon preliminary evidence from controlled human
17 exposure studies, address inconsistencies observed in epidemiologic studies of cardiovascular
18 morbidity, and determine the extent to which Oi is directly implicated or works together with
19 other pollutants in causing adverse cardiovascular effects in both at-risk and the general
20 populations.
21 (6) Most epidemiologic studies of short-term exposure effects have employed time-series
22 or case-crossover study designs and have been conducted in large populations. These study
23 designs remain subject to uncertainty due to use of ambient fixed-site data serving as a surrogate
24 for ambient exposures, and to the difficulty of determining the impact of any single pollutant
25 among the mix of pollutants in the ambient air. Measurements made at stationary outdoor
26 monitors have been used as independent variables for air pollution, but the accuracy with which
27 these measurements actually reflect subjects' exposure is not yet fully understood. Also,
28 additional research is needed to improve the characterization of the degree to which discrepancy
29 between stationary monitor measurements and actual pollutant exposures introduces error into
30 statistical estimates of pollutant effects in epidemiologic studies.
31 (7) Recent studies of "long-term" 63 often evaluate associations with daily maximum
32 concentrations, averaged over the Os season. Research is needed to better understand the extent
33 to which health effects associated with such long-term metrics are attributable to long-term
34 average concentrations versus the repeated occurrence of daily maximum concentrations.
35 (8) Improved understanding of human exposures to ambient O?, and to related
36 copollutants is an important research need. Population-based information on human exposure for
4-59
-------
1 healthy adults and children and at-risk populations, including people with asthma, to ambient O^
2 concentrations, including exposure information in various microenvironments, is needed to better
3 evaluate current and future O^ exposure models. Such information is needed for sufficient
4 periods to facilitate evaluation of exposure models throughout the 63 season.
5 (9) Information is needed to improve inputs to current and future population-based Os
6 exposure and health risk assessment models. Collection of time-activity data over longer time
7 periods is needed to reduce uncertainty in the modeled exposure distributions that form an
8 important part of the basis for decisions regarding NAAQS for Os and other air pollutants.
9 Research addressing energy expenditure and associated breathing rates in various population
10 groups, particularly healthy children and children with asthma, in various locations, across the
11 spectrum of physical activity, including sleep to vigorous exertion, is needed.
12 (10) An important consideration in the Os NAAQS review is the characterization of
13 background levels. There still remain substantial uncertainties in the characterization of 8-hr
14 daily max Os background concentrations. Further research to improve the evaluation of the
15 global and regional models which have been used to characterize estimates of background levels
16 would improve understanding of the role of non-U.S. anthropogenic emissions on Os levels over
17 the U.S.
18 4.8 SUMMARY OF PRELIMINARY STAFF CONCLUSIONS ON PRIMARY
19 STANDARD
20 In this section, we summarize our preliminary conclusions regarding the primary 63
21 standard. Staff conclusions in the final PA will be informed by our consideration of the available
22 scientific evidence as assessed in the ISA, air quality/exposure/risk information assessed in the
23 final HREA, recommendations received from CAS AC based on their review of this second draft
24 document, and comments received from members of the public.
25 As an initial matter, as discussed in section 3.4, staff concludes that reducing precursor
26 emissions to achieve Os concentrations that meet the current standard will provide important
27 improvements in public health protection. This initial conclusion is based on (1) the strong body
28 of scientific evidence indicating a wide range of adverse health outcomes attributable to
29 exposures to O?, concentrations found in the ambient air and (2) estimates indicating decreased
30 Oj, exposures and health risks upon meeting the current standard, compared to recent air quality.
31 Strong support for this conclusion is provided by the available health evidence and HREA
32 estimates of Os exposures of concern and Os-induced lung function risks. Some support for this
33 conclusion is also provided by HREA estimates of (^-associated mortality and morbidity.
34 In considering the available evidence and information, staff further concludes that the Os-
35 attributable health effects estimated to be allowed by air quality that meets the current primary
4-60
-------
1 standard for O^ can reasonably be judged important from a public health perspective. Thus, we
2 conclude that the available health evidence and exposure/risk information call into question the
3 adequacy of the public health protection provided by the current standard. This conclusion is
4 based on consideration of the scientific evidence assessed in the ISA, including controlled human
5 exposure and epidemiologic studies, as well as animal toxicology studies; the air quality,
6 exposure, and risk analyses presented in the second draft HREA for air quality that just meets the
7 current standard; and advice received from CAS AC in their review of the first draft PA and in
8 previous reviews.
9 In reaching the above conclusion regarding the current standard, we also reach
10 preliminary conclusions for the Administrator's consideration in making decisions on the
11 elements of a potential alternative primary O^ standard, as summarized below. We recognize that
12 selecting from among potential alternative standards will necessarily reflect consideration of
13 qualitative and quantitative uncertainties inherent in the relevant evidence and in the assumptions
14 of the quantitative exposure and risk assessments. Any such standard should protect public health
15 against health effects associated with exposure to 63, alone or in combination with related
16 photochemical oxidants, taking into account both evidence-based and exposure- and risk-based
17 considerations, and the nature and degree of uncertainties in such information. In reaching
18 conclusions about these ranges of potential alternative standards for consideration, we are
19 mindful that the Act requires primary standards that, in the judgment of the Administrator, are
20 requisite to protect public health with an adequate margin of safety. The primary standards are to
21 be neither more nor less stringent than necessary. Thus, the Act does not require that primary
22 NAAQS be set at zero-risk levels, but rather at levels that reduce risk sufficiently to protect
23 public health with an adequate margin of safety.
24 The degree of public health protection provided by the standard is due to the collective
25 impact of the elements of the standard, including the indicator, averaging time, level, and form.
26 Staffs preliminary conclusions on each of these elements are summarized below.
27 (1) It is appropriate to continue to use O^ as the indicator for a standard that is intended to
28 address effects associated with exposure to 63, alone or in combination with related
29 photochemical oxidants. Based on the available information staff preliminarily concludes
30 that there is no basis for considering any alternative indicator at this time. Meeting an 03
31 standard can be expected to provide some degree of protection against potential health
32 effects that may be independently associated with other photochemical oxidants, even
33 though such effects are not discernible from currently available studies indexed by 63
34 alone. Staff notes that control of ambient 63 levels is generally understood to provide the
35 best means of controlling photochemical oxidants in general, and thus of protecting
4-61
-------
1 against effects that may be associated with individual species and/or the broader mix of
2 photochemical oxidants, independent of effects specifically related to 63.
O
4 (2) It is appropriate to continue to use an 8-hour averaging time for the primary 63 standard.
5 (a) Staff preliminarily concludes that an 8-hour averaging time remains appropriate
6 for addressing health effects associated with short-term exposures to ambient 03.
7 An 8-hour averaging time is similar to the exposure periods evaluated in
8 controlled human exposure studies, including recent studies that provide evidence
9 for respiratory effects following exposures to 63 concentrations below the level of
10 the current standard. In addition, epidemiologic studies provide evidence for
11 health effect associations with 8-hour 03 concentrations, as well as with 1-hour
12 and 24-hour concentrations. A standard with an 8-hour averaging time (combined
13 with an appropriate standard form and level) would also be expected to provide
14 substantial protection against health effects attributable to 1- and 24-hour
15 exposures.
16
17 (b) Staff also preliminarily concludes that a standard with an 8-hour averaging time
18 can provide protection against respiratory effects associated with longer term 63
19 exposures. Analyses in the HREA show that as air quality is adjusted to just meet
20 the current 8-hour standard, most study areas are estimated to experience
21 reductions in respiratory mortality associated with long-term Os concentrations,
22 indicating that an 03 standard with an 8-hour averaging time can reduce
23 respiratory mortality reported to be associated with long-term 63 concentrations.
24 Moreover, the large majority of the U.S. population lives in locations where
25 reducing NOx emissions would be expected to result in modest decreases in warm
26 season averages of daily 8-hour ambient 63 concentrations. This suggests that
27 reductions in precursor emissions in order to meet a standard with an 8-hour
28 averaging time would also be expected to reduce the types of long-term 63
29 concentrations that have been associated with respiratory morbidity in
30 epidemiologic studies.
31
32 In addition, an analysis in the PA of whether just meeting the current or
33 alternative 63 standards, with 8-hour averaging times, would be expected to
34 maintain long-term 63 concentrations (i.e., seasonal average of 1-hour daily max)
35 below those present in most of the cities that provided the basis for a positive and
36 statistically significant association with respiratory mortality (Jerrett et al., 2009).
4-62
-------
1 This suggests that a standard with an 8-hour averaging time can maintain seasonal
2 averages of 1-hour daily max 63 concentrations below those that provided the
3 basis for the association with respiratory mortality (and below the concentration at
4 which study authors noted limited evidence of an effects threshold). Taken
5 together, these analyses suggest that a standard with an 8-hour averaging time,
6 coupled with the current 4* -highest form and an appropriate level, could provide
7 appropriate protection against the long-term 63 concentrations reported to be
8 associated with respiratory morbidity and mortality.
9
10 (3) It is appropriate to revise the level of the standard since the evidence and information
11 from the exposure and risk assessments in this review provide strong support for
12 consideration of an 63 standard with a level that would provide increased health
13 protection for at-risk groups, including people with asthma, especially children; the
14 lifestages of children and older adults; people with certain genetic variants; people with
15 reduced intake of certain nutrients; and outdoor workers against an array of adverse
16 health effects. These health effects range from decreased lung function, pulmonary
17 inflammation, and respiratory symptoms to serious indicators of respiratory morbidity
18 including ED visits and hospital admissions for respiratory causes, respiratory and all-
19 cause, non-accidental mortality. We also conclude that exposures of concern and health
20 risks projected to remain upon meeting the current standard, based on the exposure and
21 risk assessments, are indicative of risks to these populations and lifestages that can
22 reasonably be judged to be important from a public health perspective. This reinforces
23 our conclusion that consideration should be given to revising the level of the standard so
24 as to provide increased public health protection.
25
26 It is appropriate to consider a standard level within the range of 70 ppb to 60 ppb,
27 reflecting our judgment that a standard set within this range could provide an appropriate
28 degree of public health protection and would result in important improvements in
29 protecting the health of at-risk populations and lifestages. Standard levels within this
30 range that were considered in staff analyses of air quality, exposure, and risk include 70,
31 65 and 60 ppb, representative of levels within the upper, middle, and lower parts of this
32 range, respectively. Further, it would not be appropriate to consider increasing the level
33 of the current standard, thereby decreasing public health protection.
34
35 (4) It is appropriate to continue to use the 4* -highest daily max form of the standard. Staff
36 notes that the 4th-highest daily max was selected in 1997 in recognition of the public
4-63
-------
1 health protection provided by this form, when coupled with an appropriate averaging
2 time and level, combined with its stability for implementation programs. The currently
3 available evidence and exposure/risk information does not call into question these
4 conclusions from previous reviews. Therefore, we reach the preliminary conclusion that
5 it is appropriate to consider retaining the current 4* -highest daily max form for an Os
6 standard with an 8-hour averaging time and a revised level, as discussed above.
7 4.9 REFERENCES
8 Adams, W. C. (2006) Comparison of chamber 6.6 hour exposures to 0.04-0.08 ppm ozone via square-wave and
9 triangular profiles on pulmonary responses. Inhalation Toxicol. 18: 127-136.
10 http://dx.doi.org/10.1080/08958370500306107
11 Bell, ML; Peng, RD; Dominici, F. (2006). The exposure-response curve for ozone and risk of mortality and the
12 adequacy of current ozone regulations. Environ Health Perspect 114: 532-536.
13 Brown, JS; Bateson, TF; McDonnell, WF. (2008). Effects of exposure to 0.06 ppm ozone on FEV1 in humans: A
14 secondary analysis of existing data. Environ Health Perspect 116: 1023-1026.
15 http://dx.doi.org/10.1289/ehp.11396
16 Cakmak, S; Dales, RE; Judek, S. (2006). Respiratory health effects of air pollution gases: Modification by education
17 and income. Arch Environ Occup Health 61: 5-10. http://dx.doi.org/10.3200/AEOH.61.L5-10
18 Dales, RE; Cakmak, S; Doiron, MS. (2006). Gaseous air pollutants and hospitalization for respiratory disease in the
19 neonatal period. Environ Health Perspect 114: 1751-1754. http://dx.doi.org/10.1289/ehp.9044
20 Darrow, LA; Klein, M; Sarnat, JA; Mulholland, JA; Strickland, MJ; Sarnat, SE; Russell, AG; Tolbert, PE. (201 la).
21 The use of alternative pollutant metrics in time-series studies of ambient air pollution and respiratory
22 emergency department visits. JExpo Sci Environ Epidemiol 21: 10-19.
23 http://dx.doi.org/10.1038/jes.2009.49
24 Frey, C.; Samet, J.M. (2012). CASAC Review of the EPA's Policy Assessment for the Review of the Ozone National
25 Ambient Air Quality Standards (First External Review Draft - August 2012). EPA-CASAC-13-003.
26 November 26, 2012. Available online at:
27 Jerrett, M; Burnett, RT; Pope, CA, III; Ito, K; Thurston, G; Krewski, D; Shi, Y; Calle, E; Thun, M. (2009). Long-
28 term ozone exposure and mortality. N Engl J Med 360: 1085-1095.
29 http://dx.doi.org/10.1056/NEJMoa0803894
30 Katsouyanni, K; Samet, JM; Anderson, HR; Atkinson, R; Le Tertre, A; Medina, S; Samoli, E; Touloumi, G;
31 Burnett, RT; Krewski, D; Ramsay, T; Dominici, F; Peng, RD; Schwartz, J; Zanobetti, A. (2009). Air
32 pollution and health: A European and North American approach (APHENA). (Research Report 142).
33 Boston, MA: Health Effects Institute. http://pubs.healtheffects.org/view.php?id=327
34 Kim, CS; Alexis, NE; Rappold, AG; Kehrl, H; Hazucha, MJ; Lay, JC; Schmitt, MT; Case, M; Devlin, RB; Peden,
35 DB; Diaz-Sanchez, D. (2011). Lung function and inflammatory responses in healthy young adults exposed
36 to 0.06 ppm ozone for 6.6 hours. Am J Respir Crit Care Med 183: 1215-1221.
37 http://dx.doi.org/10.1164/rccm.201011-1813OC
3 8 Mar, TF; Koenig, JQ. (2009). Relationship between visits to emergency departments for asthma and ozone exposure
39 in greater Seattle, Washington. Ann Allergy Asthma Immunol 103: 474-479.
4-64
-------
1 McDonnell, WF; Stewart, PW; Smith, MV. (2010). Prediction of ozone-induced lung function responses in humans.
2 Inhal Toxicol 22: 160-168. http://dx.doi.org/10.3109/08958370903089557
3 Samet, J.M. (2010) Review of EPA's Proposed Ozone National Ambient Air Quality Standard (Federal Register,
4 Vol. 75, Nov. 11, January 19, 2010). EPA-CASAC-10-007. February 19, 2010. Available online at:
5 Samet, J.M. (2011) Clean Air Scientific Advisory Committee (CASAC) Response to Charge Questions on the
6 Reconsideration of the 2008 Ozone National Ambient Air Quality Standards. EPA-CASAC-11-004.
7 March 30, 2011. Available online at:
8 http://yosemite.epa.gov/sab/sabproduct.nsf/0/F08BEB48C1139E2A8525785E006909AC/$File/EPA-
9 CASAC-ll-004-unsigned+.pdf
10 Schelegle, ES; Morales, CA; Walby, WF; Marion, S; Allen, RP. (2009). 6.6-hour inhalation of ozone concentrations
11 from 60 to 87 parts per billion in healthy humans. Am J Respir Crit Care Med 180: 265-272.
12 http://dx.doi.org/10.1164/rccm.200809-1484OC
13 Smith, RL; Xu, B; Switzer, P. (2009b). Reassessing the relationship between ozone and short-term mortality in U.S.
14 urban communities. Inhal Toxicol 21: 37-61. http://dx.doi.org/10.1080/08958370903161612
15 Stieb, DM; Szyszkowicz, M; Rowe, BH; Leech, JA. (2009). Air pollution and emergency department visits for
16 cardiac and respiratory conditions: A multi-city time-series analysis. Environ Health Global Access Sci
17 Source 8: 25. http://dx.doi.org/10.1186/1476-069X-8-25
18 U.S. Environmental Protection Agency. (2012). Health Risk and Exposure Assessment for Ozone, First External
19 Review Draft, U.S. Environmental Protection Agency, Research Triangle Park, NC. EPA 452/P-12-001.
20 U.S. Environmental Protection Agency. (2013). Integrated Science Assessment for Ozone and Related
21 Photochemical Oxidants (Final Report). U.S. Environmental Protection Agency, Washington, DC,
22 EPA/600/R-10/076F.
23 U.S. Environmental Protection Agency. (2014). Health Risk and Exposure Assessment for Ozone, Second External
24 Review Draft. Office of Air Quality Planning and Standards, Research Triangle Park, NC. EPA-452/P-14-
25 004a.
4-65
-------
1 5 ADEQUACY OF THE CURRENT SECONDARY STANDARD
2 This chapter presents staffs considerations and preliminary conclusions regarding the
3 adequacy of the current secondary Os NAAQS. In doing so, we pose the following overarching
4 question:
5 • Does the currently available scientific evidence- and exposure/risk-based
6 information, as reflected in the ISA and WREA, support or call into question the
7 adequacy and/or appropriateness of the protection afforded by the current
8 secondary Os standard?
9 In addressing this overarching question, we pose a series of more specific questions, as
10 discussed in sections 5.1 through 5.5 below. We consider the nature of Os-induced effects,
11 including the nature of the exposures that drive the biological and ecological response and
12 related biologically-relevant exposure metrics (section 5.1); the scientific evidence and
13 exposure/risk information, including that of associated ecosystem services, regarding (a) tree
14 growth, productivity and carbon storage (section 5.2), (b) crop yield loss (section 5.3), (c) visible
15 foliar injury (section 5.4), and (d) other welfare effects (section 5.5). Section 5.6 describes
16 advice and recommendations received from CASAC. In section 5.7, we revisit the overarching
17 question of this chapter and present preliminary staff conclusions on the adequacy of the current
18 secondary standard.
19 5.1 NATURE OF EFFECTS AND BIOLOGICALLY-RELEVANT EXPOSURE
20 METRIC
21 • Does the current evidence alter our conclusions from the previous review
22 regarding the nature of Os-induced welfare effects?
23 The current body of Os welfare effects evidence confirms the conclusions reached in the
24 last review on the nature of Os-induced welfare effects and is summarized in the ISA as follows
25 (U.S. EPA, 2013, p. 1-8):
26 The welfare effects ofOs can be observed across spatial scales, starting at the
27 subcellular and cellular level, then the whole plant and finally, ecosystem-level
28 processes. Ozone effects at small spatial scales, such as the leaf of an individual
29 plant, can result in effects along a continuum of larger spatial scales. These
30 effects include altered rates of leaf gas exchange, growth, and reproduction at the
31 individual plant level, and can result in broad changes in ecosystems, such as
32 productivity, carbon storage, water cycling, nutrient cycling, and community
33 composition.
5-1
-------
1 This body of evidence has been expanding since the phytotoxic effects of Os were first
2 identified by Richards, et al in 1958 who showed that "63 was a constituent of smog that caused
3 foliar injury on grapes in California" (Flagler, 1998). In the half century that has followed, a
4 large number of studies have been conducted both in and outside of the U.S. to examine the
5 impacts of 63 on plants and their associated ecosystems. Taken together, these studies
6 demonstrate that Os -induced effects that occur at the subcellular and cellular levels, at sufficient
7 magnitudes propagate up to produce larger scale effects that affect the whole organism. In
8 addition, the studies assessed in this review have further increased our understanding of the
9 molecular, biochemical and physiological mechanisms that explain how plants are affected by
10 63, in the absence of other stressors, particularly in the area of genomics (U.S. EPA, 2013,
11 Chapter 9, section 9.3). Based on its assessment of this extensive body of science, the ISA
12 determined that a causal relationship exists between exposure to 03 in ambient air and visible
13 foliar injury effects on vegetation, reduced vegetation growth, reduced productivity in terrestrial
14 ecosystems, reduced yield and quality of agricultural crops and alteration of below-ground
15 biogeochemical cycles (U.S. EPA 2013, Table 1-2). Additionally, the ISA determined that a
16 likely to be causal relationship exists between exposures to 63 in ambient air and reduced carbon
17 sequestration in terrestrial ecosystems, alteration of terrestrial ecosystem water cycling and
18 alteration of terrestrial community composition (U.S. EPA, 2013, Table 1-2).
19 From this set of effects that the ISA has concluded to be causally or likely causally
20 related to 03 in ambient air, we focus primarily on three categories of effects (i.e., visible foliar
21 injury; impacts on tree growth, productivity and carbon storage; and crop yield loss). As
22 recognized in the ISA, controlled exposure studies, which are the best method for isolating or
23 characterizing the role of 63 in inducing the observed plant effects, "have clearly shown that
24 exposure to 63 is causally linked to visible foliar injury, decreased photosynthesis, changes in
25 reproduction, and decreased growth" in vegetative species (U.S. EPA 2013, p. 1-15). These plant
26 level effects are also linked to a cascade of other ecosystem level effects. For example, studies at
27 larger spatial scales support the controlled exposure study results and indicate that "ambient Os
28 exposures can effect ecosystem productivity, crop yield, water cycling, and ecosystem
29 community composition" (U.S. EPA 2013, p. 1-15). Thus, 63 effects on vegetation may have
30 implications at the individual, species, population and whole ecosystem level, including
31 associated ecosystem services.l
1 Ecosystem services are the benefits that people obtain from ecosystems and have been stated to include
"provisioning services such as food and water; regulating services such as regulation of floods, drought, land
degradation, and disease; supporting services such as soil formation and nutrient cycling; and cultural services such
as recreational, spiritual, religious, and other nonmaterial benefits" according to the Millennium Ecosystem
Assessment (UNEP, 2003; U.S. EPA, 2013, p. Ixxii).
5-2
-------
1 Our consideration of O^ welfare effects and the significance or weight to place on a given
2 study or exposure/risk assessment result, is informed by the understanding, based on the entire
3 body of vegetation effects science, that a variety of other factors can either mitigate or exacerbate
4 the Os-plant interactions and are recognized sources of uncertainty and variability. These
5 include: 1) multiple genetically-influenced determinants of 63 sensitivity; 2) changing sensitivity
6 across growth stages; 3) co-occurring stressors and/or modifying environmental factors; 4) a
7 paucity of information on most of the 43,000 U.S. plant species (U.S. EPA 2013, section 9.4.8;
8 U.S. EPA 2006; U.S. EPA, 2007, section 7.4.2).
9 • Does the current evidence continue to support a cumulative, seasonal exposure
10 index as a biological-relevant and appropriate metric for assessment of the
11 evidence and exposure/risk information?
12 In this review, the ISA assessment of the full body of currently available evidence stated
13 the following regarding biological indices (U.S. EPA, 2013, p. 2-44):
14 The main conclusions from the 1996 and 2006 0j AQCDs regarding indices
15 based on ambient exposure remain valid. These key conclusions can be restated
16 as follows:
17 • ozone effects in plants are cumulative;
18 • higher O3 concentrations appear to be more important than lower
19 concentrations in eliciting a response;
20 • plant sensitivity to Os varies with time of day and plant development
21 stage;
22 • quantifying exposure with indices that cumulate hourly Os concentrations
23 and preferentially weight the higher concentrations improves the
24 explanatory power of exposure/response models for growth and yield,
25 over using indices based on mean and peak exposure values.
26 Thus, the current evidence, as in other recent reviews, continues to support a cumulative,
27 seasonal exposure index as a biologically-relevant and appropriate metric for assessment of the
28 evidence and exposure/risk information. To this point, the available body of evidence provides a
29 wealth of information on the aspects of 63 exposure that are most important in influencing plant
30 response. While a variety of "factors with known or suspected bearing on the exposure-response
31 relationship, including concentration, time of day, respite time, frequency of peak occurrence,
32 plant phenology, predisposition, etc.," have been identified (U.S. EPA, 2013, section 9.5.2), the
33 importance of the duration of the exposure and the relatively greater importance of higher
34 concentrations over lower in determining plant response to 63 have been well documented (U.S.
35 EPA, 2013, section 9.5.3). Much of this work was completed by the mid-1990s, and was
5-3
-------
1 summarized in the 1996 Criteria Document (U.S. EPA, 1996), while the additional newer work
2 is assessed in the subsequent 2006 Criteria Document and 2013 ISA.
3 In conjunction with this research on plant response to Os exposures, others have
4 developed "mathematical approaches for summarizing ambient air quality information in
5 biologically meaningful forms for 63 vegetation effects assessment purposes ..." (U.S. EPA,
6 2013, section 9.5.3). A large set of exposure indices have been developed that use a variety of
7 functions to weight factors that have been shown to influence vegetation exposure-response
8 relationships (U.S. EPA, 2013, section 9.5.2). As discussed in the ISA, several indices have
9 been developed to attempt to incorporate some of the biological, environmental, and exposure
10 factors that influence the magnitude of the biological response and contribute to observed
11 variability (U.S. EPA, 2013, section 9.5.2). As with any summary statistic, these exposure
12 indices retain information on some, but not all, characteristics of the original observations.
13 Flux models have been identified in recent years to take into account more of the factors
14 that influence the response and contribute to observed variability (U.S. EPA, 2013, section
15 9.5.4). We note that "some researchers have claimed that using flux models can be used to better
16 predict vegetation responses to 63 than exposure-based approaches..." because flux models
17 estimate the ambient 03 concentration that actually enters the leaf (i.e., flux or deposition) (U.S.
18 EPA, 2013, p. 9-114). However, "[f]lux calculations are data intensive and must be carefully
19 implemented" (U.S. EPA, 2013, p. 9-114). Further, "[t]his uptake-based approach to quantify
20 the vegetation impact of Os requires inclusion of those factors that control the diurnal and
21 seasonal 63 flux to vegetation (e.g., climate patterns, species and/or vegetation-type factors and
22 site-specific factors)" (U.S. EPA, 2013, p. 9-114). Each species has different amounts of internal
23 detoxification potential that may protect species to differing degrees. This balance between 63
24 flux and detoxification processes has been termed the "effective flux". Accordingly, the
25 "models have to distinguish between stomatal and non-stomatal components of 03 deposition to
26 adequately estimate actual concentration reaching the target tissue of a plant to elicit a response"
27 and "[determining this Os uptake via canopy and stomatal conductance relies on models to
28 predict flux and ultimately the 'effective' flux" (U.S. EPA, 2013, pp. 9-114). The lack of
29 detailed species- and site-specific data required for flux modeling in the U.S. and the lack of
30 understanding of detoxification processes have made this technique less viable for use in
31 vulnerability and risk assessments at the national scale in the U.S. (U.S. EPA, 2013, section
32 9.5.4).
33 Thus, in the last two reviews of the O^ secondary standard, completed in 1997 and 2008,
34 the EPA concluded that the risk to vegetation comes primarily from cumulative exposures to 63
35 over a season or seasons and, in both reviews, the EPA proposed, as one alternative, a secondary
36 standard set in terms of such a form (61 FR 65716, 72 FR 37818). Although in both reviews the
5-4
-------
1 secondary standard was revised to be identical to a revised primary standard (with an 8-hour
2 averaging time), the Administrator, in both cases, also concluded, consistent with CASAC
3 advice, that a cumulative, seasonal index was the most biologically relevant way to relate
4 exposure to plant growth response (62 FR 38856, 73 FR 16436). Most recently, in the 2010
5 proposed reconsideration of the 2008 decision, the EPA again proposed to conclude that 63
6 exposure indices that cumulate differentially weighted hourly concentrations are the best
7 candidates for relating exposure to plant growth responses and proposed to set the secondary
8 standard only in terms of one such form, the W126 (75 FR 2938).
9 Based on the long-established conclusions and long-standing supporting evidence
10 described above, we continue to focus on the aspects of ambient 63 exposures that have
11 biological relevance and the biologically-relevant exposure indices or metrics that have been
12 designed in light of this consideration, i.e., cumulative seasonal indices. Since the review
13 completed in 1997, which was the first to focus on cumulative indices, attention has been given
14 primarily to two different cumulative index forms: SUM06 and W126. The SUM06 index is a
15 threshold-based approach described as the sum of all hourly 63 concentrations greater or equal to
16 0.06 ppm observed during a specified daily and seasonal time window (U.S. EPA, 2013, section
17 9.5.2). The W126 index is a non-threshold approach described as the sigmoidally weighted sum
18 of all hourly 63 concentrations observed during a specified daily and seasonal time window,
19 where each hourly Os concentration is given a weight that increases from 0 to 1 with increasing
20 concentration (Lefohn et al, 1988; Lefohn and Runeckles, 1987; U.S. EPA, 2013, section 9.5.2).
21 The EPA used the W126 index to consider welfare effects in the last review, as well as the 2010
22 proposed reconsideration of the 2008 decision; this approach received support from CASAC, as
23 discussed in section 5.6 below. Consistent with the ISA conclusions regarding the
24 appropriateness of considering cumulative exposure indices for 63 effects of concern based on
25 the evidence available in this review, we again conclude that the current evidence continues to
26 support a cumulative, seasonal exposure index as a biologically-relevant and appropriate metric
27 for assessment of the evidence and exposure/risk information, and in particular, the W126
28 cumulative, seasonal metric (U.S. EPA, 2013, p. 2-44 and section 9.5.2).
29 • Within what paradigm may it be appropriate to consider the potential adversity
30 of public welfare effects of Os?
31 The Clean Air Act requires that a secondary standard be protective against those known
32 or anticipated 63 effects that are "adverse" to the public welfare, not all identifiable (Vinduced
33 effects. Unlike the use of the terms adverse, injury or damage in the scientific literature, in the
34 NAAQS policy context, these terms have been interpreted in a particular way. Specifically, 03-
35 induced "injury" to vegetation has been defined as encompassing all plant reactions, including
5-5
-------
1 reversible changes or changes in plant metabolism (e.g., altered photosynthetic rate), altered
2 plant quality or reduced growth that does not impair the intended use or value of the plant. In
3 contrast, "damage" has been defined to include only those injury effects that reach sufficient
4 magnitude as to also reduce or impair the intended use or value of the plant to the public and thus
5 potentially become adverse to the public welfare. Examples of vegetation effects that have been
6 classified as damage include reductions in aesthetic values (e.g., foliar injury in ornamental
7 species) as well as losses in terms of weight, number, or size of harvestable plant parts. Biomass
8 loss in tree species can also be considered damage or adverse to the public welfare if it includes
9 slower growth in species harvested for timber or other fiber uses. In the context of evaluating
10 effects on single plants or species grown in monocultures such as managed forests, this construct
11 continues to remain useful (73 FR 16492/96).
12 However, given the increasing scientific literature linking Os effects on plants or species
13 to effects at the community or ecosystem level, a more expansive construct or paradigm of what
14 constitutes Os "damage" beyond that of the individual or species level is appropriate. A number
15 of broader paradigms have been discussed in the literature (72 FR 37890; Hogsett et al., 1997;
16 Young and Sanzone, 2002). In the 2008 review, the Administrator expressed support for relying
17 on a definition of "adverse" discussed in section IV.A.3 of the proposal (62 FR 37889-37890)
18 that embeds "the concept of 'intended use' of the ecological receptors and resources that are
19 affected, and applies that concept beyond the species level to the ecosystem level" (73 FR
20 16496). For example, in the 2008 rulemaking notice, the Administrator took note of "a number
21 of actions taken by Congress to establish public lands that are set aside for specific uses that are
22 intended to provide benefits to the public welfare, including lands that are to be protected so as to
23 conserve the scenic value and the natural vegetation and wildlife within such areas, and to leave
24 them unimpaired for the enjoyment of future generations" (73 FR 16496).
25 Since the 2008 O?, review, our approach to assessing adversity to the public welfare in
26 NAAQS reviews has continued to evolve. In particular, we consider the concept of ecosystem
27 services in a broader paradigm. An extensive look at the range of services than can be provided
28 by ecosystems is described in the Millennium Ecosystem Assessment (MEA, 2005). Ecosystem
29 services can be generally defined as the benefits that individuals and organizations obtain from
30 ecosystems. The EPA has previously defined ecological goods and services for the purposes of
31 benefits assessment as the "outputs of ecological functions or processes that directly or
32 indirectly contribute to social welfare or have the potential to do so in the future. Some outputs
33 may be bought and sold, but most are not marketed" and has relied on this definition in
34 regulatory impact analyses for previous NAAQS reviews (U.S. EPA, 2006b). In the review of
35 the secondary NAAQS for oxides of nitrogen and sulfur, EPA recognized that changes in
36 ecosystem services may be used to aid in characterizing a known or anticipated adverse effect to
5-6
-------
1 public welfare and that an evaluation of adversity to the public welfare might consider the
2 likelihood, type, magnitude, and spatial scale of the effect, as well as the potential for recovery
3 and any uncertainties relating to these conditions (77 FR 20232). In the context of this review,
4 ecosystem services are being evaluated and assessed in the REA as one way to characterize the
5 possible public welfare benefits received from ecosystem resources and how those services
6 might be expected to change under air quality scenarios representing the current and potential
7 alternative secondary standards (U.S. EPA, 2014, chapter 5). Thus, in considering the evidence
8 and exposure risk information associated with the welfare endpoints identified below, in the
9 context of consideration of adequacy of the current standard in this chapter and potential
10 alternative standards in chapter 6, we consider also how they fit within this paradigm.
11 5.2 FOREST TREE GROWTH, PRODUCTIVITY AND CARBON STORAGE
12 This section considers the current evidence and exposure/risk information to inform
13 consideration of the adequacy of the protection provided by the current standard from known and
14 anticipated adverse welfare effects of Os related to growth, productivity, and carbon storage of
15 forest trees, and other associated effects. Trees are important from a public welfare perspective
16 because they provide many valued services to humans. In addition to the aesthetic value
17 discussed below in 5.4, these include: food, fiber, timber, other forest products, habitat,
18 recreational opportunities, climate regulation, erosion control, air pollution removal, hydrologic
19 and fire regime stabilization (U.S. EPA, 2014, section 6.1, Figure 6-1, section 6.4, Table 6-12).
20 This section includes a discussion of the policy-relevant evidence and weight-of-evidence
21 conclusions discussed in the ISA (section 5.2.1) and the exposure/risk results, including
22 associated ecosystem services (section 5.2.2) described in the second draft WREA. Important
23 uncertainties and limitations in the available information are discussed throughout the sections.
24 These discussions highlight the information we consider relevant to answering the overarching
25 question and associated policy-relevant questions included in this section.
26 5.2.1 Evidence-based Considerations
27 • To what extent has scientific information become available that alters or
28 substantiates our prior conclusions of Os-related effects on forest tree growth,
29 productivity and carbon storage and of factors that influence associations between
30 Os concentrations and these effects?
31 Research published since the 2006 AQCD substantiates prior conclusions regarding 63-
32 related effects on forest tree growth, productivity and carbon storage. Information supporting
33 these previous conclusions comes from a variety of different types of studies and which cover an
34 array of different species, endpoints and exposure methods. One subset of studies focused on
5-7
-------
1 underlying mechanisms as they relate to growth, productivity and carbon storage including:
2 reduced carbon dioxide uptake due to stomatal closure (U.S. EPA 2013, section 9.3.2.1); the
3 upregulation of genes associated with plant defense, signaling, hormone synthesis and secondary
4 metabolism (U.S. EPA 2013, section 9.3.3.2); the down regulation of genes related to
5 photosynthesis and general metabolism (U.S. EPA 2013, section 9.3.3.2); loss of carbon
6 assimilation capacity due to declines in the quantity and activity of key proteins and enzymes
7 (U.S. EPA, 2013, section 9.3.5.1); and negative impacts on the efficiency of the photosynthetic
8 light reactions (U.S. EPA, 2013, section 9.3.5.1). These new studies "have increased knowledge
9 of the molecular, biochemical and cellular mechanisms occurring in plants in response to (V,
10 adding "to the understanding of the basic biology of how plants are affected by oxidative
11 stress..." (U.S. EPA, 2013, p. 9-11).
12 The recent studies cover a variety of exposure methods, species, and settings. In
13 particular, a recent meta-analysis indicates a relationship between 63 concentrations in the
14 northern hemisphere and effects with the potential to affect growth (i.e., stomatal conductance
15 and photosynthesis) (U.S. EPA, 2013, section 9.4.3.1; Wittig et al., 2007). A second meta-
16 analysis, which quantitatively compiled peer-reviewed studies from the past 40 years, found that
17 ambient concentrations reported in those studies significantly decreased annual total biomass
18 growth (7%) across the studies (U.S. EPA, 2013, section 9.4.3.1). The ISA states that these two
19 meta-analyses demonstrate the coherence of Os effects on plant photosynthesis and growth
20 across numerous studies and species using a variety of experimental techniques", and including a
21 recent study, that "recent meta-analyses have generally indicated that 63 reduced carbon
22 allocation to roots (Wittig et al., 2009; Grantz et al., 2006)" (U.S. EPA, 2013, pp. 9-45 to 9-46).
23 Recent field-based studies also have added to the evidence base. For example, a study
24 conducted in forest stands in the southern Appalachian Mountains found that "the cumulative
25 effects of ambient levels of Os decreased seasonal stem growth by 30-50% for most tree species
26 in a high Oj, year in comparison to a low 63 year (McLaughlin et al., 2007a). The authors also
27 reported that high ambient Os concentrations can increase whole-tree water use and in turn
28 reduce late-season streamflow (McLaughlin et al., 2007b)" (U.S. EPA, 2013, p. 9-43).
29 Additionally, a recent study has provided concentration-response information for tree seedlings
30 of an additional species beyond the 11 previously studied (Gregg, et al., 2003). This study on
31 cottonwood expands the dataset of studied species such that the 12 species for which C-R
32 functions are available in this review include deciduous, coniferous, eastern, western, sensitive
33 and tolerant species (U.S. EPA 2013, section 9.6.2; U.S. EPA, 2014, section 6.2, Figure 6-2,
34 Table 6-1).
35 The "previous Os AQCDs concluded that there is strong evidence that exposures to Os
36 decreases photosynthesis and growth in numerous plant species" and that "[s]tudies published
5-8
-------
1 since the 2008 review support those conclusions" (U.S. EPA, 2013, p. 9-42). The previously
2 available strong evidence included the development of robust concentration-response (C-R)
3 functions for tree seedling biomass loss in 11 species under the National Health and
4 Environmental Effects Research Laboratory-Western Ecology Division (NHEERL-WED)
5 program. This series of experiments used open-top-chambers (OTC) to study seedling growth
6 response under a variety of Os exposures (ranging from near background to well above current
7 ambient concentrations) and growing conditions (U.S. EPA 2013, section 9.6.2, Lee and Hogsett,
8 1996).
9 We additionally recognize that, because trees are long lived, in addition to the effects of
10 Oj, exposures over the annual growing season, trees and other perennials can also cumulate
11 effects across multiple years. For example, growth affected by a reduction in carbohydrate
12 storage in one year may result in the limitation of growth in the following year, so that effects
13 "carry over" from one year to another (U.S. EPA, 2013, section 9.4.8; Andersen, et al., 1997). In
14 past reviews, such carry-over effects have been documented in the growth of some tree seedlings
15 and in roots. A number of recent studies based on the FACE exposure method in a planted forest
16 at the Aspen FACE site in Wisconsin have augmented and supported the earlier information.
17 These studies observed tree growth responses over seven years beyond the seedling growth stage
18 and growing in field settings more similar to natural forest stands than OTC studies. In addition
19 to affecting tree heights, diameters, and main stem volumes in the aspen community, elevated Os
20 was reported to change intra- and inter-species competition (Kubiske et al., 2006; Kubiske et al.,
21 2007). For example, Oi treatments increased the rate of conversion from a mixed aspen-birch
22 community to a birch dominated community, potentially changing intra- and inter-species
23 competition.
24 The EPA comparison of biomass results from the first seven years of the recent study by
25 Kubiske et al (2007) to that predicted using the C-R function established from the earlier OTC
26 experiments indicated close agreement (U.S. EPA 2013, Section 9.6.3). Accordingly, the ISA
27 concludes that "[o]verall, the studies at the Aspen FACE experiment were consistent with many
28 of the open-top chamber (OTC) studies that were the foundation of previous Oj NAAQS
29 reviews" and that "[tjhese {recent} results strengthen the understanding of Os effects on forests
30 and demonstrate the relevance of the knowledge gained from trees grown in OTC studies" (U.S.
31 EPA 2013, p. 2-38). In addition to growth effects, these scientists also found that elevated Os
32 decreased birch seed weight, germination, and bud starch levels as well as aspen bud size. The
33 effects on birch seeds could lead to a negative impact on species regeneration, while the bud
34 effects may have been related to the observed delay in spring leaf development and have the
35 potential to alter carbon metabolism of overwintering buds. These latter effects likely have
36 implications for the subsequent growing season (i.e. carry-over effects) in the following year,
5-9
-------
1 including effects on forest biomass, buds and seeds that carry into subsequent years (U.S. EPA,
2 2013, section 9.4).
3 The recent studies, in combination with the entire body of evidence, form the basis for
4 the ISA determinations that: 1) there is a causal relationship between 63 exposures and reduced
5 vegetative growth; 2) there is a causal relationship between O?, exposures and biomass
6 accumulation, including altered carbon allocation to below ground tissues, rates of leaf and root
7 production, and turnover and decomposition that can alter below-ground biogeochemical cycles;
8 and 3) there is likely to be a causal relationship between O?, exposure and a reduction in carbon
9 sequestration in terrestrial ecosystems (U.S. EPA, 2013, Table 2-2). Therefore, the current
10 evidence base substantiates prior conclusions of Cb-related effects on tree growth, productivity
11 and carbon storage. The results from some of these studies are discussed more fully under
12 different questions below.
13 • To what extent have important uncertainties in the evidence identified in the last
14 review been reduced and/or new uncertainties emerged?
15 As stated above, the ISA concludes that the new evidence confirms, strengthens or
16 expands our understanding of Os effects on plants. Much of this new evidence is focused on the
17 molecular and genetic level, providing very important new mechanistic information that in some
18 cases enhances our understanding of the complexity of the Os-plant response. This information
19 has, in general, reduced overall uncertainties at the subcellular and cellular scales. However,
20 because these studies were primarily conducted using artificial exposure conditions and model
21 plants, uncertainties remain regarding the extent to which these plant responses reflect those in
22 other plant species and exposure conditions (U.S. EPA, 2013, section 9.3.6). With regard to O^
23 impacts at the whole plant, species, and ecosystem scales, recent information has informed our
24 understanding of associated uncertainties in a variety of ways.
25 Importantly, one key uncertainty has been significantly reduced. This relates to the C-R
26 functions we have used in previous reviews to estimate tree seedling biomass loss for different
27 Os exposure conditions (U.S. EPA, section 9.6). As these functions were derived from OTC
28 experiments, an associated uncertainty has been with regard to the extent to which they reflected
29 concentration-response relationships for field conditions. In the current review, EPA staff have
30 conducted an analysis comparing OTC data with field-based data for one crop and one tree
31 species (U.S. EPA, 2013, section 9.6.3.2). One comparison was done using soybean OTC data
32 from the National Crop Loss Assessment Network (NCLAN)2 and field-based data (Soy FACE),
2 The NCLAN program was conducted from 1980 to 1987 at five different locations across the US. At
each site, open top chambers were used to expose plants to Os treatments that represented the range of
5-10
-------
1 as discussed in section 5.3 below. The second was done using aspen seedling OTC data from the
2 NHEERL-WED and field-based data (Aspen FACE). The result of the aspen analysis, similar to
3 that for soybean, showed very close agreement between the predictions based on NHEERL-
4 WED data and Aspen FACE observations, even when comparing the results of experiments that
5 used different exposure methodologies, different genotypes, locations, and durations. Based on
6 this analysis, the ISA additionally stated that "the [C-R] function based on one year of growth
7 was applicable to subsequent years" (of the six-year dataset) (U.S. EPA, 2013, section 9.6.3.2).
8 This information reduces the uncertainties associated with potential impacts of other
9 experimental factors on the Os-plant response. Other studies, such as the meta-analyses
10 discussed in the ISA and below, also demonstrate the coherence of 63 growth effects across
11 numerous studies and species that used a variety of experimental techniques. While recognizing
12 that uncertainties may remain for some individual species for which the database is relatively
13 less robust (such as the more recent information on cottonwood), taken together, this information
14 substantially reduces uncertainties associated with use of the tree seedling OTC-derived C-R
15 functions to predict the response of trees beyond the seedling stage in field settings. Thus, in the
16 current review, the ISA and WREA have continued to use these functions to estimate tree growth
17 response for different cumulative O3 exposures (U.S. EPA, 2013, section 9.6.2; U.S. EPA, 2014,
18 section 6.2).
19 Other uncertainties associated with studying or modeling Os impacts on trees, including
20 those identified in the last review, still remain, due in part to a lack of additional research but
21 also due to the growth characteristics of trees which present a unique set of experimental
22 challenges. For example, while trees are long-lived, with life spans which range from decades to
23 centuries, most studies are designed to take place within an annual or 2-3 year timeframe, which
24 represents only a small fraction of the lifetime of a tree. Further, trees grow very large, making it
25 difficult to use controlled exposure environments beyond the seedling or sapling growth stage.
26 Additionally, sensitivity to 63 varies over the life of the tree so that different growth stages of the
27 tree may be more or less sensitive and this variation in growth-stage sensitivity is species-
28 specific. Thus, while some limited information exists regarding tree sensitivity beyond the
29 seedling growth stage (e.g. aspen, cottonwood) and in some species for both seedling and mature
30 trees within a species (e.g., red oak), it remains uncertain to what degree effects observed on
31 trees during one growth stage (e.g., seedling) can be extrapolated to trees at other growth stages.
32 An analysis in the WREA comparing seedling to adult tree biomass loss, discussed in 5.2.2
concentrations that occur in different areas of the world. The NCLAN focused on the most important U.S
agricultural crops (Heagle et al, 1989; http://www.ars.usda.gov/Main/docs.htm?docid=12462).
5-11
-------
1 below, informs our consideration of this remaining uncertainty (U.S. EPA, 2014, section
2 6.2.1.1).
3 Lastly, we recognize that exposures in one year have the potential to cause effects not
4 observed until a subsequent year (termed "carry over" effects). While recent studies provide
5 evidence of some carry-over effects and the potentially serious implications they could have for
6 associated ecosystems and services, the true extent of this effect is unclear because most studies
7 on the effects of 63 on growth do not measure or take into account the possibility of carry-over
8 effects in subsequent years. Additionally, uncertainties remain regarding the extent of
9 compounding of growth effects across multiple years for different air quality conditions. These
10 uncertainties affect our characterization of such impacts, particularly the quantitative aspects, for
11 differing exposure scenarios. Therefore our characterization of subsequent growing season
12 effects is uncertain, affecting our ability to fully describe impacts of observed annual biomass
13 losses, particularly in quantitative terms. In section 5.2.2 below, the potential variation of
14 growth effect compounding across multiple years is discussed further drawing on the WREA
15 evaluation of this (U.S. EPA, 2014, section 6.2.1.4). Further discussion of these tree-related
16 uncertainties that are relevant in the context of informing our understanding of the quantitative
17 and qualitative exposure and risk results are discussed in the appropriate sections below.
18 • What are the ecosystem services potentially affected by Os effects on tree growth,
19 productivity and carbon storage and to what extent are they important from a
20 public welfare perspective?
21 A variety of ecosystem services are potentially affected by 63 impacts on tree growth,
22 productivity and carbon storage. The ISA identifies as causal the relationship of 03 and reduced
23 productivity in terrestrial ecosystems and alteration of below ground biogeochemical cycles. It
24 further identifies as likely to be a causal relationship (^impacts on reduced carbon sequestration
25 in terrestrial ecosystems; alteration of terrestrial ecosystem water cycling; alteration of terrestrial
26 community composition (U.S. EPA, 2013, Table 9-19). These effects are important to the
27 public welfare, and in particular include those associated with national parks, national refuges
28 and other protected areas ranging to the harvesting of timber for commercial uses. The
29 ecosystem services most directly affected by biomass loss include: (1) habitat provision for
30 wildlife, (2) carbon storage, (3) provision of food and fiber, and (4) pollution removal (see also
31 U.S. EPA, 2014, Figure 6-1).3 Less direct impacts can occur on process-related effects such as
32 nutrient and hydrologic cycles.
3 The impacts of O3 in ambient air on some of these services are characterized in the WREA, as described
in section 5.2.2 below.
5-12
-------
1 With regard to the extent of Os impact on carbon transport or allocation to roots,
2 information from a few recent individual studies have shown mixed results, including negative,
3 non-significant, and positive effects on root biomass and root: shoot ratio. However, as assessed
4 in the ISA, "[gjenerally, there is clear evidence that Os reduces carbon transport or allocation to
5 roots", although (U.S. EPA, 2013, p. 9-44).
6 • To what extent does the available evidence indicate the occurrence of Os-related
7 effects on forest growth, productivity and carbon storage attributable to
8 cumulative exposures at lower ambient Os concentrations than previously
9 established or to exposures that might be expected to occur under the current
10 standard?
11 The evidence available in this review indicates that Cb-induced effects on tree growth,
12 productivity and carbon storage occur as a result of cumulative exposure concentrations similar
13 to those identified in the previous review and that these effects can occur at exposures associated
14 with air quality conditions that might be expected to meet the current standard. We first consider
15 the evidence on Oj, effects on growth, particularly for tree seedlings, and that supports the C-R
16 functions describing the relationship between cumulative 63 exposure and reduction in growth
17 (relative biomass loss). As described above, this evidence base currently includes functions for
18 12 species.4 Figure 5-1, below (based on species-specific composite functions described in the
19 ISA and WREA), illustrates the appreciable variation in sensitivity across individual species
20 (U.S. EPA 2013, section 9.6.2; U.S. EPA, 2014, section 6.2, Table 6-1 and Figure 6-2). For
21 example, in seven or the 12 species, the W126 index value for which 2% seedling biomass loss is
22 estimated is below 8 ppm-hours and in the other five species, the W126 value for which 2%
23 biomass loss is estimated is above 18 ppm-hours. Within the group of seven more sensitive
24 species, the most sensitive are cottonwood and black cherry (U.S. EPA 2014, section 6.2).
4 Among the 12 species, in addition to the functions for 11 species studied in the FACE research, is the
field study by Gregg et al (2003) for cottonwood. As noted in discussion of uncertainties above, there is only the
single study available for this species which focused on a gradient of O3 exposure extending from the New York
metropolitan area which may contribute uncertainty to the quantitative characterization of C-R.
5-13
-------
DQ
Ct.
CO
d>
CD
CJ
o _
CNj
CD
CD
CD
Red Maple
Sugar Maple
Red Alder
Tulip Poplar
Ponderosa Pine
White Pine
Loblolly Pine
d Virginia Pine
d Cottonwood
• Aspen
• Black Cherry
• Douglas Fir
0
10
20
30
40
50
W126
2 Figure 5-1. Relative biomass loss in seedlings for 12 studied species in response to seasonal
3 ozone concentrations in terms of seasonal W126 index values (U.S. EPA 2014,
4 Figure 6-2).
5 In considering the evidence for the two most sensitive species, we recognize the more
6 limited quantitative information for cottonwood. The C-R function for this species is based on
7 the study by Gregg et al (2003) which described 63 effects on growth on eastern cottonwood
8 (Populus deltoides) saplings along an Os gradient between urban New York City and downwind
9 rural areas for the years 1992, 1993 and 1994. While the authors additionally studied
10 cottonwood response in a companion controlled OTC experiment, the C-R function described
11 here is based on the field data and reflects the single study (U.S. EPA, 2013, sections 9.2.5 and
12 9.6.3). Given these factors, as noted in consideration of uncertainties above, we recognize there
13 may be additional uncertainty in this function.
14 Using the 11 concentration-response (C-R) functions for tree seedlings from the FACE
15 research, and with the addition of the cottonwood C-R function, Figure 5-2 below presents three
16 approaches for characterizing the median response function (U.S. EPA, 2014, section 6.2.1.2 and
17 Figure 6-4). The first approach used a composite function developed from the C-R functions for
18 all 52 tree seedling studies (across the 12 species) combined (red line). The second approach
19 used only the median C-R function, when available, for each of the 12 tree species included
20 (green line)5. The third approach used a stochastic sampling method to randomly select a C-R
' For some species, only one study was conducted so that C-R function was used.
5-14
-------
1
2
3
4
5
6
7
8
9
10
11
12
function for each species from the studies conducted for that species. For some species only one
study was available (e.g., Red Maple), and for other species there were as many as 11 studies
available (Ponderosa Pine). The process was repeated 1,000 times (grey lines), and the median
value was plotted for biomass loss values of 1% to 7%, and 10% (the bar associated with each
median point denotes the 25th and 75th percentile values. This third approach provides an
illustration of the effect of within-species variability on estimates of the median species. The
median W126 values are similar, when using all of the studies or just the composite C-R function
for each species; however, the median value is higher when within-species variability is included
(U.S. EPA, 2014, section 6.2.1.2). Among these three approaches, the median seasonal W126
index value for which a two percent biomass loss is estimated in seedlings for the studied species
ranges between approximately 7 and 14 ppm-hrs.
E °
S. "
CL
0 1
5 Q 1 8 9 10 11 12 13 14 15
Percent Biomass Loss
13 Figure 5-2. Relationship of tree seedling percent biomass loss with seasonal W126 index.
14 (From U.S. EPA 2014, Figure 6-4)
15 We further consider several recent studies that have provided important evidence of
16 growth effects occurring in the field at ambient exposure concentrations, including some
17 exposure concentrations which are at or below the level of the current standard. For example,
18 the study by McLaughlin et al., (2007a, b) investigated the effects of ambient 63 on tree growth
19 (measured by change in stem circumference) in the field at 3 forest sites (Cade's Cove, Look
5-15
-------
1 Rock and Oak Ridge) in the southern Appalachian Mountains for three years (2001, 2002, 2003)
2 (U.S. EPA, 2013, section 9.4.5). Compared to the growth observed in 2001 (base year), the
3 authors reported that ambient Os exposures, decreased seasonal stem growth by 30-50% for most
4 tree species in a high O^ year (2002), relative to the 2001 baseline. As shown in Table 5-6 below
5 (modified from Table 6-4 in U.S. EPA, 2014), the seasonal W126 index values ranged from
6 approximately 9 to 20 ppm-hours in the low Os year (2003) and from approximately 18 to 40
7 ppm-hours in the high Oj, year (2002), across the three sites. When these exposure
8 concentrations for the three years are expressed in terms of the 4th highest daily maximum 8-hour
9 average concentration, the values range from 71 to 90 ppb during the low year (2003) to 82 -
10 102 ppb during the high year (2002). The 3-year average of these 4th highest daily maximums6
11 for the period 2001-2003 are 76 ppb, 93 ppb, and 87 ppb for Cade's Cove, Look Rock, and Oak
12 Ridge, respectively. Thus, while two of the years at Cade's Cove had annual values at or below
13 75 ppb, the concentrations at all sites for the 2001- 2003 period appear to exceed the current
14 standard. Only one species, tulip poplar, has data for all three years and was present at all three
15 sites (U.S. EPA, 2014, Table 6-4). This subset of data is what is shown in table 5-1 below.
16 Table 5-1. Ozone concentrations associated with effects on tulip poplar in southern
17 Appalachian Mountains (2001-2003).
Ambient Os concentrations for Mclaughlin et al., 2007A
2001
2002
2003
2001-
2003
Cade's Cove
Max4
high
(ppm)
75
82
71
76
W126
(ppm-
hrs)
15
18
9
N/A
AAir quality values obtaim
rounded to nearest whole \
circumference compared t<
B Study results for percent
Percent change
in
circumference6
baseline
-62%
N/A
Look Rock
Max4
high
(ppm)
86
102
90
93
W126
(ppm-
hrs)
23
40
20
28
Percent change
in
circumference6
baseline
-26%
-38%
Oak Ridge
Max4
high
(ppm)
85
99
78
87
W126
(ppm-
hrs)
20
32
11
21
Percent change
in
circumference6
baseline
-49.6%
7.5%
;d from http://www.epa.sov/ttn/airs/airsaqs/detaildata/downloadaqsdata.htm. W126
>pm-hr. 2002 and 2003 study results measured as % change in tulip poplar stem
3 2001 baseline.
change in circumference of Tulip Poplar from WREA Table 6-4.
18
19
20
Based on the table above, appreciable growth reductions occurred in tulip poplar at all
three sites during a high Os year (2002), though it spanned a rather large range of -26% to -62%.
6 This is the form of the current standard. Thus, the current standard is met when the 3-year average of the
4th highest daily maximum 8-hour concentrations is at or below 75 ppb.
5-16
-------
1 The results of this study are hard to interpret given the different exposure base lines, magnitude
2 of changes in subsequent year exposures, and other undocumented site-specific factors. The
3 study findings do show, however, that higher Os concentrations (above the current standard)
4 significantly reduced growth compared to that in a lower O^ year.
5 With regard to evidence of effects associated with exposures that may be allowed by the
6 current standard, we additionally consider the study by Gregg et al (2003) referenced above that
7 documented Cb-induced growth loss on eastern cottonwood (Populus deltoides) saplings along
8 an Os gradient between urban New York City and downwind rural areas for the years 1992, 1993
9 and 1994 (U.S. EPA, 2013, section 9.2.5). Fourth highest daily max 8-hour average
10 concentrations during the experimental period (July 7 - Sept 20) ranged from 0.056 ppm to
11 0.098 ppm along the gradient (data in 2008 Os docket) while W126 index values ranged from 1.9
12 to 12 ppm-hrs. At some sites, the highest daily maximum 8-hour average ambient concentrations
13 (0.056 ppm, 0.072 ppm, and 0.073 ppm) were below the level of the current standard, indicating
14 that the standard would likely have been met. Figure 5-3 below shows the growth suppression
15 response of the cottonwood saplings to the range of Os exposures occurring along the gradient.
ro
m
90
80
70-
50
40-
30-
20
10-
0
= -6.1902x + 87.149 R2 = 0.8622
0.073
0.078
.088
0.098
.086
0.0 2.0 4.0 6.0 8.0 10.0 12.0 14.0
Ozone Exposure
16 (W126, ppm-hrs)
17 Figure 5-3. Association of Os with cottonwood biomass in gradient downwind of New
18 York City (Gregg et al., 2003). Symbol Key: Squares (1992); Circles (1993);
19 Triangles (1994); Shaded (urban); open (rural). Figure modified from Gregg et al.,
20 2003 and U.S. EPA, 2007 Figure 7-17 to add 8-hour average concentrations (red
21 font).
5-17
-------
1 We additionally recognize some limitations in the extent to which some studies that look
2 at Os-induced effects on tree growth, productivity and carbon storage can inform consideration
3 of effects that might occur under air quality conditions that meet the current standard. In some
4 cases this is because studies were designed to look only at what is happening under ambient
5 conditions and/or above ambient conditions and the ambient conditions often did not meet the
6 standard. In addition, because tree growth is known to be affected by cumulative exposures,
7 studies do not commonly provide exposure concentration data pertaining to daily 8-hour
8 averages. Unless hourly 63 concentration data are provided, translation from one index to
9 another is not usually feasible. Thus, to consider effects that might occur under conditions that
10 meet the current standard we have traditionally relied on monitored hourly 63 air quality data to
11 calculate both types of exposure metrics (i.e. cumulative and 8-hour average) used in
12 combination with the robust tree seedling C-R functions developed in OTC studies (as described
13 above) to predict the magnitude of the growth effects associated with exposure concentrations at
14 or below that of the current standard. These tree seedling C-R functions are also used in the
15 WREA estimates of effects on tree growth that might be expected to remain under air quality
16 conditions that just meet the current secondary standard (discussed in section 5.2.2 below).
17 • To what extent does currently available evidence suggest locations where the
18 vulnerability of sensitive species, ecosystems and/or their associated services to
19 Os-related effects on forest tree growth, productivity and carbon storage would
20 have special significance to the public welfare?
21 Areas with special significance to the public welfare include federally designated Class I,
22 non-Class I national parks, and other areas set aside to provide similar public welfare benefits.
23 Therefore Table 5-2 provides some examples of Class I sites where the current secondary
24 standard is met but W126 index values fall above 15 ppm-hours. At W126 index values above
25 15 ppm-hours, relative biomass loss in seedlings might be expected to range from less than 1%
26 to more than 30% across the 12 species for which C-R functions have been developed (as
27 described above). Relative biomass loss above 5% might be expected in seedlings of six of the
28 12 species and above 2% in seedlings (but less than 5%) in a seventh, indicating that if present in
29 these specially protected areas, meeting the current standard may not prevent W126 index values
30 that may cause these levels of growth effect.
31
32
5-18
-------
1 Table 5-2. Examples of counties containing Class I areas where recent air quality might be
2 expected to meet the current standard and where 3-yr W126 index values are
3 above 15 ppm-hrs.
Monitor ID
806710041
806710041
806710041
80830101
80830101
4005800
4005800
4005800
49053013
49053013
49053013
49053013
49053013
49037010
49037010
49037010
Years
2008-2010
2009-2011
2010-2012
2006-2008
2007-2009
2006-2008
2007-2009
2010-2012
2006-2008
2007-2009
2008-2010
2009-2011
2010-2012
2006-2008
2007-2009
2010-2012
3-yr 8-
hrmax
71
74
73
71
69
70
68
72
71
70
70
70
73
71
70
69
3-yr W1 26
(annual values)
15
(16,11,18)
17
(11,18,21)
19
(18,21,18)
18
(24, 18, 14)
16
(18, 14, 15)
19
(22,19, 22)
15
(19,22,11)
18
(15, 18, 20)
21
(24, 19, 20)
18
(19, 20, 15)
18
(20, 15, 21)
18
(15,21,18)
20
(21,18,22)
18
(19, 18, 18)
16
(18, 18, 13)
15
(14, 14, 17)
State/Co.
Colorado/ La Plata
Colorado/ La Plata
Colorado/ La Plata
Colorado/
Montezuma
Colorado/
Montezuma
Arizona/ Coconino
Arizona/ Coconino
Arizona/ Coconino
Utah/ Washington
Utah/ Washington
Utah/ Washington
Utah/ Washington
Utah/ Washington
Utah/ San Juan
Utah/ San Juan
Utah/ San Juan
Class I Area
Weminuche
WAa
Weminuche WA
Weminuche WA
Mesa Verde NPb
Mesa Verde NP
Grand Canyon
NP
Grand Canyon
NP
Grand Canyon
NP
Zion NP
Zion NP
Zion NP
Zion NP
Zion NP
Canyonlands
NP
Canyonlands
NP
Canyonlands
NP
Based on data from http://www.epa.qov/ttn/airs/airsaqs/detaildata/downloadaqsdata.htm
a = Wilderness Area; b= National Park
5-19
-------
2
3
4
5
6
7
5.2.2 Exposure/Risk-based Considerations
The WREA presents a number of quantitative analyses of cumulative exposure and risk
related to tree growth, productivity and carbon storage for air quality scenarios intended to
inform our consideration of exposure and risk for considerations associated with the current
standard (see Table 5-3 below, U.S. EPA, 2014, chapter 6).
Table 5-3. Exposure, risk and ecosystem services analyses related to tree growth,
productivity and carbon storage.
WREA
estimates*
Species Level
Effects
Ecosystem Level
Effects
Percent of total
geographic area6 with
annual relative
biomass loss above
2%
Number of assessed
Class 1 areas with
annual relative
biomass loss above
2%
Ecosystem Services
• Economic surplus to timber producers and
consumers (WREA, Table 6-11)
• Carbon storage, nationally (WREA, Table 6-18)
• Carbon storage, in 5 urban areas (WREA, Table
6-20)
• Air pollutant removal in 5 urban areas (WREA,
Table 6-21)
ft See WREA chapter 6.
BThe total geographic area includes the contiguous U.S..
9
10
11
12
13
14
15
16
17
18
19
20
21
22
The relevant quantitative exposure and risk analyses for tree biomass loss, productivity
and carbon storage include:
1) Species-specific and composite biomass loss estimates.
2) National-scale assessments for: a) basal area weighted relative biomass loss for tree
seedlings; b) timber production; c) carbon sequestration.
3) Case study-scale assessments for: a) carbon sequestration; b) air pollution removal.
• For what air quality scenarios were exposures and risks estimated? What
approaches were used to estimate W126 exposures for those conditions? What
are associated limitations and uncertainties?
In the analyses related to Os effects on tree growth, productivity and carbon storage,
quantitative estimates were developed for five air quality scenarios by the methodology
summarized in Table 5-4 below. In general, this methodology involved two steps. The first is
derivation of the average W126 value (across the three years) at each monitor location. This
value is based on unadjusted data for recent conditions and model-adjusted concentrations for the
5-20
-------
1 four other scenarios. The development of model adjusted concentrations was done for each of 9
2 regions independently (see U.S. EPA, 2014, section 4.3.4.1). In the second step, national-scale
3 spatial surfaces (W126 values for each model grid cell) were created using the monitor-location
4 values and the Voronoi Neighbor Averaging (VNA) spatial interpolation technique (details on
5 the VNA technique are presented in U.S. EPA, 2014 Appendix 4A).
6 Table 5-4. Summary of methodology by which national surface of average W126 index
7 values was derived for each air quality scenario.
Scenario
Development of W126 values for Each Air Quality Scenario
Monitor-location-specific calculations
and any model adjustment
Derivation of national surface of
average W126 index values
Recent Conditions
(2006-2008)
An annual W126 index value is calculated for each
year at each monitor location, using the highest 3-
month period. A location-specific 3-year W126 was
calculated by averaging annual W126 index values
from 3 consecutive years which may have used
different 3-month periods..
Current Standard
2006-2008 hourly Oa concentrations at each monitor
location are model-adjustedA to create a three year
record of Oa concentrations that just meets the
current standard (see WREA, section 4.3.4).
A seasonal W126 index value is calculated for each
year at each monitor location using the same 3-
month period for each year (which is the highest as a
3-yr average and is highest in at least one of the
years). A location-specific average is derived from
these three index values.
Average W126 Index
of 15ppm-hrs
Average W126 Index
of 11 ppm-hrs
Average W126 Index
of 7 ppm-hrs
Hourly Oa concentrations at each monitor location,
within each modeling region, are model-adjusted to
create a record for which the highest location-specific
average index value in the region (the controlling
location) just meets the scenario target index value.
A seasonal W126 index value is calculated for each
year (of 2006-2008 period) at each monitor location,
using the same 3-month period for each year (which
is the highest in at least one of the years). A location-
specific average is derived from these three index
values.
The VNA method is applied to
the monitor-location average
W126 values to create a
national distribution of W126
values within model grid-cells
for each scenario.
AThe model-based adjustment approach is based on regional emission reduction scenarios at monitor sites
followed by spatial interpolation for broader spatial coverage. See WREA, chapters 4 and 7, and Appendices
4A and 7A).
5-21
-------
1
2 During the recent conditions period (2006 through 2008), the average W126 index values
3 (across the three-year recent conditions period) at the monitor locations ranged from below 10
4 ppm-hrs to 48.6 ppm-hrs (U.S. EPA 2014, Figure 4-4 and Table 4-1). Across the nine modeling
5 regions, the maximum average W126 index values ranged from 48.6 ppm-hrs in the west region
6 down to 6.6 ppm-hrs in the northwest region. After model-adjustment of the 2006-2008 data to
7 just meet the current standard in each region, the region-specific maximum values range from
8 18.9 ppm-hrs in the west region to 2.6 ppm-hrs in the northeast region (U.S. EPA, 2014, Table 4-
9 1). After application of the VNA technique to the current standard scenario monitor location
10 values, the average W126 values was below 15 ppm-hrs across the national surface with the
11 exception of a very small area of the southwest region (near Phoenix) where the average W126
12 values was near or just above 15 ppm-hrs. A lowering of the highest values occurred with
r\
13 application of the interpolation method as a result of estimating W126 values at a 12x12 km grid
14 resolution rather that at the exact location of a monitor. This indicates one uncertainty associated
15 with this aspect of the approach to estimating W126 values for the model-adjusted air quality just
16 meeting the current standard.
17 The WREA also recognizes other sources of uncertainty for the W126 estimates for each
18 air quality scenario and qualitatively characterizes the magnitude of uncertainty and potential for
19 directional bias. These sources of uncertainty are described in more detail in the WREA Chapter
20 4 and summarized below. Because the W126 estimates generated in the air quality analyses are
21 inputs to the vegetation risk analyses for biomass loss and foliar injury, any uncertainties in the
22 air quality analyses are propagated into the those analyses (U.S. EPA 2014, section 8.5).
23 An important large uncertainty in the analyses is the assumed response of the W126
24 concentrations to emissions reductions needed to meet the existing standard (U.S. EPA, 2014,
25 section 8.5.1). We note that any approach to characterizing 63 air quality over broad geographic
26 areas based on concentrations at monitor locations will convey inherent uncertainty. The model-
27 based adjustments, based on U.S.-wide emissions reductions in oxides of nitrogen (NOx), do not
28 represent air quality distributions from an optimized control scenario that just meets the current
29 standard (or target W126 values for other scenarios), but characterize one potential distribution
30 of air quality across a region when all monitor locations meet the standard (U.S. EPA 2014,
5-22
-------
1 section 4.3A2).7 An additional uncertainty comes from the creation of a national W126 surface
2 using the VNA technique to interpolate recent air quality measurements of 63. In general, spatial
3 interpolation techniques perform better in areas where the Os monitoring network is denser.
4 Therefore, the W126 estimated in the rural areas in the West, Northwest, Southwest, and West
5 North Central with few or no monitors (Figure 2-1) are more uncertain than those estimated for
o
6 areas with more dense monitoring . Additionally, the surface is created from the three-year
7 average at the monitor locations, rather than creating a surface for each year and then averaging
8 across years at each grid cell; the potential impact of this on the resultant estimates is considered
9 in the WREA (U.S. EPA, 2014, Appendix 4A).
10 • What are the nature and magnitude of exposure- and risk-related estimates for
11 tree growth, productivity, and carbon storage under recent conditions or
12 conditions remaining upon meeting the current standard? To what extent are
13 these exposures and risks important from a public welfare perspective?
14 The WREA used the C-R functions for 12 species described above with information on
15 the distribution of those species across the U.S., and average W126 exposure estimates to
16 estimate relative biomass loss for each of the studied species for each national air quality
17 scenario (U.S. EPA, 2014, section 6.2.1.3 and Appendix 6A). For example, the estimates of
18 relative biomass loss of Ponderosa Pine for air quality model-adjusted to just meet the current
19 standard are illustrated in Figure 5-4 below. While relative biomass loss below 2% is estimated
20 for most areas where this species is found, estimates in some areas of the southwest fall in a bin
21 defined as 2.01-4% biomass loss (U.S. EPA 2014, Figure 6-7 and Appendix 6A).
22
23
24
25
26
27
7 Because our analyses used U.S.-wide NOx emissions reductions to simulate just meeting the existing
standard independently in each region, there are broad regional reductions in O3 even in meeting standards in urban
areas when targeting a few high-ozone urban monitors for reductions. However, the assumption of broad regional or
national NOx reductions are not unreasonable given current EPA regulations such as the Clean Air Interstate Rule
(CAIR), which requires NOx cuts across the Eastern U.S. to reduce regional ozone transport, and the multitude of
onroad and offroad mobile source rules that will lead to reduction in NOx from these sources across the country in
future years.
5-23
-------
Relative BiomassLoss
| 1.01% -2%
I 2,01%-4%
4.01%-7%
7 01%-10%
10.01%-15%
15.01%-20%
I 20.01%-20%
2 Figure 5-4. Relative biomass loss of Ponderosa Pine for air quality model-adjusted to just
3 meet the current standard (U.S. EPA 2014, Figure 6-7).
4 The WREA also developed national-scale estimates of 63 biomass ecosystem-level
5 impacts considering the 12 studied species together (U.S. EPA 2014, section 6.8, Table 6-22).
6 This was done using the species-specific biomass loss C-R functions, information on prevalence
7 of the studied species across the U.S., and a weighting approach based on proportion of the basal
8 area that each species contributes. The RBL values for multiple tree species were weighted by
9 their basal area and combined into a weighted RBL value. The weighted relative biomass loss
10 (wRBL) is intended to inform our understanding of the potential magnitude of the ecological
11 effect that could occur in some ecosystems. Specifically, the more basal area that is affected in a
12 given ecosystem, the larger the potential ecological effect. A wRBL value for each grid cell is
13 generated by weighting the RBL value for each studied tree species found within that grid cell by
14 the proportion of basal area it contributes to the total basal area of all tree species within the grid
15 cell, and then summing those individual wRBLs. The percent of total basal area that exceeds a
16 2% weighted relative biomass loss in the recent conditions scenario is 10.1% (U.S. EPA 2014,
17 Table 6-24). Based on the average W126 values estimated for the air quality scenario just
5-24
-------
1 meeting the current standard across the contiguous U.S., the WREA estimates 0.8 % of the total
2 geographic area to have a wRBL above 2% (U.S. EPA 2014, Table 6-24).
3 We also consider WREA estimates (quantitative and qualitative) of effects on several
4 ecosystem services. First, impacts on growth related to 63 concentrations in federally-designated
5 Class I areas were derived from an average weighted RBL value (discussed above) for 145 of the
6 155 Class I areas (U.S. EPA 2014, section 6.8.1). Given established objectives for Class I areas
7 (e.g., to maintain in perpetuity), effects in Class I areas may be considered to have the potential
8 to adversely affect the intended use of the ecosystem, i.e., to maintain in pristine natural
9 conditions for future generations. For the recent conditions scenario, this analysis estimates
10 wRBL values above 2% in 13 of the 145 assessed Class I areas. In comparison, the analysis
11 estimates wRBL values above 2% in only 2 areas based on average W126 values estimated for
12 the current standard scenario (U.S. EPA 2013, Table 6-26).
13 The WREA also presents national-scale estimates of the effects of biomass loss on timber
14 production and agricultural harvesting, as well as on carbon sequestration. The WREA used the
15 Oj, C-R functions for tree seedlings to calculate relative biomass loss. Because the forestry and
16 agriculture sectors are related, and trade-offs occur between the sectors, the WREA also
17 calculated the resulting market-based welfare effects of O?, exposure in the forestry and
18 agriculture sectors. In the analyses for commercial timber production, based on the average
19 W126 values estimated for the air quality scenario just meeting the existing standard, relative
20 biomass losses (RBL) estimates were below one percent in all regions except the Southwest,
21 Southeast, Central, and South regions (U.S. EPA, 2014, section 6.3, Table 6-8) (see U.S. EPA,
22 2014, Table 6-14 for clarification on region names)). Relative biomass losses remain above one
23 percent for the average W126 scenarios for 15 and 11 ppm-hrs in parts of the Southeast, Central,
24 and South regions, and for the 7 ppm-hr scenario in the Southeast and South regions (U.S. EPA,
25 2014, section 6.3, Table 6-8).
26 In addition to estimating changes in forestry and agricultural yields, the WREA presents
27 estimated changes in consumer and producer/farmer surplus associated with the change in yields.
28 Changes in biomass affect individual tree species, but the overall effect on forest ecosystem
29 productivity depends on the composition of forest stands and the relative sensitivity of trees
30 within those stands. Economic welfare impacts resulting from just meeting the existing and
31 alternative standards were largely similar between the forestry and agricultural sectors —
32 consumer surplus, or consumer gains, generally increased in both sectors because higher
33 productivity under lower 63 concentrations increased total yields and reduced market prices.
34 Comparisons are not straightforward to interpret due to market dynamics. For example, because
35 demand for most forestry and agricultural commodities is not highly responsive to changes in
5-25
-------
1 price, there were more examples for which producer surplus, or producer gains declines. In
2 some cases, lower prices reduce producer gains more than can be offset by higher yields.9 In
3 general, the WREA estimates benefits for agricultural yield of approximately 1 billion dollars
4 from differences in average W126 index values representing recent conditions from those
5 representing model-adjusted air quality for the current standard (U.S. EPA 2014, section 6.3).
6 The national-scale analysis of carbon dioxide (CO2) sequestration estimates substantially more
7 storage under the current standard compared to recent conditions (U.S. EPA 2014, section 6.6.1,
8 Table 6-18, Appendix 6B). In considering the significance of the potential climate and
9 ecosystem service impact, we also take note of the large uncertainties associated with this
10 analysis (see U.S. EPA 2014, Table 6-27).
11 We additionally consider the WREA estimates of tree growth and ecosystem services
12 provided by urban trees over a 25-year period, for five urban areas based on case-study scale
13 analyses that quantified the effects of biomass loss on carbon sequestration and pollution
14 removal (U.S. EPA 2014, sections 6.6.2 and 6.7).10 The urban areas included in this analysis
15 represent diverse geography in the Northeast, Southeast, and Central regions, although do not
16 include an urban area in the western part of the U.S. Estimates of the effects of Os-related
17 biomass loss on carbon sequestration indicate the potential for an increase of somewhat more
18 than a million metric tons of COz equivalents for average W126 values associated with meeting
19 for the current standard scenario as compared to recent conditions. Somewhat smaller increases
20 are estimated for the three W126 scenarios in comparison to the current standard scenario (U.S.
21 EPA 2014, section 6.6.2 and Appendix 6D).
22 In addition to the quantitative assessments discussed above, qualitative assessments for
23 some ecosystem services, such as commercial non-timber forest products and recreation (U.S.
24 EPA, 2014, section 6.4), aesthetic and non-use values (U.S. EPA, 2014, section 6.4), increased
25 susceptibility to insect attack and fire damage (U.S. EPA, 2014, sections 5.3 and 5.4,
26 respectively), were also conducted. Other ecological effects that are causally or likely causally
27 associated with Os exposure such as terrestrial productivity, water cycle, biogeochemical cycle,
28 and community composition (U.S. EPA 2013, Table 9-19) were not directly addressed in the
29 WREA due to a lack of sufficient quantitative information.
9 The relative biomass loss estimates were also used with the Forest and Agricultural Sector Optimization
Model with Greenhouse Gases (FASOMGHG). FASOMGHG is a national-scale model that provides a complete
representation of the U.S. forest and agricultural sectors' impacts of meeting alternative standards. FASOMGHG
simulates the allocation of land over time to competing activities in both the forest and agricultural sectors.
FASOMGHG results include multi-period, multi-commodity results over 60 to 100 years in 5-year time intervals
when running the combined forest-agriculture version of the model. See Chapter 6, Section 6.3 of the WREA for a
discussion of economic welfare and consumer and producer surplus.
10 The model, iTree, a peer-reviewed suite of software tools provided by USFS was used in this analysis.
5-26
-------
1 There is substantial heterogeneity in plant responses to 03, both within species, between
2 species, and across regions of the U.S. The (Vsensitive tree species are different in the eastern
3 and western U.S. — the eastern U.S. has far more species. Ozone exposure and risk is somewhat
4 easier to assess in the eastern U.S. because of the availability of more data and the greater
5 number of species to analyze. In addition, there are more 63 monitors in the eastern U.S. but
6 fewer national parks (U.S. EPA, 2014, chapter 8).
7 • What are the uncertainties associated with both quantitative and qualitative
8 information?
9 Several key limitations and uncertainties, which may have a large impact on both overall
10 confidence and confidence in individual analyses, are discussed here. Key uncertainties
11 associated with the assessment of impacts on ecosystem services at the national and case-study
12 scales, as well as across species, U.S. geographic regions and future years include those
13 associated with the following seedling C-R functions, as well as interpolated and model-adjusted
14 Os concentrations used to estimate W126 exposures in the REA air quality scenarios. The
15 uncertainties in the W126 exposure estimates are discussed above at the beginning of section
16 5.2.2.
17 With regard to the seedling C-R functions, the description of Figure 5-2 above provides
18 some characterization of the variability of individual study results and the impact of that on
19 estimates of W126 index values that might elicit different percentages of biomass loss in tree
20 seedlings (U.S. EPA, 2014, section 6.2.1.2). Even though the evidence shows that there are
21 additional species adversely affected by Ch-related biomass loss, the WREA only has C-R
22 functions available to quantify this loss for 12 tree species and 10 crop species. This absence of
23 information only allows a partial characterization of the Os-related biomass loss impacts in trees
24 and crops associated with recent 63 index values and with just meeting the existing and
25 potential alternative secondary standards. In addition, there are uncertainties inherent in these C-
26 R functions, including the extrapolation of relative biomass loss rates from tree seedlings to
27 adult trees and information regarding within-species variability. The overall confidence in the
28 C-R function varies by species based on the number of studies available for that species. Some
29 species have low within-species variability (e.g., many agricultural crops) and high
30 seedling/adult comparability (e.g., Aspen), while other species do not (e.g., Black Cherry). The
31 uncertainties in the C-R functions for biomass loss and in the air quality analyses are propagated
32 into the analysis of the impact of biomass loss on ecosystem services, including provisioning
33 and regulating services (U.S. EPA, 2014, Table 6-27). The WREA characterizes the direction
34 of potential influence of C-R function uncertainty as unknown, yet its magnitude as high,
5-27
-------
1 concluding that further studies are needed to determine how accurately the assessed species
2 reflect the larger suite of (Vsensitive tree species in the U.S. (U.S. EPA, 2014, Table 6-27).
3 Another uncertainty associated with interpretation of the WREA biomass loss-related
4 estimates concerns the potential for underestimation of compounding of growth effects across
5 multiple years of varying concentrations. Though tree biomass loss impacts were estimated
6 using air quality scenarios of 3-year average W126 values, the WREA also conducted an
7 analysis to compare the impact of using a variable compounding rate based on yearly variations
8 in W126 exposures to that of using a W126 index value averaged across three years. The WREA
9 compared the compounded values for two examples. In both examples, one species (Tulip Polar
10 and Ponderosa Pine) and one climate region where that species occurred (Southeast and
11 Southwest regions) were chosen and air quality values associated with just meeting the existing
12 standard of 75 ppb were used. Within each region the WREA analysis used both the W126 value
13 at each monitor in the region for each year and the three-year average W126 value using the
14 method described in Chapter 4. The results show that the use of the three-year average W126
15 index value may underestimate RBL values slightly, but the approach does not account for
16 moisture levels or other environmental factors that could affect biomass loss (U.S. EPA, 2014,
17 section 6.2.1.4 and Figure 6-13). In considering these results, we note that in both regions and in
18 all three years, the three-year average W126 value is sometimes above and sometimes below the
19 individual year W126 index value.
20 In the national-scale analyses of timber production, agricultural harvesting, and carbon
21 sequestration, the WREA used the FASOMGHG model, which includes functions for carbon
22 sequestration, assumptions regarding proxy species, and non-W126 C-R functions for three
23 crops. However, FASOMGHG does not include agriculture and forestry on public lands,
24 changes in exports due to O^ into international trade projections, or forest adaptation. Despite
25 the inherent limitations and uncertainties, the WREA concludes that the FASOMGHG model
26 reflects reasonable and appropriate assumptions for a national-scale assessment of changes in
27 the agricultural and forestry sectors due to changes in vegetation biomass associated with O3
28 exposure (U.S. EPA, 2014, sections 6.3, 6.5, 6.6, and 8.5.2).
29 In the case study analyses of five urban areas, the WREA used the iTree model, which
30 includes an urban tree inventory for each area and species-specific pollution removal and carbon
31 sequestration functions. However, iTree does not account for the potential additional VOC
32 emissions from tree growth, which could contribute to Oi formation. Despite the inherent
33 limitations and uncertainties, the WREA concludes that the iTree model reflects reasonable and
5-28
-------
1 appropriate assumptions for a case study assessment of pollution removal and carbon
2 sequestration for changes in biomass associated with O^ exposure (U.S. EPA, 2014, sections
3 6.6.2, 6.7, and 8.5.2).
4 The overall effect of the combined set of uncertainties on confidence in the interpretation
5 of the results of the analyses is difficult to quantify. Due to differences in available information,
6 the degree to which each analysis was able to incorporate quantitative assessments of uncertainty
7 differed.
8 5.3 CROP YIELD LOSS
9 This section considers the current evidence and exposure/risk information to inform
10 consideration of the adequacy of the protection provided by the current standard from known and
11 anticipated adverse welfare effects of 63 related to crop yield and other associated effects. Crops
12 are important from a public welfare perspective because they provide food and fiber services to
13 humans. This section includes a discussion of the policy-relevant science and weight-of-
14 evidence conclusions discussed in the ISA (section 5.3.1) and the exposure/risk results (section
15 5.3.2) described in the second draft WREA. Important uncertainties and limitations in the
16 available information are discussed throughout the sections. These discussions highlight the
17 information we consider relevant to answering the overarching question and associated policy-
18 relevant questions included in this section.
19 5.3.1 Evidence-based Considerations
20 Ozone can interfere with carbon gain (photosynthesis) and allocation of carbon. As a
21 result of decreased carbohydrate availability, fewer carbohydrates are available for plant growth,
22 reproduction, and/or yield. For seed-bearing plants, these reproductive effects will culminate in
23 reduced seed production or yield. The detrimental effect of Os on crop production has been
24 recognized since the 1960s, and current Os concentrations in many areas across the U.S. are high
25 enough to cause yield loss in a variety of agricultural crops including, but not limited to,
26 soybeans, wheat, potatoes, watermelons, beans, turnips, onions, lettuces, and tomatoes.
27 Increases in Oj concentration may further decrease yield in these sensitive crops while also
28 causing yield losses in less sensitive crops (U.S. EPA 2013, section 9.4.4). The ISA concluded
29 that the evidence is sufficient to determine that there is a causal relationship between Oj,
30 exposure and reduced yield and quality of agricultural crops (U.S. EPA 2013, Table 2-2).
5-29
-------
1 • To what extent has scientific information become available that alters or
2 substantiates our prior conclusions regarding Os-related crop yield loss and of
3 factors that influence associations between Os levels and crop yield loss?
4 In general, the vast majority of the new scientific information has substantiated our prior
5 conclusions regarding 63 crop yield loss. On the whole, this evidence supports previous
6 conclusions that exposure to 63 decreases growth and yield of crops. The ISA describes average
7 yield loss reported across a number of meta-analytic studies have been published recently for
8 soybean wheat, rice, semi-natural vegetation, potato, bean and barley (U.S. EPA 2013, section
9 9.4.4.1). Meta-analysis allows for the objective development of a quantitative consensus of the
10 effects of a treatment across a wide body of literature. Further, several new exposure studies
11 continue to show decreasing yield and biomass in a variety of crops with increased 63 exposure
12 (U.S. EPA 2013, section 9.4.4.1, Table 9-17). Research has linked increasing OT, concentration
13 to decreased photosynthetic rates and accelerated aging (U.S. EPA 2013, section 9.4.4) in leaves,
14 which are related to yield. Recent research has highlighted the effects of 63 on crop quality.
15 Increasing O^ concentration decreases nutritive quality of grasses, decreases macro- and micro-
16 nutrient concentrations in fruits and vegetable crops (U.S. EPA 2013, section 9.4.4). The
17 findings of these studies did not change our understanding of (Vrelated crop loss and little
18 information has emerged on factors that influence associations between 03 levels and crop yield
19 loss.
20 • To what extent have important uncertainties identified in the last review been
21 reduced and/or new uncertainties emerged?
22 Important uncertainties have been reduced regarding the confidence placed in using crop
23 exposure-response functions, especially for soybean. In general, the ISA reports consistent
24 results across exposure techniques and across crop varieties.
25 Two important uncertainties have been reduced regarding the C-R functions for yield
26 effects of Os in crop species, especially for soybean. First, in the last several reviews, the extent
27 to which C-R functions developed in OTC predicted plant responses in the field and under
28 different exposure conditions was not clear. In this review, staff from the EPA's ORD/NCEA
29 performed an analysis comparing OTC data with field-based data for one crop and one tree
30 species (U.S. EPA, 2013, section 9.6.3.2). The crop comparison was done using soybean OTC
31 data from NCLAN and field-based data from Soy FACE. The NCLAN studies were undertaken
32 in the early to mid 1980's and provide the largest, most uniform database on the effects of Os on
33 agricultural crop yields (U.S. EPA 1996; U.S. EPA 2006; U.S. EPA 2013, sections 9.2, 9.4, and
5-30
-------
1 9.6). n The Soy FACE experiment was a chamberless field-based exposure study in Illinois that
2 was conducted from 2001 - 2009 (U.S. EPA 2013, section 9.2.4). Yield loss in soybean from O3
3 exposure at the Soy FACE field experiment was reliably predicted by soybean C-R functions
4 developed in NCLAN (U.S. EPA, 2013, Section 9.6). This analysis supports the robustness and
5 use of the C-R functions developed in NCLAN to predict relative yield loss to O3 exposure.
6 A second area of uncertainty that was reduced is that regarding the application of the
7 NCLAN C-R functions, developed in the 1980s, to more recent cultivars currently growing in the
8 field. However, recent studies continue to find yield loss levels in crop species studied
9 previously under NCLAN that reflect the earlier findings. There has been little new evidence
10 that crops are becoming more tolerant of O3 (U.S. EPA, 2006a; U.S. EPA 2013). This is
11 especially evident in the research on soybean. In a meta-analysis of 53 studies, Morgan et al.
12 (2003) found consistent deleterious effects of O3 exposures on soybean from studies published
13 between 1973 and 2001. Further, Betzelberger et al. (2010) has recently utilized the SoyFACE
14 facility to compare the impact of elevated O3 concentrations across 10 soybean cultivars to
15 investigate intraspecific variability of the O3 response. The C-R functions derived for these 10
16 current cultivars were similar to the response functions derived from the NCLAN studies
17 conducted in the 1980s (Heagle, 1989), suggesting there has not been any selection for increased
18 tolerance to O3 in more recent cultivars. The 2013 ISA reported comparisons between yield
19 predictions based on data from cultivars used in NCLAN studies, and yield data for modern
20 cultivars from SoyFACE (U.S. EPA, 2013, section 9.6.3). They confirm that the average
21 response of soybean yield to O3 exposure has not changed in current cultivars. Thus, staff
22 concludes that at least for soybean, uncertainties associated with use of the NCLAN generated C-
23 R functions to estimate biomass loss in recent cultivars has been reduced.
24 • To what extent does the available evidence indicate the occurrence of Os-related
25 effects on crop yield loss attributable to cumulative exposures at lower ambient Os
26 concentrations than previously established or to exposures at or below the level of
27 the current standard?
28 Very little evidence has emerged to indicate a lower level cumulative exposures that can
29 affect crop yield (levels of concern), that have been based on C-R functions from OTC
11 The NCLAN protocol was designed to produce crop exposure-response data representative of the areas
in the U.S. where the crops were typically grown. In total, 15 species (e.g., corn, soybean, winter wheat, tobacco,
sorghum, cotton, barley, peanuts, dry beans, potato, lettuce, turnip, and hay [alfalfa, clover, and fescue]), accounting
for greater than 85 percent of U.S. agricultural acreage planted at that time, were studied. Of these 15 species, 13
species including 38 different cultivars were combined in 54 cases representing unique combinations of cultivars,
sites, water regimes, and exposure conditions. Crops were grown under typical farm conditions and exposed in
open-top chambers to ambient O3, sub-ambient O3, and above ambient O3.
5-31
-------
1 experiments. As described above, the new evidence has strengthened the basis for using the
2 information from the C-R functions.
3 Where the current evidence on crop yield loss is not in terms of parts per billion
4 concentrations over a specific exposure period such as eight hours, assessing whether O^
5 concentrations associated with meeting the current standard would allow crop yield effects is
6 more complex. Where feasible, we have attempted to characterize the Os exposures associated
7 with crop yield loss in terms of seasonal W126 index and we have separately considered the
8 extent to which such index values might be expected to occur in agricultural locations that meet
9 the current standard. For example, Sedgwick and Sumner counties in Kansas met the level of the
10 3-year 8-hr standard of 75 ppb in 2011. However, the W126 for 2011 in those counties was 19
11 ppm-hours and would be predicted to result in a 9% yield loss for soybean grown in those
12 counties.
13 • To what extent does currently available evidence suggest locations where the
14 vulnerability of sensitive species, ecosystems and/or their associated services to
15 Os-related crop yield loss would have special significance to the public welfare?
16 During the previous NAAQS reviews, there were very few studies that estimated Os
17 impacts on crop yields at large geographical scales (i.e., regional, national or global). Recent
18 modeling studies of the historical iimpact of 63 concentrations found that increased 63 generally
19 reduced crop yield, but the impacts varied across regions and crop species (U.S. EPA, 2013,
20 Section 9.4.4.1). The largest Cb-induced crop yield losses were estimated to occur in high-
21 production areas exposed to elevated 63 concentrations, such as the Midwest and the Mississippi
22 Valley regions of the United States. Among crop species, the estimated yield loss for wheat and
23 soybean were higher than rice and maize. Additionally, satellite and ground-based 63
24 measurements have been used to assess yield loss caused by Os over the continuous tri-state area
25 of Illinois, Iowa, and Wisconsin. The results indicate that O?, concentrations during the assessed
26 period reduced soybean yield, which correlates well with the previous results from FACE- and
27 OTC-type experiments (U.S. EPA 2013, section 9.4.4.1).
28 Thus, the recent scientific literature continues to support the conclusions of the 1996 and
29 2006 Criteria Documents and 2013 ISA that ambient 63 concentrations can reduce the yield of
30 major commodity crops in the U.S. and support the use of crop C-R functions based on OTC
31 experiments. Agricultural areas that would be likely to have the most significance to the public
32 welfare would be those high production areas for sensitive crops that also are exposed to high OT,
33 concentrations, such as areas in the Midwest and Mississippi Valley regions.
5-32
-------
1 5.3.2 Exposure/Risk-based Considerations
2 Two main analyses are conducted in the second draft WREA to estimate Os impacts
3 related to crop yield. Annual yield losses are estimated for 10 commodity crops and these
4 estimates are then additionally used to estimate 63 impacts on producer and consumer economic
5 surpluses (Table 5-5 below; U.S. EPA, 2014, sections 6.2, 6.5).
6 Table 5-5. Exposure, risk and ecosystem services analyses related to crop yield.
REA estimates
Crop-level impact*
Annual Relative Yield Loss
Corn, Cotton, Potato, Sorghum,
Soybean, Winter Wheat
Agri-Ecosystem Services6
Economic surplus to crop producers
and consumers
ASee section 6.2 WREA. 3See section 6.5 WREA.
7
8 • For what air quality scenarios were exposures and risks estimated? What
9 approaches were used to estimate W126 exposures for those conditions? What
10 are associated limitations and uncertainties?
11 The WREA crop analyses described here were performed for five air quality scenarios by
12 the methodology summarized in Table 5-4 above. In general, this methodology involved two
13 steps. The first is derivation of the average W126 value (across the three years) at each monitor
14 location. This value is based on unadjusted data for recent conditions and model-adjusted
15 concentrations for the 4 other scenarios. The development of model adjusted concentrations was
16 done for each of 9 regions independently (see U.S. EPA, 2014, section 4.3.4.1). In the second
17 step, national-scale spatial surfaces (W126 values for each model grid cell) were created using
18 the monitor-location values and the Voronoi Neighbor Averaging (VNA) spatial interpolation
19 technique (details on the VNA technique are presented in U.S. EPA, 2014 Appendix 4A). The
20 results of model adjustments on estimated average W126 values and grid cell estimates produced
21 from the VNA interpolation approach is summarized in section 5.2.2 above. A lowering of the
22 highest values occurred with application of the interpolation method as a result of estimating
9
23 W126 values at a 12x12 km grid resolution rather that at the exact location of a monitor. This
24 indicates one uncertainty associated with this aspect of the approach to estimating W126 values
25 for the model-adjusted air quality just meeting the current standard. Other areas of uncertainty
26 associated with the model adjustment and VNA interpolation approach are briefly summarized in
27 section 5.2.2 above and described in more detail in the WREA (U.S. EPA, 2014, chapter 4 and
28 Appendix 4A).
5-33
-------
1 • What is the nature and magnitude of the cumulative exposure- and risk-related
2 estimates for crop yield loss associated with remaining upon simulating just
3 meeting the current Os standard? What are the uncertainties associated this
4 information?
5 The WREA presents estimates of crop yield loss for the five air quality scenarios
6 described above using C-R functions for commodity crops across the country (U.S. EPA, 2014,
7 section 6.5). The largest reduction in Oj, exposure-related crop yield loss occurs when moving
8 from the recent conditions scenario to that for just meeting the current standard (U.S. EPA, 2014,
9 section 6.5). In the analyses for agricultural harvest, the largest estimates of yield changes also
10 occur when comparing the recent conditions scenario to that for the current standard. Under
11 recent conditions, the West, Southwest, and Northeast regions generally have the highest yield
12 losses. For the three average W126 scenarios, relative yield losses for winter wheat12 are less
13 than one percent. For soybeans, yield losses for these scenarios range from just above 1 percent
14 to below one percent (U.S. EPA 2014, section 6.5).
15 The WREA estimates of Os-attributable percent yield loss based on average W126 values
16 estimated for just meeting the current standard are relatively small (0.0 - 2.72%, U.S. EPA 2014,
17 section 6.5, Appendix 6B). In considering these estimates, we recognize the significant
18 uncertainties associated with several aspects of the analyses. Because the W126 estimates
19 generated in the air quality analyses are inputs to the vegetation risk analyses for biomass loss,
20 crop loss and foliar injury, any uncertainties in the air quality analyses are propagated into the
21 those analyses (U.S. EPA 2014, Table 6-27, section 8.5).
22 • To what extent are the exposures and risks remaining upon simulating just
23 meeting the current Os standard important from a public welfare perspective?
24 From a public welfare prospective, the O?, attributable risks to crops estimated for
25 conditions that just meet the current standard are small. As discussed in the WREA and
26 summarized above, there are multiple areas of uncertainty associated with these estimates,
27 including those associated with the model-based adjustment methodology as well as those
28 associated with projection of yield loss at the estimated 63 concentrations (U.S. EPA, 2014,
29 Table 6-27, section 8.5).
12 Among the major crops, because winter wheat and soybeans are more sensitive to ambient O3 levels than
other crops we include these crops for this discussion.
5-34
-------
1 • What are the ecosystem services potentially affected by Os-related crop yield loss
2 and to what extent are they important from a public welfare perspective? To
3 what degree can the magnitude of the O3 effect on these services be qualitatively
4 or quantitatively characterized?
5 The WREA presents national-scale estimates of the effects of biomass loss on timber
6 production and agricultural harvesting, as well as on carbon sequestration (U.S. EPA 2014,
7 section 6.5). Because the forestry and agriculture sectors are related, and trade-offs occur
8 between the sectors, the WREA also calculated the resulting market-based welfare effects of Os
9 exposure in the forestry and agriculture sectors. Overall effect on agricultural yields and
10 producer and consumer surplus depends on the (1) ability of producers/farmers to substitute
11 other crops that are less Os sensitive, and (2) responsiveness, or elasticity, of demand and supply
12 (U.S. EPA, 2014, sections 6.5, 8.2.1.3). Estimated (Vattributable economic welfare impacts on
13 agricultural sectors associated with air quality conditions model-adjusted to just meet the existing
14 and potential alternative W126 standard levels were largely similar between the forestry and
15 agricultural sectors. Estimates of consumer surplus, or consumer gains, were generally higher
16 under those conditions (compared to recent conditions) in both sectors because higher
17 productivity under lower O?, concentrations increased total yields and reduced market prices
18 (U.S. EPA 2014, Table 6-16). Because demand for most forestry and agricultural commodities
19 is not highly responsive to changes in price, there were more examples for which producer
20 surplus, or producer gains, decline. For agricultural welfare, annualized combined consumer and
21 producer surplus gains were estimated to be $2.6 trillion for model adjustment to meet the
22 current standard. Combined gains were essentially unchanged in comparisons of the current
23 standard scenario to the average W126 scenario for 15 ppm-hrs, but gains increased by $21
24 million for the W126 scenario for 11 ppm-hrs and $231 million for the W126 scenario for 7
25 ppm-hrs. In some cases, lower prices reduce producer gains more than can be offset by higher
26 yields (U.S. EPA, 2014, Table 6-17).
27 The WREA discusses multiple areas of uncertainty associated with these estimates (also
28 summarized above), including those associated with the model-based adjustment methodology as
29 well as those associated with projection of yield loss at the estimated 63 concentrations (U.S.
30 EPA, 2014, Table 6-27, section 8.5).
31 5.4 VISIBLE FOLIAR INJURY
32 Visible foliar injury resulting from exposure to 63 has been well characterized and
33 documented over several decades of research on many tree, shrub, herbaceous, and crop species
34 (U.S. EPA, 2013, 2006, 1996, 1984, 1978). The significance of O3 injury at the leaf and whole
35 plant levels depends on an array of factors and there is difficulty in relating visible foliar injury
5-35
-------
1 symptoms to other vegetation effects such as individual tree growth, or effects at population or
2 ecosystem characteristics (U.S. EPA, 2013, p. 9-39). Visible foliar injury by itself, however, can
3 impact the public welfare through damaging or impairing the intended use of the affected entity
4 or the service it provides. For example, ways by which Os-induced visible foliar injury may
5 impact the public welfare thus include: 1) visible damage to ornamental species or leafy crops
6 (spinach, lettuce, tobacco) that affects the economic value, yield, or usability of that plant (U.S.
7 EPA 2007, section 7.4.1; Abt Associates, Inc. 1995); 2) visible damage to plants with special
8 cultural significance (e.g., those used in tribal practices); 3) visible damage to species occurring
9 in natural settings valued for their scenic beauty and/or recreational appeal, including in areas
10 specially designated for more protection (i.e., federal Class I areas) (73 FR 16490). Given
11 limitations in the available information pertaining to the first two categories,13 the discussions of
12 the evidence and exposure/risk information in sections 5.4.1 and 5.4.2 below focus primarily on
13 what is known about visible foliar injury that has been shown to occur in natural settings valued
14 for their scenic beauty and/or recreational appeal.
15 At the time of the last review, the following was known:
16 1) Ozone causes diagnostic visible injury symptoms on studied bioindicator species.
17 2) Soil moisture is a major confounding effect that can decrease the incidence and
18 severity of visible foliar injury under dry conditions and visa versa.
19 3) The most extensive dataset regarding visible foliar injury incidence across the U.S.
20 was that collected by the U. S. Forest Service (USFS) Forest Health Monitoring/Forest
21 Inventory and Analysis (FHM/FIA) Program.
22 4) Staff analyses of county level air quality data and USFS biomonitoring data showed
23 that for each year within a four year period (2001 - 2004) the percentage of counties
24 having a biosite with visible foliar injury ranged between 11-30% at an 8-hour average
25 annual level of 0.074 ppm (U.S. EPA, 2007, section 7.6.3.2).
26 In the remainder of this section, we consider how the currently available evidence and
27 exposure/risk information informs our understanding of the relationship that exists between
28 visible foliar injury and exposures to Os in ambient air and consideration of the adequacy of
29 protection provided by the current standard. The policy-relevant evidence and weight-of-
30 evidence conclusions drawn from the ISA are discussed in section 5.4.1, and the exposure/risk
31 and associated ecosystem services estimates from the second draft WREA, are discussed in
32 section 5.4.2. Important uncertainties and limitations in each type of available information are
33 also discussed in these two sections.
13 Qualitative information regarding potential cultural impacts of O3-induced visible foliar injury is noted in
section 5.5 and Appendix 5A).
5-36
-------
1 5.4.1 Evidence-based Considerations
2 • To what extent has scientific information become available that alters or
3 substantiates our previous conclusions of Os-related visible foliar injury and of
4 factors that influence associations between Os exposures or concentrations and
5 visible foliar injury?
6 Recent research continues to build and substantiate the previous conclusions and findings
7 drawn from several decades of research on many tree, shrub, herbaceous, and crop species (U.S.
8 EPA, 2013, 2006, 1996, 1984, 1978) that O3-induced visible foliar injury symptoms are well
9 characterized and considered diagnostic on certain bioindicator plant species. Diagnostic usage
10 for these plants has been verified experimentally in exposure-response studies, using exposure
11 methodologies such as continuous stirred tank reactors (CSTRs), open-top chambers (OTCs),
12 and free-air fumigation (FACE). Although there remains a lack of robust exposure-response
13 functions that would allow prediction of visible foliar injury severity and incidence under
14 varying air quality and environmental conditions, experimental and observational evidence has
15 clearly established a consistent association of the presence of visible injury symptoms with Os
16 exposure, with greater exposure often resulting in greater and more prevalent injury (U.S. EPA
17 2013, section 9.4.2). This new research includes: 1) controlled exposure studies conducted to
18 test and verify the Os sensitivity and response of potential new bioindicator plant species; 2)
19 multi-year field surveys in several National Wildlife Refuges documenting the presence of foliar
20 injury in valued areas; 3) ongoing data collection and assessment by the USDA Forest Service
21 FUM/FIA program, including multi-year trend analysis (U.S. EPA 2013, section 9.4.2). These
22 recent studies, in combination with the entire body of available evidence, thus form the basis for
23 the ISA determinations of a causal relationship between ambient Os exposure and the occurrence
24 of Os-induced visible foliar injury on sensitive vegetation across the U.S. (U.S. EPA 2013, p. 9-
25 42).
26 With regard to evidence from controlled exposure studies, a recent study of 28 plant
27 species confirmed prior findings of 63 causing predictable diagnostic visible foliar injury
28 symptoms on some species of plants. This study selected 28 plant species, most of which grow
29 naturally throughout the northeast and midwest US, including in national parks and wilderness
30 areas, that were suspected of being Oj, sensitive, and exposed them to four different 63
31 concentrations (30, 60, 90, and 120 ppb) in continuously stirred tank reactor (CSTR) chambers
32 (Kline et al., 2008). Two experiments were conducted in each year of the study (2003 and 2004).
33 Plants were exposed to 63 for 7 hours a day, five days a week over the course of each
34 experiment. Specifically, in 2003, the first experiment lasted from July 14 to August 21and
35 included 29 days of 63 exposure and the second from September 9 to 30 and included 16
36 exposure days. In 2004, the first experiment was conducted from July 13 to August 10 with 21
5-37
-------
1 Os exposure days and the second from August 27 to September 24, including 21 days of Os
2 exposure. Though the exposures were cumulative over the course of the study, exposures were
3 reported only in terms of the target exposure concentration for each experiment. The study
4 reported Os-induced responses in 12, 20, 28 and 28 of the 28 tested species at the 30, 60, 90 and
5 120 ppb exposure concentrations14, respectively. Based on their findings, the authors suggest
6 that American sycamore, aromatic sumac, bee-balm, buttonbush, common milkweed, European
7 dwarf elderberry, New England aster, snowberry and swamp milkweed would make the most
8 useful bioindicator species. Some of these species are native to U.S. locations designated of
9 national significance (discussed further below). The staff additionally concludes that given that
10 the exposure protocol was designed to create a continuous exposure level, not a fluctuating one,
11 this study shows that Cb-induced foliar injury can occur from 7-hour exposures repeated over
12 multiple days at Oj, concentrations that are below the 75 ppb level of the current standard.15
13 While this type of controlled study provides clear evidence of cause and effect, it also has
14 limitations. The authors, recognizing this cautioned that "extrapolation of these CSTR results to
15 the field must be done carefully, since CSTR/greenhouse conditions ... are not representative of
16 natural environmental conditions" (Kline et al., 2008).
17 A string of recently published multi-year field studies provide a complimentary line of
18 field-based evidence by documenting the incidence of visible foliar injury symptoms on a variety
19 of Os-sensitive species over multiple years and across a range of cumulative, seasonal exposure
20 values in several eastern and midwestern Class I national wildlife refuges (NWRs) (U.S. EPA
21 2013, section 9.4.2.1; Davis and Orendovinci 2006; Davis 2007a, b; Davis 2009). Some of these
22 studies also included information regarding soil moisture stress using the Palmer Drought
23 Severity Index (PDSI). While environmental conditions and species varied across the four
24 NWRs, visible foliar injury was documented to a greater or lesser degree at each site. As
25 discussed further below, visible foliar injury incidence in these types of areas has greater
26 significance to the public welfare.
27 • To what extent have important uncertainties identified in the last review been
28 reduced and/or new uncertainties emerged?
29 The studies mentioned above also provide additional information regarding an important
30 uncertainty indentified in the previous review, i.e., the role of soil moisture in influencing visible
14 Two of the target exposure levels, 30 and 60 ppb, fall below the level of the current standard (75 ppb).
The mean exposure concentrations achieved in the CTSRs for the 30 ppb target level for each year and study were
27.9, 26.3, 27.1, and 29.3 ppb and for the 60 ppb target level were 56.6, 55.8, 57.9, and 59.0 ppb, for 2003 study 1,
2003 study 2, 2004 study 1, and 2004 study 2, respectively.
15 The current standard is met when the 3-year average of the 4th highest daily maximum 8-hour average
concentrations is at or below 75 ppb.
5-38
-------
1 foliar injury response (U.S. EPA 2013, section 9.4.2). These studies confirm that adequate soil
2 moisture creates an environment conducive to greater visible foliar injury in the presence of 63
3 than drier conditions. As stated in the ISA, "[a] major modifying factor for Os-induced visible
4 foliar injury is the amount of soil moisture available to a plant during the year that the visible
5 foliar injury is being assessed ... because lack of soil moisture generally decreases stomatal
6 conductance of plants and, therefore, limits the amount of Os entering the leaf that can cause
7 injury" (U.S. EPA, 2013, p. 9-39). As a result, "many studies have shown that dry periods in
8 local areas tend to decrease the incidence and severity of (Vinduced visible foliar injury;
9 therefore, the incidence of visible foliar injury is not always higher in years and areas with higher
10 O3, especially with co-occurring drought (Smith, 2012; Smith et al., 2003)" (U.S. EPA, 2013, p.
11 9-39). This "... partial 'protection' against the effects of Os afforded by drought has been
12 observed in field experiments (Low et al., 2006) and modeled in computer simulations
13 (Broadmeadow and Jackson, 2000)" (U.S. EPA, 2013, p. 9-87). In considering the extent of any
14 protective role of drought conditions, however, the ISA also notes that other studies have shown
15 that "drought may exacerbate the effects of 63 on plants (Pollastrini et al., 2010; Grulke et al.,
16 2003)" and that "[t]here is also some evidence that 63 can predispose plants to drought stress
17 (Maier-Maercker, 1998)". Accordingly, the ISA concludes that "the nature of the response is
18 largely species-specific and will depend to some extent upon the sequence in which the stressors
19 occur" (U.S. EPA, 2013, p. 9-87). Such uncertainties associated with describing the potential for
20 foliar injury and its severity or extent of occurrence for any given air quality scenario due to
21 confounding by soil moisture levels make it difficult to identify an appropriate degree of annual
22 protection (as well as ambient Os exposure conditions that might be expected to provide that
23 protection).
24 • To what extent does the available evidence indicate the occurrence of Os-related
25 visible foliar injury attributable to cumulative exposures at lower ambient Os
26 concentrations than previously established or to exposures at or below the level of
27 the current standard?
28 Recently available evidence confirms that available in previous reviews that visible foliar
29 injury can occur when sensitive plants are exposed to elevated O^ concentrations in a
30 predisposing environment (i.e. adequate soil moisture (U.S. EPA, 2013, section 9.4.2). Recent
31 evidence also continues to indicate the occurrence of visible foliar injury at cumulative ambient
32 Os exposures previously established. Since the 2006 03 AQCD, results from several multi-year
33 field surveys and experimental screenings of (Vinduced visible foliar injury on vegetation also
34 show that visible foliar injury can occur under conditions where the annual 8-hour average Os
35 concentrations are at or below the level of the current standard, as discussed here. Limited
5-39
-------
1 information exists regarding the incidence of visible foliar injury occurring in areas that have
2 design values that meet the current 3-year average 8-hour standard.
3 To facilitate comparison with other studies reporting foliar injury response to W126
4 cumulative exposures, we obtained air quality data from EPA's AQS database for monitors in
5 each study location and calculated the 12-hr W126 values and obtained the maximum 4* high 8-
6 hour average values for a subset of the most recent years included in each study (Table 5-6). As
7 the shaded rows in Table 5-6 below show, in the years 2002/2003 and 2004 in the Cape Romain
8 National Wildlife Refuge in SC, and the Seney National Wildlife Refuge in Michigan,
9 respectively, the 4th highest daily maximum 8-hour average 63 concentrations were at or below
10 the level of the current standard. We additionally note that the Cape Romain site met the current
11 standard of 75 in every 3-year period during the study and has consistently met the standard from
12 2001 to 2012.16 Under these air quality conditions, three species exhibited (Vinduced stipple
13 (winged sumac, Chinese tallow tree, and wild grape). In 2002, 32% of the examined wild grape
14 plants, 20% of the winged sumac plants, and 4.6% of the Chinese tallow tree plants, respectively,
15 were symptomatic (Davis 2009). At the same time, the 12 hour W126 value was 20 ppm-hrs. In
16 2003, when air quality was somewhat improved, foliar injury declined, with only 13.3% of wild
17 grape showing ozone stipple at a maximum 4th high 8-hr of 74 ppb and a W126 of 11 ppm-hr.
18 The PSDI values were 0.27 and 2.45 in 2002 and 2003, respectively. These values show that
19 2003 was a wetter year than 2002, though 2002 would have been considered within the normal
20 soil moisture range.
21 At the Seney NWR site, by comparison, the annual W126 level was similar in 2004 to
22 that at Cape Romain in 2003, and the annual 8-hour average level was below that of the current
23 standard, though the 3-year average design values were above that of the current standard for that
24 year. Not surprisingly, given the lower 63 air quality in 2004, the Seney study reported injury
25 ranging from about 2% on common milkweed to about 6% on spreading dogbane. Though this
26 study does not provide the PDSI values, the authors provided some discussion of a possible
27 relationship stating that "the incidence of ozone injury on spreading dogbane, but not other
28 species, was weakly, but not significantly, related to the drought index (PDSI)... .However this
29 relationship was too weak to be used for predictive purposes." The authors then conclude that
30 "[nevertheless, the threshold SUM06 ozone level needed to induce stipple on sensitive plants
31 within the Seney refuge is likely 5000 ppb-hrs under the environmental conditions of these
32 surveys." On the basis of the above, the staff concludes that these studies confirm that visible
33 foliar injury has been shown to occur in the field at W126 index values ranging down to 10 ppm-
1 ''Design values (concentrations in the form of the standard) for this monitoring site during this period are
presented in the file available at: http://www.epa.gov/airtrends/values.html
5-40
-------
1 hrs and provide limited evidence that such foliar injury can occur in areas with special public
2 welfare significance during periods that meet the current standard.
3 Table 5-6. Visible foliar injury incidence in four National Wildlife Refuges.
Name/ Site #/ Ref .A
Cape Remain NWR, SCI
4501 90046 (Davis, 2009)
Moosehorn NWR, ME/ 230090102
(Davis, 2007a)
Seney NWR, Ml/ 261 530001
(Davis, 2007b)
NWR, NJ/ 34001 0005/
(Davis and Orendovici, 2006)
Year6
2002
2003
2002
2003
2004
2002
2003
2004
2001
2002
2003
AStudies (cited above) reported exposures in terms
calculated exposures in terms of the current 8-hr an
B0nly recent years with available W126 data were ir
4th highest daily maximum
8-hr average
0.075 ppm
0.074 ppm
0.1 ppm
0.083 ppm
0.082 ppm
0.083 ppm
0.076 ppm
0.074 ppm
0.095 ppm
0.092 ppm
0.085 ppm
12hr.W126
20 ppm-hr
11 ppm-hr
24 ppm-hr
22 ppm-hr
14 ppm-hr
11 ppm-hr
15 ppm-hr
10 ppm-hr
39 ppm-hr
53 ppm-hr
36 ppm-hr
% Plants with
visible injury
5-32
3-13
0-17
0-13
3-10
0-13
1-6
2-6
0-45
0-4
0-4
Df SUM06 form. EPA staff, using AQS data for the same monitors,
d W126 forms : http://www.epa.gov/ttn/airs/airsaqs/
eluded in Table.
4
5
6
1
8
9
10
11
12
13
14
15
By far the most extensive field-based dataset of visible foliar injury incidence is that
obtained by of the USDA Forest Service FHM/FIA biomonitoring network program. A trend
analysis of data from the sites located in the Northeast and North Central U.S. for the 16 year
period (1994-2009) (Smith, 2012) provides additional evidence of foliar injury occurrence in the
field as well as some insight into the influence of changes in air quality and soil moisture on
visible foliar injury and the difficulty inherent in predicting foliar injury response under different
air quality/soil moisture scenarios (Smith, 2012; U.S. EPA 2013, section 9.2.4.1). In this study
ambient exposures were expressed in terms of the SUM06 cumulative index coupled with a
measure of the number of peak hourly concentrations above 100 ppb (N100). Soil moisture
conditions were generated using both the Palmer Drought Severity Index (PSDI) and the plant
moisture availability index (MI). Foliar injury was expressed in terms of the biosite index (BI)17.
Biosite index (BI) is the average score (proportion of leaves with injury "amount" x mean severity of
symptoms on injured leaves "severity") for each species averaged across all species on the biosite multiplied by
1,000.
5-41
-------
1 The authors observed that over this 16-year period, "injury indices have fluctuated annually in
2 response to seasonal ozone concentrations and site moisture conditions. Sites with and without
3 injury occur at all ozone exposures but when ambient concentrations are relatively low, the
4 percentage of uninjured sites is much greater than the percentage of injured sites; and regardless
5 of ozone exposure, when drought conditions prevail, the percentage of uninjured sites is much
6 greater than the percentage of injured sites" (Smith, 2012). The authors further note that while
7 "both site moisture and ozone exposure play a role in foliar injury expression ... the interplay
8 among these three factors is unique for each year and possibly each site. Extreme moisture
9 deficits decrease foliar injury, ... [and] ... [i]n no year do high ozone exposures override the
10 controlling effect of site moisture, although at the other end of the scale, injury severity is
11 minimized under conditions of low ozone exposure regardless of site moisture conditions. This
12 implies a necessary threshold of ozone exposure for injury to occur...." "In a similar analysis,
13 Rose and Coulston (2009) reported a high percentage of biosites with injury across the Southern
14 region in 2003, a year when SUM06 values >10 ppm-h were widespread at the same time that
15 the land area was in moisture surplus or balance." Thus, Rose and Coulston (2009) also "found
16 evidence that it is the co-occurrence of sufficient moisture and elevated ozone that determine
17 whether injury occurs to bioindicator plants, not ozone exposure alone." Regarding the role of
18 peak ozone concentrations (>100 ppb 63), Smith (2012) reported that over the 16-year period
19 concentrations above 100 ppb have declined, and that this "... may account for the observed
20 decrease in the severity of ozone-induced foliar injury to ozone sensitive bioindicator plants in
21 eastern forests." They also note that "[tjhere is no compelling evidence, however, that moderate
22 ozone concentrations, as reflected in seasonal mean SUM06 data, are on the decline...." "This
23 may explain why injury continues to be detected on many of the same sites every year..." The
24 authors thus conclude that, "[although it is reasonable to remain concerned about long-term
25 impacts of ozone pollution on our forest ecosystems, the findings of this biomonitoring survey
26 point to a declining risk of probable impact on eastern forests over the 16-year period from 1994
27 to 2009"
28 In a similar assessment of the USDA Forest Service FHM/FIA data in the West, six years
29 (2000 to 2005) of biomonitoring data for 63 injury were evaluated for the three coastal states of
30 California, Oregon and Washington (Campbell et al., 2007; U.S. EPA 2013, section 9.4.2.1).
31 Campbell et al., 2007 found that "... ozone injury occurs frequently (25 to 37 percent of sampled
32 biosites) in California forested ecosystems demonstrating that ozone is present at phytotoxic
33 levels." This study concluded that, "in California, an estimated 1.3 million acres of forest land
34 and 596 million cubic feet of wood are at moderate to high risk to impacts from ozone.
35 However, [m]ore years of data are needed to discern any trends." Though this study does not
36 discuss the role of soil moisture in describing the results, the criteria used to select the
5-42
-------
1 biomonitoring sites include one that considers soil conditions. The best sites are identified as
2 those with low drought potential and good fertility. Thus, given the relatively high 63
3 concentrations that occur in California and the likelihood that many of the biomonitoring sites
4 occur in areas that have sufficient soil moisture, the high percentage of sampled biosites with
5 foliar injury is not unexpected.18
6 These recent studies continue to provide evidence of Cb-induced foliar injury occurring in
7 many areas across the U.S. and augment our understanding of Ch-related visible foliar injury and
8 of factors that influence associations between 63 exposures or concentrations and visible foliar
9 injury such as soil moisture.
10 • To what extent does currently available evidence suggest locations where the
11 vulnerability of sensitive species, ecosystems and/or their associated services to
12 Os-related visible foliar injury would have special significance to the public
13 welfare?
14 As mentioned above, federally designated Class I areas are afforded stringent protections
15 under the 1977 amendments to the Clean Air Act (Act). The Act gives federal land managers of
16 Class I areas "the responsibility to protect all air quality related values (AQRVs).. .from
17 deterioration.... In order to determine if deterioration is occurring, baseline AQRVs must be
18 established" (Davis, 2009). Because of this need and the significance of these areas, studies
19 often focus on these sites. For example, a study by Kohut (2007) was undertaken to assess the
20 risks of (Vinduced visible foliar injury on 63 bioindicators (i.e., (Vsensitive vegetation) in 244
21 parks managed by the NFS. Kohut (2007) estimated Os exposure using hourly O^ monitoring
22 data collected at 35 parks from 1995 to 1999, estimated 63 exposure at 209 additional parks
23 using kriging, a spatial interpolation technique, and qualitatively assessed risk. Kohut (2007)
24 applied a subjective evaluation based on three criteria: (1) the frequency of exceedance of foliar
25 injury thresholds19'20 using several 63 exposure metrics (i.e., SUM06, W126 and N100), (2) the
26 extent that low soil moisture constrains Os uptake during periods of high exposure, and (3) the
27 presence of 63 sensitive species within each park. Based on these criteria, Kohut (2007)
18 Staff additionally notes that a large proportion of O3 monitoring sites in California did not meet the
current standard during the study period (see: http://www.epa.gov/airtrends/values.html).
19 Kohut (2007) uses the term "foliar injury thresholds". In the WREA foliar injury analyses, we use the
term "benchmarks" or "benchmark criteria" to avoid implying that foliar injury could not occur below these levels.
20 Consistent with advice from CAS AC (Frey and Samet, 2012a), the WREA modified the approach used
by Kohut (2007) to apply the W126 metric. The WREA analysis, described in section 5.2.2 below developed and
considered these different W126 benchmarks for foliar injury after further investigation into the benchmarks applied
in Kohut (2007), which were derived from biomass loss rather than visible foliar injury. Kohut cited a threshold of
5.9 ppm-hrs for highly sensitive species from Lefohn (1997), which was based on the lowest W126 estimate
corresponding to a 10% growth loss for black cherry. For soil moisture, Kohut (2007) qualitatively assessed whether
there appeared to be an inverse relationship between soil moisture and high O3 exposure.
5-43
-------
1 concluded that the risk of visible foliar injury was high in 65 parks (27 percent), moderate in 46
2 parks (19 percent), and low in 131 parks (54 percent). Thus, while this study suggests that there
3 may be a reason for concern in as much as 46% of the parks, there were a number of important
4 limitations associated with this study (described in footnotes 8 and 9 below) that weakened this
5 conclusion. Given the importance of this kind of assessment, the WREA used Kohut (2007) as
6 the conceptual basis for the subsequent WREA national-scale and screening-level analyses,
7 though numerous modifications were made to the approach to make it applicable to the context
8 of this Os NAAQS review (see section 5.4.2 below).
9 In addition, as described above, several recently published studies (U.S. EPA 2013,
10 section 9.4.2.1; Davis and Orendovici 2006; Davis 2007a, b; Davis 2009, Kohut 2007) were
11 conducted in federally protected areas including federally designated Class I areas and national
12 parks. These studies confirm that visible foliar injury has been observed in these areas under
13 annual air quality conditions with ambient concentrations at or below the level of the current
14 standard and at W126 index values within the CAS AC range recommended in past reviews.
15 This evidence continues to suggest that Os-sensitive species and their associated ecosystems and
16 services continue to remain vulnerable to visible foliar injury incidence in areas that have been
17 afforded special protection by Congress and that have special significance to the public welfare.
18 5.4.2 Exposure-/Risk-based Considerations
19 The WREA presents a number of analyses considering air quality conditions associated
20 with increased prevalence of visible foliar injury and potential associated welfare impacts (see
21 Table 5-7 below, U.S. EPA, 2014, Chapter 7). An initial analysis included the development of
22 benchmark criteria reflecting different prevalences of visible foliar injury or of the occurrence of
23 elevated injury in conjunction with different W126 exposures and in some cases, soil moisture
24 conditions. These criteria were then used in a national scale screening level assessment to
25 characterize potential risk of foliar injury incidence under 2006-2010 conditions in 214 national
26 parks. The last analysis was a case study assessment on three national parks, which also
27 provides limited characterization of the associated ecosystem services. Despite the limitations
28 and uncertainties associated with these analyses, and recognizing that the air quality conditions
29 in most cases (prior to any model adjustment) did not meet the current standard, staff believes
30 that they help inform our understanding of the relationship between soil moisture and foliar
31 injury incidence, as well as provide limited support for our conclusion of a risk of visible foliar
32 injury incidence under air quality conditions likely to meet the current standard in areas of
33 special significance to the public welfare.
5-44
-------
1 Table 5-7. Exposure, risk and ecosystem services analyses related to visible foliar injury.
WREA
estimates
Ecosystem Level Effects
Proportion of USFS biosites with
foliar injury incidence, degree of
injury, and soil moisture at various
W126-based benchmarks
Percent of 214 national parks
exceeding criteria based on USFS
biosite dataset analysis during years
considered
Ecosystem Services
Case study of 3 national parks characterized impacts
using available visitor and use data, including monetary
data for activities and visitor expenditures:
• Utilized Willingness-to-Pay studies for scenic
impairment;
• Assessed the overall cover of sensitive species;
• Compared sensitive species cover to trails and
overlooks; and,
• Estimated percent of park area with Oa concentrations
above different W126 index values averaged over
three consecutive years.
B The screening-level assessment of 214 national parks additionally included observations based on the model-
based adjustments to just meet the current standard and targets for the three W126 scenarios (discussed below) but
did not conduct a full analysis using these data.
3 • For what air quality scenarios were exposures and risks estimated? What
4 approaches were used to estimate W126 exposures for those conditions? What
5 are associated limitations and uncertainties?
6 Three types of foliar injury analyses were performed in the WREA and are considered
7 below. They include an analysis using U.S. Forest Service biosite data, a National Park
8 screening-level assessment and a National Park case study (focused on three parks). The
9 analysis of U.S. Forest Service (USFS) biosite data was done using 03 concentrations estimated
10 for a national-scale surface of concentrations (at a 12 km grid cell resolution in contiguous U.S.)
11 using interpolation methodology applied to concentrations at Os monitor locations (U.S. EPA,
12 2014, section 4.3.2, Appendix 4A). The analysis of USFS data used surfaces for each year from
13 2006 through 2010 (U.S. EPA, 2014, Appendix 4A, section 4.2). In both the National Park
14 screening-level assessment and the case study analysis, observations related to air quality were
15 made for five air quality scenarios by the methodology summarized in Table 5-4 above.21
16 The W126 index values in the individual years from 2006 to 2010 ranged from less than
17 5 ppm-hrs to 25 ppm-hrs. Concentration estimates varied appreciably across the five years with
In general, this methodology involved two steps. The first is derivation of the average W126 value
(across the three years) at each monitor location. This value is based on unadjusted data for recent conditions and
model-adjusted concentrations for the 4 other scenarios. The development of model adjusted concentrations was
done for each of 9 regions independently (see U.S. EPA, 2014, section 4.3.4.1). In the second step, national-scale
spatial surfaces (W126 values for each model grid cell) were created using the monitor-location values and the
Voronoi Neighbor Averaging (VNA) spatial interpolation technique (details on the VNA technique are presented in
U.S. EPA, 2014 Appendix 4A).
5-45
-------
1 the median index values across grid cells ranging from a low of 5.5 ppm-hrs in 2009 to 11 ppm-
2 hrs in 2006 (U.S. EPA, 2014, Appendix 4A, section 4.2). During the recent conditions period
3 (2006 through 2008), the average W126 index values (across the three-year recent conditions
4 period) at the monitor locations ranged from below 10 ppm-hrs to 48.6 ppm-hrs (U.S. EPA 2014,
5 Figure 4-4 and Table 4-1). After model-adjustment of the 2006-2008 data to just meet the
6 current standard in each region, and subsequent application of the VNA technique to the current
7 standard scenario monitor location values, the average W126 values were below 15 ppm-hrs
8 across the national surface with the exception of a very small area of the southwest region (near
9 Phoenix) where the average W126 values was near or just above 15 ppm-hrs. A lowering of the
10 highest values occurred with application of the interpolation method as a result of estimating
9
11 W126 values at a 12x12 km grid resolution rather that at the exact location of a monitor. This
12 indicates one uncertainty associated with this aspect of the approach to estimating W126 values
13 for the model-adjusted air quality just meeting the current standard. Other uncertainties are
14 summarized in section 5.2.2 above.
15 • What are the nature and magnitude of the cumulative exposure- and risk-related
16 estimates for visible foliar injury under recent conditions or conditions meeting
17 the current Os standard?
18 As an initial matter, we consider the analysis of the biomonitoring site data from the
19 USDA Forest Service FHM/FIA Network, described in section 7.2 of the WREA.22 Using this
20 dataset and associated data for soil moisture conditions during the sample years along with
21 ambient air 63 concentrations based on monitoring data from 2006 to 2010 and spatial
22 interpolation methodology (as described above), the proportion of sites with any or elevated
23 foliar injury are observed to increase with increasing annual W126 index values up to specific
24 values after which there is little change in proportion of affected sites with higher W126 values
25 (see Figure 5-5 below; U.S. EPA, 2014, section 7.2). The proportion of sites metric is derived by
26 first ordering the data (across sites and sample years) by W126 index value estimated for that site
27 and year. Then for each W126 index value the proportion of sites exceeding the selected biosite
28 index value for all observations at or below that W126 index value is calculated. The WREA
29 repeated this using a biosite index value greater than zero, indicating presence of any foliar
30 injury, and an index value >5, corresponding to a USFS cutoff for "low" or more injury (USFS,
31 2011).
32 When looking only at presence or absence of foliar injury ("any injury") with the
33 exception of 2008, the proportion of sites across all W126 index values with foliar injury exceeds
22 Data were not available for several western states (Montana, Idaho, Wyoming, Nevada, Utah, Colorado,
Arizona, New Mexico, Oklahoma, and portions of Texas).
5-46
-------
1 15 percent; in 2006, it exceeds 20 percent, while in 2008 the proportion of sites with foliar injury
2 across all W126 index values was just below 15 percent (U.S.EPA, 2014, section 7.2.3, Figure 7-
3 8). When categorized by moisture levels, the data demonstrate a distinct pattern. In general, the
4 WREA concludes that the results of these foliar injury analyses demonstrate a similar pattern -
5 the proportion of sites showing the presence of any foliar injury (biosite>0) or at least little foliar
6 injury (biosite>5) increases from zero to about 20% and 6% respectively (Figure 5-5 below).
7 This increase occurs with increasing W126 index values up to approximately 10 ppm-hrs for any
8 foliar injury (biosite index >0), with little change in proportion of sites with any injury at higher
9 W126 values. The data for sites during normal moisture years are very similar to the dataset as a
10 whole, with an overall proportion of close to 18 percent for presence of any foliar injury, and
11 close to 6 percent for sites exceeding a biosite index of 5. Among the sites with relatively wet
12 season (average Palmer Z => 1), the highest proportion of sites observed is much higher for both
13 index categories of injury and the relationship with annual W126 index value is much steeper.
14 Much lower proportions of sites are reached for both injury categories at sites with relatively dry
15 seasons (average Palmer Z < -1.24), potentially indicating that drought may provide some
16 protection from foliar injury as discussed in the ISA (U.S. EPA, 2014, section 7.2.3, Figures 7-10
17 and 7-11). This information provides insight into the relationship between soil moisture and
18 foliar injury and the issue of whether drought provides protection from foliar injury. For both
19 categories of injury, there is relatively little change in the proportion of sites beyond a W126 of
20 20 ppm-hrs. There are two important observations that can be made from these analyses: (1) the
21 proportion of sites exhibiting foliar injury rises rapidly at increasing W126 index values below
22 10 ppm-hrs, and (2) there is relatively little change in the proportions above W126 index values
23 of 20 ppm-hrs.
5-47
-------
Biosite Index > 0
2
Q_
O
2
Q_
BO 00
10
20 30
W126(ppm-hrs)
Biosite Index > 5
40
10
20 30
W126(ppm-hrs)
40
3 Figure 5-5. Cumulative proportion of sites with a) any foliar injury or b) elevated injury
4 present, by moisture category (U.S. EPA 2014, Figures 7-10 and 7-11).
5 We additionally consider the WREA screening-level assessment in 214 parks in the
6 contiguous U.S. that employed benchmark criteria developed from the above analysis (Table 5-
5-48
-------
9^ 9 A
1 8). , For example, annual Oj, concentrations corresponding to a W126 seasonal index value of
2 10.46 ppm-hrs represents the 63 exposure concentration above which a consistent percentage of
3 all biosites (17.7 percent) showed any injury, without consideration of soil moisture. The W126
4 benchmarks across the six scenarios range from 3.05 ppm-hrs (five percent of biosites, normal
5 moisture, any injury) up to 46.87 ppm-hrs (five percent of biosites, dry, elevated injury). For the
6 scenario of 10% biosites with injury, W126 values were approximately 4, 6, and 25 ppm-hours
7 for wet, normal and dry years, respectively. The national-scale screening-level assessment
8 includes 42 parks with 63 monitors and 214 parks with 63 exposure estimated from the
9 interpolated O3 surface for individual years from 2006 to 2010 (U.S. EPA, 2014, Appendix 7A).
10 These data were combined with lists from the NFS of the parks containing Os sensitive
11 vegetation species (NFS, 2003, 2006). Based on NFS lists, 95 percent of the parks in this
12 assessment contain at least one Os-sensitive species. This analysis for recent air quality
13 conditions, estimates that 58 percent of parks exceeded the benchmark criteria corresponding to
14 the base scenario (W126>10.46 ppm-hrs, 17.7 percent of biosites, all moisture categories, any
15 injury) for at least three years in the period from 2006 to 2010, and 34 percent of parks would
16 exceed the benchmark criteria for the elevated injury scenario (five percent of biosites, multiple
17 moisture categories, elevated foliar injury) for at least three years (U.S. EPA, 2014, section
18 7.3.2).25 Based on model-adjustments to meet the current standard, none of the 214 parks have
19 average W126 index values that would exceed the annual benchmark criteria for the base
20 scenario (W126>10.46 ppm-hrs).
21
23 The parks assessed here include lands managed by the NFS in the continental U.S., which includes
National Parks, Monuments, Seashores, Scenic Rivers, Historic Parks, Battlefields, Reservations, Recreation Areas,
Memorials, Parkways, Military Parks, Preserves, and Scenic Trails.
24 The WREA applied different foliar injury benchmarks in this assessment after further investigation into
the benchmarks applied in Kohut (2007), which were derived from biomass loss rather than visible foliar injury.
Kohut cited a threshold of 5.9 ppm-hrs for highly sensitive species from Lefohn (1997), which was based on the
lowest W126 estimate corresponding to a 10% growth loss for black cherry. For soil moisture, Kohut (2007)
qualitatively assessed whether there appeared to be an inverse relationship between soil moisture and high O3
exposure.
25 The lack of national surfaces of O3 concentrations for each of three years in each scenario (as described
earlier and summarized in Table 5-3) precluded derivation of similar estimates (of parks with W126 index values
above any of the benchmark criteriaafter model-adjustment of air quality to just meet the current standard, although
we note that the W126 averaged across three years in all 214 parks was below 10.46 ppm-hrs (the individual year
benchmark criteria for the base scenario) (U.S. EPA, 2014, section 7.3.2).
5-49
-------
1 Table 5-8. Benchmark criteria for Os exposure and relative soil moisture used in
2 screening-level assessment of parks (from U.S. EPA 2014, Table 7-5).
Description
Any injury, without
considering soil
moisture conditions
(Base Scenario)
Any injury,
considering soil
moisture conditions
5% of biosites,
Injury s 5
17.7% of all biosites with any foliar
injury (consistent proportion of all
biosites showed injury in years with
W126 index value above this value)
5% of biosites showing any foliar
injury, by relative moisture category
10% of biosites showing any foliar
injury, by relative moisture category
15% of biosites showing any foliar
injury, by relative moisture category
20% of biosites showing any foliar
injury, by relative moisture category
5% of biosites showing foliar injury
equal or greater than 5 on the biosite
injury index (e.g., 5% of leaf shows
injury in 10% of the leaves), by
relative moisture category
Normal
(-1.251)
Dry
(Palmer Z<-1)
W126>10.46
(soil moisture not considered)
W126>3.05
W126>5.94
W126>8.18
N/A
W126>12.23
W126>3.76
W126>4.42
W126>4.69
W126>5.65
W126>7.02
W126>6.16
W126>24.61
N/A
N/A
W126>46.87
4 Lastly, we consider the WREA case study analysis which focused on characterizing the
5 ecosystem services potentially associated with visible foliar injury in three specific national
6 parks (case study assessment). The parks included were Great Smoky Mountains National Park
7 (GRSM), Rocky Mountain National Park (ROMO), and Sequoia/Kings National Parks (SEKI).
8 For each park, the potential impact of (Vrelated foliar injury on recreation (cultural services)
9 was considered in light of information on visitation patterns, recreational activities and visitor
10 expenditures. For example, visitor spending in 2011 exceeded $800 million, $170 million and
11 $97 million dollars in GRSM, ROMO and SEKI, respectively. This assessment also included
12 percent cover of species sensitive to foliar injury and focused on the overlap between recreation
13 areas within the park and elevated W126 concentrations. Ozone concentrations in GRSM have
14 been among the highest in the eastern U.S. In the recent conditions scenario, the grid cells
15 representing 44 percent of GRSM had three year average W126 value above 15 ppm-hrs. After
16 adjustments to just meet the current standard, no grid cell had a three-year average W126 value
17 above 7 ppm-hrs. In the recent conditions scenario for ROMO, three-year average W126 values
18 for all grid cells were abovelS ppm-hrs. In the current standard scenario, values for 59 percent
5-50
-------
1 of the park were below 7 ppm-hrs. For SEKI, three-year average W126 values for all grid cells
2 were above 15 ppm-hrs in the recent conditions scenario, but dropped below 7 ppm-hrs for the
3 current standard scenario (U.S. EPA, 2014, section 7.4).
4 In summary, these analyses indicate that O?, concentrations in U.S. national parks in
5 recent years correspond to W126 index values at which some foliar injury may occur, with
6 variation associated with relative soil moisture conditions. None of the 214 parks assessed are
7 estimated to exceed the annual benchmark criteria for the base scenario (W126 >10.46 ppm-hrs)
8 after adjusting air quality to meet the current standard. Although model-adjusted scenarios to
9 just meet the current standard indicate substantial reductions in three-year average W126 index
10 values estimated by the VNA approach, some individual year values may range higher. The case
11 study analysis of three parks indicates the potential for appreciable ecosystem services impact
12 associated with foliar injury, although it might be expected that such impact would relate more to
13 severe and/or widespread foliar injury occurrences. While these analysis indicate the potential
14 for foliar injury to occur under conditions that meet the current standard, the extent of foliar
15 injury that might be expected under such conditions is unclear from these analyses.
16 • To what extent are the exposures and risks remaining upon simulating just
17 meeting the current Os standard important from a public welfare perspective?
18 The screening level assessment, as described above, indicates that risk of visible foliar
19 injury is likely to be lower in some national parks after simulating just meeting the current
20 standard. Based on the national-scale analysis, visible foliar injury would likely continue to
21 occur at lower Oj, exposures, including some sensitive species growing in areas (e.g., National
22 Parks and other Class I areas) that may provide important cultural ecosystem services to the
23 public. Staff notes that such occurrences might reasonably be considered to have some
24 importance from a public welfare perspective, as discussed in section 5.1 above.
25 • What are the ecosystem services potentially affected by visible foliar injury, to
26 what degree can the magnitude of these effects be qualitatively or quantitatively
27 characterized, and to what extent are they important from a public welfare
28 perspective?
29 The ecosystem services most likely to be affected by Cb-induced foliar injury are cultural
30 services, including aesthetic value and outdoor recreation. Aesthetic value and outdoor
31 recreation depend on the perceived scenic beauty of the environment. Many outdoor recreation
32 activities directly depend on the scenic value of the area, in particular scenic viewing, wildlife-
33 watching, hiking, and camping. These activities and services are of significant importance to
34 public welfare as they are enjoyed by millions of Americans every year and generate millions of
35 dollars in economic value (U.S. EPA, 2014, Chapter 5, Chapter 7). These aesthetic values are at
5-51
-------
1 risk of impairment because of Os-induced damage directly due to foliar injury. Other ecosystem
2 services that have also been found to be associated with Os-sensitive plants include those that fall
3 under the categories of provisioning. For example, several tribes have indicated that many of the
4 known confirmed Os-sensitive species (including bioindicator species) are culturally significant
5 (see Appendix 5A). Although data are not available to explicitly quantify these negative effects
6 on ecosystem services, several qualitative analyses conducted in the WREA are summarized
7 below.
8 To assess the effects of visible foliar injury on recreation, the WREA reviewed the
9 National Survey on Recreation and the Environment (NSRE), as well as the 2006 National
10 Survey of Fishing, Hunting, and Wildlife-Associated Recreation (FHWAR) and a 2006 analysis
11 done for the Outdoor Industry Foundation (OIF). According to the NSRE, some of the most
12 popular outdoor activities are walking, including day hiking and backpacking; camping; bird
13 watching; wildlife watching; and nature viewing. Participant satisfaction with these activities
14 can depend on the quality of the natural scenery, which can be adversely affected by Os-related
15 visible foliar injury. According to the FHWAR and the OIF reports, the total expenditures across
16 wildlife watching activities, trail-based activities, and camp-based activities are approximately
17 $230 billion dollars annually. While the WREA could not quantify the magnitude of the impacts
18 of Os damage to the scenic beauty and outdoor recreation, the existing losses associated with
19 current Os-related foliar injury are reflected in reduced outdoor recreation expenditures (U.S.
20 EPA, 2014, section 7.1).
21 The WREA also assessed Os impacts on cultural ecosystem services related to foliar
22 injury at three national parks - Great Smoky Mountains National Park, Rocky Mountain
23 National Park, and Sequoia/Kings National Parks - by considering information on visitation
24 patterns, recreational activities and visitor expenditures. The analysis included percent cover of
25 species sensitive to foliar injury and focused on the overlap between recreation areas within the
26 park and elevated W126 concentrations. All three of these park units are in areas that are known
27 to have high ambient Os concentrations, have vegetation maps, and have species that are
28 considered Os sensitive. Using GIS, the NFS vegetation maps were compared to the national Os
29 surface to illustrate where foliar injury may be occurring, particularly with respect to park
30 amenities such as trails (U.S. EPA, 2014, section 7.4).
31 Great Smoky Mountains National Park is prized, in part, for its rich species diversity.
32 The large mix of species includes 37 Os-sensitive species and many areas contain several
33 sensitive species. With 3.8 million hikers using the trails every year and those hikers willing to
34 pay (WTP) over $266 million for that activity, even a small benefit of reducing Os damage in the
35 park could result in a significant economic value. Ozone concentrations in Great Smoky
36 Mountains National Park have been among the highest in the eastern U.S. - at times twice as
5-52
-------
1 high as neighboring cities such as Atlanta (U.S. EPA, 2014, p. 7-52). Unlike Great Smoky
2 Mountains National Park, sensitive species cover in Rocky Mountain National Park is driven by
3 a few Os-sensitive species (7 species) and most notably by Quaking Aspen. This is significant in
4 that many of the visitors to Rocky Mountain National Park visit specifically to see this tree in its
5 fall foliage. Given 1.5 million hikers in Rocky Mountain National Park and their $70 million
6 WTP for the hiking experience, even a small improvement in the scenic value could be
7 economically significant (U.S. EPA, 2014, section 7.4.2, p. 7-56). Sequoia/Kings National Parks
8 is home to 12 identified sensitive species. Again, although the EPA is not able to quantify the
9 impact of this scenic damage on hiker satisfaction for hikers in Sequoia/Kings National Parks
10 and their $26 million WTP for the experience, even a small improvement in the scenic value
11 could be economically significant ((U.S. EPA, 2014, section 7.4.3, p. 7-63).
12 • What are the uncertainties associated with this information and what is the level
13 of confidence associated with those estimates?
14 Uncertainties associated with these analyses are discussed in the WREA, sections 7.3.2
15 and 8.5.3, and in WREA Table 7.23. As discussed in the WREA (section 8.5.3), evaluating soil
16 moisture is more subjective than evaluating Oj, exposure because of its high spatial and temporal
17 variability within the Os season, and there is considerable subjectivity in the categorization of
18 relative drought. The WREA generally concludes that the spatial and temporal resolution for the
19 soil moisture data is likely to underestimate the potential of foliar injury that could occur in some
20 areas. In addition, there is lack of a clear threshold for drought below which visible foliar injury
21 would not occur. In general, low soil moisture reduces the potential for foliar injury, but injury
22 could still occur, and the degree of drought necessary to reduce potential injury is not clear. Due
23 to the absence of biosite injury data in the Southwest region and limited biosite data in the West
24 and West North Central regions, the benchmarks applied may not be applicable to these regions.
25 There are also important uncertainties in the estimated Os concentrations for the different
26 air quality scenarios evaluated (U.S. EPA, 2014, section 8.5), as discussed earlier in this section.
27 In general, this interpolation method under-predicts higher 12-hr W126 exposures. Due to the
28 important influence of higher exposures in determining risks to plants, the potential for the VNA
29 interpolation approach to under-predict higher W126 exposures could result in an under-
30 estimation of risks to vegetation in some areas. The WREA applied the benchmarks from the
31 national-scale analysis to a screening-level analysis of 214 national parks and case studies of
32 three national parks. Therefore, uncertainties in the foliar injury benchmarks and in the air
33 quality analyses are propagated into the national park analyses. The uncertainties associated
34 with air quality assessments include those resulting from use of an unevenly distributed
35 monitoring network with fewer monitors in rural and western sites to drive a VNA interpolation
5-53
-------
1 and the use of model-adjusted air quality to generate alternative air quality scenarios including
2 just meeting the current standard (these uncertainties are described in more detail above).
3 Additional uncertainties are associated with the national park analyses. Specifically,
4 survey estimates of participation rates, visitor spending/economic impacts, and willingness-to-
5 pay are inherently uncertain. These surveys potentially double-count impacts based on the
6 allocation of expenditures across activities but also potentially exclude other activities with
7 economic value. Each survey uses different survey methods, so it is not appropriate to generalize
8 across the surveys. In general, the national level surveys apply standard approaches, which
9 minimize potential bias. Other sources of uncertainty are associated with the mapping, including
10 park boundaries, vegetation species cover, and park amenities, such as scenic overlooks and
11 trails. In general, the WREA concludes that there is high confidence in the park mapping (U.S.
12 EPA, 2014, Table 7-23)..
13 5.5 OTHER WELFARE EFFECTS
14 In addition to the welfare effects discussed in the previous sections, there is evidence of
15 other Oj, effects, such as those related to climate impacts that we consider here. In this section,
16 the WREA national-scale analyses of the effects of insect damage to forests related to elevated
17 Os exposures are considered in section 5.5.1, and a case study-scale characterization of the effect
18 community composition changes on forest susceptibility and fire regulation in California is
19 considered in section 5.5.2. As above, these sections, where possible, consider the WREA
20 information regarding risk remaining under model adjusted conditions just meeting the current
21 standard and associated uncertainties (U.S. EPA 2014, section 8.5). Chapters 5, 6, and 7 of the
22 WREA also qualitatively assessed additional ecosystem services, including regulating services
23 such as hydrologic cycle and pollination; provisioning services such as commercial non-timber
24 forest products; and cultural services with aesthetic and non-use values. The information
25 associated with these latter effects is insufficient to inform the target protection of the standard.
26 The effects of Os on climate are also considered in section 5.5.3 below, drawing primarily on the
27 evidence presented in the ISA (U.S. EPA 2013, chapter 10).
28 5.5.1 Forest Susceptibility to Insect Infestation
29 Ozone in ambient air can contribute to increased susceptibility of some forests to
30 infestation by some chewing insects, including the southern pine beetle and western bark beetle
31 (U.S. EPA 2013, chapter 9; U.S. EPA 2014, sections 5.3.3 and 5.4). These infestations can cause
32 economically significant damage to tree stands and the associated timber production. The
33 WREA described the potential impacts of this effect on timber markets (U.S. EPA 2014, section
34 5.4). In the short-term, the immediate increase in timber supply that results from the additional
5-54
-------
1 harvesting of damaged timber depresses prices for timber and benefits consumers. In the longer-
2 term, the decrease in timber available for harvest raises timber prices, potentially benefitting
3 producers. The United States Forest Service (USFS) reports timber producers have incurred
4 losses of about $1.4 billion (2010$), and wood-using firms have gained about $966 million, due
5 to beetle outbreaks between 1977 to 2004. It is not possible to attribute a portion of these
6 impacts resulting from the effect of Os on trees' susceptibility to insect attack; however, the
7 losses are embedded in the estimates cited.
8 To provide some quantitative estimates related to insect infestation-related risks, the
9 WREA reported the estimates of 3-year average W126 values in areas estimated to be at risk of
10 greater than 25% timber loss (high loss) due to pine beetle infestation. This was done for all six
11 WREA air quality scenarios. For example, for the recent conditions scenario, approximately 57
12 percent of the at-risk area has W126 estimates above 15 ppm-hrs, with the percentage dropping
13 to approximately five percent in the current standard scenario (U.S. EPA 2014, section 5.4).
14 5.5.2 Fire Regulation
15 Evidence indicates that fire regime regulation may also be negatively affected by Os
16 exposure (U.S. EPA 2013, chapter 9; U.S. EPA 2014, section 5.3.3). For example, Grulke et al.
17 (2008) reported various lines of evidence indicating that 63 exposure may contribute to southern
18 California forest susceptibility to wildfires by increasing leaf turnover rates and litter, increasing
19 fuel loads on the forest floor. According to the National Interagency Fire Center, in the U.S. in
20 2010 over 3 million acres burned in wildland fires and an additional 2 million acres were burned
21 in prescribed fires. From 2004 to 2008, Southern California alone experienced, on average, over
22 4,000 fires per year burning, on average, over 400,000 acres per fire. The California Department
23 of Forestry and Fire Protection (CAL FIRE) estimated that losses to homes due to wildfire were
24 over $250 million in 2007 (CAL FIRE, 2008). In 2008, CAL FIRE's costs for fire suppression
25 activities were nearly $300 million (CAL FIRE, 2008).
26 The WREA developed maps that overlay the mixed conifer forest area of California with
27 areas of moderate or high fire risk defined by CAL FIRE and with recent W126 concentrations
28 and surfaces adjusted to just meet existing and alternative standards. The highest fire risk and
29 highest 63 concentrations overlap with each other, as well as with significant portions of mixed
30 conifer forest. In the recent concentrations scenario, over 97 percent of mixed conifer forest area
31 has average W126 values over 7 ppm-hrs with a moderate to severe fire risk, and 74 percent has
32 average W126 values over 15 ppm-hrs with a moderate to severe fire risk. The scenario for air
33 quality adjusted to just meet the current standard, almost all of the mixed conifer forest area with
34 a moderate to high fire risk shows a reduction in Os to below a W126 index value of 7 ppm-hrs
35 (average across three years of scenario). In the scenario for an average W126 index value of 15
5-55
-------
1 ppm-hrs, all but 0.18 percent of the area has average index values below 7 ppm-hrs, and for the
2 W126 scenarios of 11 and 7 ppm-hrs, all of the moderate to high fire threat area has estimated
3 average W126 values below 7 ppm-hrs (U.S. EPA 2014, section 5.3.3, figure 5-3).
4 5.5.3 Ozone Effects on Climate
5 Tropospheric O3 is a major greenhouse gas, third in importance after carbon dioxide
6 (CC>2) and methane (CH4). While the developed world has successfully reduced emissions of O3
7 precursors in recent decades, many developing countries have experienced large increases in
8 precursor emissions and these trends are expected to continue, at least in the near term (U.S.
9 EPA 2013, section 10.3.6.2). Projections of radiative forcing due to changing O3 over the 21st
10 century show wide variation, due in large part to the uncertainty of future emissions of source
11 gases (U.S. EPA 2013, section 10.3.6.2). In the near-term (2000-2030), projections of O3
r\
12 radiative forcing range from near zero to +0.3 W/m , depending on the emissions scenario (U.S.
13 EPA 2013, section 10.3.6.2; Stevenson et al., 2006). Reduction of tropospheric O3
14 concentrations could therefore provide an important means to slow climate change in addition to
15 the added benefit of improving surface air quality (U.S. EPA, 2013, section 10.5).
16 It is clear that increases in tropospheric O3 lead to warming. However the precursors of
17 O3 also have competing effects on the greenhouse gas CH4, complicating emissions reduction
18 strategies. A decrease in CO or VOC emissions would enhance OH concentrations, shortening
19 the lifetime of CH4, while a decrease in NOx emissions could depress OH concentrations in
20 certain regions and lengthen the CH4lifetime (U.S. EPA, 2013, section 10.5).
21 Abatement of CH4 emissions would likely provide the most straightforward means to
22 address O3-related climate change since CH4 is itself an important precursor of background O3
23 (West et al., 2007; West et al., 2006; Fiore et al., 2002). A reduction of CH4 emissions would
24 also improve air quality in its own right. A set of global abatement measures identified by West
25 and Fiore (2005) could reduce CH4 emissions by 10% at a cost savings, decrease background O3
26 by about 1 ppb in the Northern Hemisphere summer, and lead to a global net cooling of 0.12
27 W/m2. West et al. (2007) explored further the benefits of CH4 abatement, finding that a 20%
28 reduction in global CH4 emissions would lead to greater cooling per unit reduction in surface O3,
29 compared to 20% reductions in VOCs or CO (U.S. EPA, 2013, section 10.5).
30 Important uncertainties remain regarding the effect of tropospheric O3 on future climate
31 change. To address these uncertainties, further research is needed to: (1) improve knowledge of
32 the natural atmosphere; (2) interpret observed trends of O3 in the free troposphere and remote
33 regions; (3) improve understanding of the CH4 budget, especially emissions from wetlands and
34 agricultural sources, (4) understand the relationship between regional O3 radiative forcing and
5-56
-------
1 regional climate change; and (5) determine the optimal mix of emissions reductions that would
2 act to limit future climate change (U.S. EPA, 2013, section 10.5).
3 The IPCC has estimated the effect of the tropospheric 63 change since preindustrial times
4 on climate has been estimated to be about 25-40% of the anthropogenic CC>2 effect and about
5 75% of the anthropogenic CTLj effect. There are large uncertainties in the radiative forcing
6 estimate attributed to tropospheric 63, making the effect of tropospheric 63 on climate more
7 uncertain than the effect of the long-lived greenhouse gases (U.S. EPA, 2013, section 10.5).
8 Radiative forcing does not take into account the climate feedbacks that could amplify or
9 dampen the actual surface temperature response. Quantifying the change in surface temperature
10 requires a complex climate simulation in which all important feedbacks and interactions are
11 accounted for. As these processes are not well understood or easily modeled, the surface
12 temperature response to a given radiative forcing is highly uncertain and can vary greatly among
13 models and from region to region within the same model (U.S. EPA, 2013, section 10.5).
14 5.5.4 Additional Effects
15 Recent information available since the last review considers the effects of Os on chemical
16 signaling in insect and wildlife interactions. Specifically, studies on 63 effects on pollination and
17 seed dispersal, defenses against herbivory and predator-prey interactions all consider the ability
18 of 63 to alter the chemical signature of VOCs emitted during these pheromone-mediated events.
19 The effects of Os on chemical signaling between plants, herbivores and pollinators as well as
20 interactions between multiple trophic levels is an emerging area of study that may result in
21 further elucidation of 63 effects at the species, community and ecosystem-level (U.S. EPA, 2013,
22 p. 9-98).
23 5.6 CASAC ADVICE
24 Following the 2008 decision to revise the secondary standard by setting it identical to the
25 revised primary standard, CASAC conveyed additional advice to the Administrator regarding
26 that decision. Shortly after that, several petitioners filed suit challenging the decision and in
27 September 2009, the EPA announced its intention to reconsider the 2008 standards, issuing a
28 notice of proposed rulemaking in January 2010 (FR 75 2938). Soon after, the EPA solicited
29 CASAC review of that proposed rule and in January 2011 solicited additional advice. This
30 proposal was based on the scientific and technical record from the 2008 rulemaking, including
31 public comments and CASAC advice and recommendations. As further described in section
32 1.2.2 above, the EPA in the fall of 2011 did not promulgate final rulemaking in that process but
33 decided to coordinate further proceedings on the reconsideration rulemaking with this ongoing
5-57
-------
1 periodic review. Accordingly, we are describing CASAC advice related to the 2008 final
2 decision and the subsequent reconsideration in this section, as well as advice on the NAAQS
3 review that was initiated in September 2008.
4 In April 2008, the members of the CASAC Ozone Review Panel sent a letter to EPA
5 stating that "[i]n our most-recent letters to you on this subject - dated October 2006 and March
6 2007 - ... the Committee recommended an alternative secondary standard of cumulative form
7 that is substantially different from the primary Ozone NAAQS in averaging time, level and form
8 — specifically, the W126 index within the range of 7 to 15 ppm-hours, accumulated over at least
9 the 12 'daylight' hours and the three maximum ozone months of the summer growing season "
10 (Henderson, 2008). The letter continued:
11 The CASAC now wishes to convey, by means of this letter, its additional,
12 unsolicited advice with regard to the primary and secondary Ozone NAAQS. In
13 doing so, the participating members of the CASAC Ozone Review Panel are
14 unanimous in strongly urging you or your successor as EPA Administrator to
15 ensure that these recommendations be considered during the next review cycle for
16 the Ozone NAAQS that will begin next year ... The CASAC was also greatly
17 disappointed that you failed to change the form of the secondary standard to
18 make it different from the primary standard. As stated in the preamble to the
19 Final Rule, even in the previous 1996 ozone review, "there was general
20 agreement between the EPA staff, CASAC, and the Administrator, ... that a
21 cumulative, seasonal form was more biologically relevant than the previous 1-
22 hour and new 8-hour average forms (61 FR 65716) "for the secondary
23 standard. Unfortunately, this scientifically-sound approach of using a
24 cumulative exposure index for welfare effects was not adopted...
25 In response to the EPA's solicitation of their advice on the Agency's proposed
26 rulemaking as part of the reconsideration, CASAC conveyed their support as follows (Samet,
27 2010).
28 CASAC also supports EPA 's secondary ozone standard as proposed: a new
29 cumulative, seasonal standard expressed as an annual index of the sum of
30 weighted hourly concentrations (i.e., the Wl26form), cumulated over 12 hours
31 per day (Sam to 8pm) during the consecutive 3-month period within the ozone
32 season with the maximum index value, set as a level within the range of 7 to [1]5
33 ppm-hours. This W126 metric can be supported as an appropriate option for
34 relating ozone exposure to vegetation responses, such as visible foliar injury and
3 5 reductions in plant grow th. We found the Agency's reasoning, as stated in the
36 Federal Register notice of January 19, 2010, to be supported by the extensive
37 scientific evidence considered in the last review cycle. In choosing the Wl26 form
38 for the secondary standard, the Agency acknowledges the distinction between the
39 effects of acute exposures to ozone on human health and the effects of chronic
40 ozone exposures on welfare, namely that vegetation effects are more dependent on
5-58
-------
1 the cumulative exposure to, and uptake of, ozone over the course of the entire
2 growing season (defined to be a minimum of at least three months).
3 In advice offered so far in the current review, which is considering an updated scientific
4 and technical record since the 2008 rulemaking, CASAC indicated that a conclusion that the
5 current standard is inadequate to protect vegetation and ecosystems is "warranted" although the
6 foundation needs to be broadened beyond analysis focused on Class I areas and trees to include
7 "effects on sensitive crops, trees in regions outside of Class I areas, and additional ecosystem
8 impacts" (Frey and Samet, 2012b, p. 2). The Panel additionally endorsed the first draft PA
9 discussions and conclusions on biologically relevant exposure metrics, stating that "the focus on
10 the W126 form is appropriate" and that "there is a strong justification made for using a
11 cumulative and weighted exposure standard for welfare effects (i.e. the W126)..." (Frey and
12 Samet, 2012b, p. 2).
13 5.7 PRELIMINARY STAFF CONCLUSIONS ON ADEQUACY OF
14 SECONDARY STANDARD
15 This section presents preliminary staff conclusions for the Administrator to consider in
16 deciding whether the existing secondary Os standard is adequate and whether it should be
17 retained or revised. Our conclusions are based on consideration of the assessment and integrative
18 synthesis of information presented in the ISA, as well as our analyses of air quality distributions;
19 analyses in the second draft WREA; and the comments and advice of CASAC and public
20 comment on an earlier draft of this document and the ISA and WREA, as discussed above.
21 Taking into consideration the responses to specific questions discussed above, we revisit the
22 overarching policy question for this chapter:
23 • Does the currently available scientific evidence and exposure/risk information, as
24 reflected in the ISA and REA, support or call into question the adequacy and/or
25 appropriateness of the protection afforded by the current secondary Os standard?
26 There is a longstanding evidence base on the phytotoxic effects of Os. Since first
27 identified by Richards, et al in 1958, numerous studies have been conducted on the effects of Os
28 on plants and their associated ecosystems. Taken together, these studies demonstrate that 63 -
29 induced effects that occur at the subcellular and cellular levels, at sufficient magnitudes
30 propagate up to produce larger scale effects that affect the whole organism. In particular in this
31 review, many of the recent studies have focused on and further increased our understanding of
32 the molecular, biochemical and physiological mechanisms that explain how plants are affected
33 by 63, in the absence of other stressors, particularly in the area of genomics (U.S. EPA, 2013,
34 Chapter 9, section 9.3).
5-59
-------
1 Based on its assessment of this extensive body of science, the ISA determined that a
2 causal relationship exists between exposure to 63 in ambient air and visible foliar injury effects
3 on vegetation, reduced vegetation growth, reduced productivity in terrestrial ecosystems, reduced
4 yield and quality of agricultural crops and alteration of below-ground biogeochemical cycles
5 (U.S. EPA 2013, Table 1-2). Additionally, the ISA determined that a likely to be causal
6 relationship exists between exposures to Os in ambient air and reduced carbon sequestration in
7 terrestrial ecosystems, alteration of terrestrial ecosystem water cycling and alteration of
8 terrestrial community composition (U.S. EPA, 2013, Table 1-2).
9 Recent studies also continue to provide evidence that adverse vegetation effects are
10 attributable to cumulative seasonal Os exposures. On the basis of the entire body of evidence in
11 this regard, the ISA concludes that "quantifying exposure with indices that cumulate hourly 63
12 concentrations and preferentially weight the higher concentrations improves the explanatory
13 power of exposure/response models for growth and yield, over using indices based on mean and
14 peak exposure values" (U.S. EPA, 2013, p. 2-44). Thus, as in other recent reviews, the evidence
15 continues to provide a strong basis for concluding that it is appropriate to judge impacts of Os on
16 vegetation, related effects and services, and the level of public welfare protection achieved, using
17 a cumulative, seasonal exposure metric, such as the W126-based metric. In addition, CAS AC
18 has consistently since the 1997 review expressed support for the use of such a metric, including
19 most recently in its letter on the first draft PA (Frey and Samet, 2012b, p. 2). Thus, based on the
20 long-established conclusions and long-standing supporting evidence described above, we
21 continue to conclude that the most appropriate and biologically-relevant way to relate O^
22 exposure to plant growth, and to determine what would be adequate protection for public welfare
23 effects attributable to the presence of Os in the ambient air, is to express exposures in terms of a
24 cumulative seasonal form, and in particular the W126 metric. Accordingly, in considering the
25 current evidence and exposure/risk information with regard to the adequacy of public welfare
26 protection it affords, staff has considered both the evidence of vegetation and welfare impacts in
27 areas of the U.S. likely to have met the current standard, as well as air quality information
28 regarding W126 index values (and evidence of effects associated with those values) in areas
29 likely to have met the current standard.
30 We further take note of CAS AC advice in the last review, where they have recognized
31 that the current secondary standard is of an averaging time and form pertaining more to 63
32 effects on human health than to the effects of 63 exposure on welfare. In so doing, they
33 emphasized the dependency of vegetation effects on the cumulative exposure to and uptake of 03
34 over the course of the entire growing season (Henderson, 2006). Accordingly, we recognize
5-60
-------
1 potential uncertainty in the level of protection provided by the current standard from vegetation
2 related effects, due to its averaging time and form.
3 Based on the considerations described in the sections above and summarized below, we
4 preliminarily conclude that the currently available evidence and exposure/risk information calls
5 into question the adequacy of public welfare protection provided by the current standard and
6 provides support for considering potential alternative standards to achieve greater public welfare
7 protection, especially for sensitive vegetation and ecosystems in federally protected Class I and
8 similar areas.
9 As an initial matter, we note that the CAA does not require that a secondary standard be
10 protective of all effects associated with a pollutant in the ambient air, but only those considered
11 adverse to the public welfare (as described in section 1.3.2 above). In judging whether particular
12 known or anticipated effects should be considered adverse, the Administrator considers a number
13 of factors, including the intended use of the affected entity(s) and the location(s) in which those
14 effects occur or are predicted to occur. In this regard, effects that have been observed in
15 federally protected Class I areas which are afforded special protections under the CAA may have
16 greater significance to the public welfare. In this context, staff has given particular consideration
17 to recent study findings of welfare-related effects in such areas, in combination with information
18 on air quality during the study periods to inform our conclusions regarding the nature and scope
19 of welfare effects in such protected areas, including areas where the current standard is met or
20 likely to have been met.
21 Regarding visible foliar injury, new research includes: 1) controlled exposure studies; 2)
22 multi-year field surveys; 3) US FS FIA biomonitoring program surveys and assessments. In
23 addition to supporting the ISA causal determination, these studies also address some
24 uncertainties indentified in the previous review (i.e., influence of soil moisture on visible injury
25 development) and further show that visible foliar injury can occur under conditions where 8-hour
26 average Os concentrations are or would be expected to be below the level of the current standard
27 (e.g., Kline et al., 2008, as discussed in section 5.4.1 above). Incidence of visible foliar injury
28 symptoms on (Vsensitive vegetation has also been documented in the field in federally
29 protected areas that have met the current standard. As one example, in Cape Romain National
30 Wildlife Refuge in SC, where air quality data indicate 63 concentrations have met the current
31 standard over the period from 2001-2012, recent study findings indicate that three species have
32 exhibited Os-induced visible foliar injury (winged sumac, Chinese tallow tree, and wild grape).
33 In at least one of the studied locations, up to 32 % of the wild grape plants examined showed
34 foliar injury symptoms during years where 8-hour Os concentrations were similar to the level of
35 the current standard (U.S. EPA 2013, section 9.4.2.1). Importantly, these Ch-induced vegetation
5-61
-------
1 effects were found in Class I areas which have identified Os-impacts to vegetation as an air
2 quality related value that should be protected commiserate with its congressionally recognized
3 public welfare significance (75 FR 3024). A staff air quality assessment identified Class I areas
4 with recent air quality that met the current standard but was above a W126 index value of 15
5 ppm-hrs (Table 5-2 above). Given evidence of the potential occurrence of visible foliar injury at
6 W126 index values of this magnitude, staff concludes that air quality levels that are at or below
7 the level of the current standard may allow levels of visible foliar injury incidence to occur in
8 areas of special significance to the public welfare.
9 Results from the exposure and risk assessments also indicate visible foliar injury at sites
10 with air quality likely to meet the current standard. In the national-scale analysis of the presence
11 or absence of foliar injury across the years 2006-2010 in U.S. National Forest sites with annual
12 W126 index values of from 7 to 15 ppm-hrs, the percent of sites indicating the presence of foliar
13 injury generally ranges from about 5 to 20 percent, varying with soil moisture conditions for the
14 year. Across all sites, regardless of the year's soil moisture conditions, these analyses find an
15 increase in the proportion of sites showing the presence of foliar injury or elevated foliar injury
16 with increasing W126 index values up to W126 values of approximately 10 ppm-hrs (see U.S.
17 EPA 2014, Figure 7-10). Using this relationship, a National Park screening assessment (214
18 parks) found that 58% of the parks analyzed exceeded a W126 of 10.46 ppm-hrs, which is
19 associated with 17.7% injury prevalence. Data indicate large reductions in injury prevalence
20 with incrementally lower W126 index values. Just meeting the current standard would bring all
21 parks under 10.46 ppm-hrs. However, as noted above, there are important uncertainties
22 associated with the model-adjusted air quality used in these assessments (U.S. EPA, 2014, Table
23 7-23, section 8.5.2) that we note in considering what the adjustment results convey with regard to
24 W126 conditions associated with just meeting the current standard.
25 We also note that visible foliar injury is associated with important cultural and
26 recreational ecosystem services to the public, such as scenic viewing, wildlife-watching, hiking,
27 and camping, that are of significance to the public welfare and enjoyed by millions of Americans
28 every year, generating millions of dollars in economic value (U.S. EPA 2014, section 7.1). In
29 addition, several tribes have indicated that many of the known confirmed (Vsensitive species
30 (including bioindicator species) are culturally significant (see Appendix 5A). These ecosystem
31 services are at risk of impairment because of Ch-induced damage directly due to foliar injury,
32 though data is not available to explicitly quantify these negative effects. Thus, while it is unclear
33 to what extent Os-induced visible foliar injury impacts these services, we conclude it is
34 appropriate to consider the potential for these effects to contribute to adverse effects on public
5-62
-------
1 welfare in considering the adequacy and appropriateness of the protection afforded by the current
2 standard.
3 Tree biomass loss is another (Vinduced effect that is important from a public welfare
4 perspective, both alone and in conjunction with associated ecosystem-level impacts. This effect
5 occurs in both Class I and non-Class I areas and its importance comes from whether this effect
6 reduces the intended use of the plant or its associated ecosystems in a way that is relevant to the
7 public welfare. Recent studies confirm and extend the evidence of Os-related effects on tree
8 growth, productivity and carbon storage. Analysis of existing data conducted by the EPA staff
9 and discussed in the ISA has substantially reduced the uncertainty associated with using OTC C-
10 R functions to predict tree growth effects in the field, as described in section 5.2.1 above (U.S.
11 EPA, 2013, section 9.6.3.2). Based on the 12 individual tree seedling C-R functions, seven of
12 the 12 species show 2% seedling biomass loss at a W126 index value below 8 ppm-hrs and in the
13 other five species at a W126 value above 18 ppm-hrs. Within the group of seven more sensitive
14 species, the most sensitive are cottonwood and black cherry (U.S. EPA 2014, section 6.2).
15 Based on the median composite C-R functions, the W126 index values associated with a 2%
16 biomass loss range from approximately 7 to 14 ppm-hrs (U.S. EPA, 2014, Figure 6-4, section
17 6.2.1.2). Recent studies have provided additional evidence on tree biomass or growth effects
18 associated with multiple year exposures in the field, including the potential for compounding and
19 carry-over effects. These effects could lead to a negative impact on species regeneration and
20 have implications for the subsequent growing season in the following year (U.S. EPA, 2013,
21 section 9.4), as recognized in recent advice from CASAC (Frey and Samet, 2012b).
22 Accordingly, in considering WREA biomass loss estimates, we also consider the WREA
23 evaluation of a potential for underestimation of such effects (U.S. EPA, 2014, section 6.2.1.4).
24 While it is not possible at this time to identify the extent or magnitude of such effects in the field
25 under exposure levels that may be associated with the current standard, their occurrence, when of
26 a magnitude of concern, on federally protected lands or any associated impacts on ecosystem
27 services such as habitat quality, hydrologic regimes, carbon storage or air pollution removal in
28 areas important to the public might reasonably be concluded to be significant to public welfare.
29 We additionally take note of the WREA findings with regard to urban areas where tree CC>2
30 sequestration and air pollutant removal are important public welfare services.
31 With respect to crops, the detrimental effect of Os on crop production has been
32 recognized since the 1960s, and recent O?, concentrations in many areas across the U.S. are high
33 enough that they might be expected to cause yield loss in a variety of agricultural crops
34 including, but not limited to, soybeans, wheat, potatoes, watermelons, beans, turnips, onions,
35 lettuces, and tomatoes (U.S. EPA, 2013, section 9.4.4). In general, the vast majority of the new
36 scientific information confirms prior conclusions that exposure to Oi can decrease growth and
5-63
-------
1 yield of crops. Recent research has highlighted the effects of Os on crop quality. Increasing O^
2 concentration decreases nutritive quality of grasses, decreases macro- and micro-nutrient
3 concentrations in fruits and vegetable crops (U.S. EPA 2013, section 9.4.4). Recent studies
4 continue to find yield loss levels in crop species studied previously under NCLAN that reflect
5 the earlier findings. There has been little published evidence that crops are becoming more
6 tolerant of O3 (U.S. EPA, 2006a; U.S. EPA 2013). This is especially evident in the research on
7 soybean. The 2013 ISA reported comparisons between yield predictions based on data from
8 cultivars used in NCLAN studies, and yield data for modern cultivars from SoyFACE (U.S.
9 EPA, 2013, section 9.6.3). They confirm that the average response of soybean yield to O3
10 exposure has not changed in current cultivars. In addition, satellite and ground-based 63
11 measurements have been used to assess yield loss caused by Os over the continuous tri-state area
12 of Illinois, Iowa, and Wisconsin. The results showed that Os concentrations reduced soybean
13 yield, which correlates well with the previous results from FACE- and OTC-type experiments
14 (U.S. EPA, 2013, section 9.4.4.1). Thus, the recently available evidence, as assessed in the ISA,
15 continues to support the conclusions of the 1996 and 2006 Criteria Documents that ambient 63
16 concentrations can reduce the yield of major commodity crops in the U.S.
17 The currently available evidence, as assessed in the ISA, continues to support the use of
18 C-R functions based on OTC experiments. Further, important uncertainties have been reduced
19 regarding the exposure-response functions for crop yield loss, especially for soybean. In general,
20 the ISA reports consistent results across exposure techniques and across crop varieties (U.S. EPA
21 2013, section 9.6.3.2). Based on the crop C-R functions, the W126 index values associated with
22 a five percent yield loss range from approximately 12 to 17 ppm-hrs. The lower end of the range
23 would be more protective of soybeans which are the second-most planted field crop in the U.S.
24 (http://www.ers.usda.gov/topics/crops/soybeans-oil-crops/background.aspx). Staff analyses of
25 recent monitoring data (2009-2011) indicate that 63 concentrations in multiple agricultural areas
26 in the U.S. that meet the current standard correspond to W126 index levels above 12 ppm-hrs.
27 The information for other welfare effects, including those with causal or likely causal
28 relationships with 03 (i.e., alteration of ecosystem water cycling, changes in climate), is limited
29 with regard to our ability to consider potential impacts of air quality conditions associated with
30 the current standard, although the WREA provides some perspective on this issue with regard to
31 susceptibility to insect attack and fire regime change. We note, however, the importance of these
32 effects categories to the public welfare.
33 Given all of the above, staff reaches the preliminary conclusion that the available
34 evidence and exposure and risk information calls into question the adequacy of public welfare
35 protection provided by the current standard, and provides strong support to giving consideration
5-64
-------
1 to revising the current secondary standard to provide increased public welfare protection. More
2 specifically, staff preliminarily concludes that it is appropriate for the Administrator to consider
3 revision of the current secondary Os standard to increase protection against Os-attributable tree
4 biomass loss, crop yield loss, and visible foliar injury, and particularly for those effects
5 associated with cumulative, seasonal exposures that occur in specially protected natural areas. In
6 reaching conclusions on options for the Administrator's consideration, we note that the final
7 decision to retain or revise the current secondary Oj, standard is a public welfare policy judgment
8 to be made by the Administrator as to what standards would be requisite (i.e., neither more nor
9 less stringent than necessary) to protect the public welfare from any known or anticipated
10 adverse welfare effects. This final decision will draw upon the available scientific evidence for
11 Os-attributable welfare effects and on analyses of vegetation and ecosystem exposures and public
12 welfare risks based on impacts to vegetation, ecosystems and their associated services, including
13 judgments about the appropriate weight to place on the range of uncertainties inherent in the
14 evidence and analyses. In determining the requisite level of protection for crops and trees, the
15 Administrator will need to weigh the importance of the predicted risks of these effects in the
16 overall context of public welfare protection, along with a determination as to the appropriate
17 weight to place on the associated uncertainties and limitations of this information.
18 As noted in section 1.3.2 above, our general approach to informing these judgments,
19 recognizes that the available welfare effects evidence reflects a continuum consisting of higher
20 Os concentrations at which scientists generally agree that welfare effects are likely to occur,
21 through lower concentrations at which the likelihood and magnitude of a response become
22 increasingly uncertain. Therefore, in presenting conclusions here, we are mindful that the
23 Administrator's ultimate judgment on the adequacy of protection afforded by the current
24 standard will most appropriately reflect an interpretation of the available scientific evidence and
25 exposure/risk information that neither overstates nor understates the strengths and limitations of
26 that evidence and information.
27 5.8 REFERENCES
28 Abt Associates, Inc. (1995). Ozone NAAQS benefits analysis: California crops. Report to U.S. EPA, July.
29 Andersen, CP; Wilson, R; Plocher, M; Hogsett, WE. (1997). Carry-over effects of ozone on root growth and
30 carbohydrate concentrations of ponderosa pine seedlings. Tree Physiol 17: 805-811.
31 Betzelberger, AM; Gillespie, KM; McGrath, JM; Koester, RP; Nelson, RL; Ainsworth, EA. (2010). Effects
32 ofchronic elevated ozone concentration on antioxidant capacity, photosynthesis and seed yield of 10
33 soybeancultivars. Plant Cell Environ 33: 1569-1581. http://dx.doi.0rg/10.llll/j.1365-3040.2010.02165.x
34 Broadmeadow, MSJ; Jackson, SB. (2000). Growth responses of Quercus petraea, Fraxinus excelsior and
35 Pinussylvestris to elevated carbon dioxide, ozone and water supply. New Phytol 146: 437-
36 451.http://dx.doi.org/10.1046/j.l469-8137.2000.00665.x
5-65
-------
1 Campbell, SJ; Wanek, R; Coulston, JW. (2007). Ozone injury in west coast forests: 6 years of monitoring -
2 Introduction. Portland, OR: U.S. Department of Agriculture.
3 Davis, DD. (2007a). Ozone-induced symptoms on vegetation within the Moosehorn National Wildlife Refuge in
4 Maine. Northeast Nat 14: 403-414. http://dx.doi.org/10.1656/1092-6194(2007)14r403:OSOVWT12.0.CO:2
5 Davis, DD. (2007b). Ozone injury to plants within the Seney National Wildlife Refuge in northern Michigan.
6 Northeast Nat 14: 415-424.
7 Davis, DD. (2009). Ozone-induced stipple on plants in the Cape Remain National Wildlife Refuge, South Carolina.
8 Southeastern Naturalist 8: 471-478.
9 Davis, DD; Orendovici, T. (2006). Incidence of ozone symptoms on vegetation within a National Wildlife Refuge in
10 New Jersey, USA. EnvironPollut 143: 555-564. http://dx.doi.0rg/10.1016/j.envpol.2005.10.051
11 Federal Register 1996. National Ambient Air Quality Standards for Ozone; Proposed Rule. 40 CFR 50; Federal
12 Register 61: 65716
13 Federal Register 1997. National Ambient Air Quality Standards for Ozone; Final Rule. 40 CFR 50; Federal Register
14 62:38856
15 Federal Register 2007. National Ambient Air Quality Standards for Ozone; Proposed Rule. 40 CFR 50; Federal
16 Register 72: 37818
17 Federal Register 2008. National Ambient Air Quality Standards for Ozone; Final Rule. 40 CFRparts 50 and 58;
18 Federal Register 73:16436
19 Federal Register 2010. National Ambient Air Quality Standards for Ozone; Proposed Rule. 40 CFR 50 and 58;
20 Federal Register 75 FR: 2938
21 Federal Register 2012. National Ambient Air Quality Standards for Oxides of Nitrogen and Sulfur; Final Rule. 40
22 CFR 50; Federal Register 77 FR 20218
23 Fiore, AM; Jacob, DJ; Field, BD; Streets, DG; Fernandes, SD; Jang, C. (2002). Linking ozone pollution and climate
24 change: The case for controlling methane. Geophys Res Lett 29: 1919.
25 http://dx.doi.org/10.1029/2002GL015601
26 Flagler, RB. (1998). Recognition of air pollution injury to vegetation: A pictorial atlas (2nd ed.). Pittsburgh, PA: Air
27 & Waste Management Association.
28 Frey, C. and Samet, J.M. (2012a) CASAC Review of the EPA's Policy Assessment for the Review of the Ozone
29 National Ambient Air Quality Standards (First External Review Draft-August 2012). EPA-CASAC-13-
30 003. November 26, 2012.
31 Frey, C. and Samet, J.M. (2012b) CASAC Review of the EPA's Health Risk and Exposure Assessment for Ozone
32 (First External Review Draft - Updated August 2012) and Welfare Risk and Exposure Assessment for
33 Ozone (First External Review Draft-Updated August 2012). EPA-CASAC-13-002. November 19, 2012.
34 Grantz, DA; Gunn, S; Vu, HB. (2006). O3 impacts on plant development: A meta-analysis of root/shoot allocation
35 and growth. Plant Cell Environ 29: 1193-1209. http://dx.doi.0rg/10.llll/j.1365-3040.2006.01521.x
36 Gregg, JW; Jones, CG; Dawson, TE. (2003). Urbanization effects on tree growth in the vicinity of New York City
37 [Letter]. Nature 424: 183-187. http://dx.doi.org/10.1038/nature01728
5-66
-------
1 Grulke, NE; Johnson, R; Esperanza, A; Jones, D; Nguyen, T; Posch, S; Tausz, M. (2003). Canopy transpiration of
2 Jeffrey pine in mesic and xeric microsites: O3 uptake and injury response. Trees Struct Funct 17: 292-298.
3 Grulke, NE; Minnich, RA; Paine, TD; Seybold, SJ; Chavez, DJ; Fenn, ME; Riggan, PJ; Dunn, A. (2008).
4 Airpollution increases forest susceptibility to wildfires: A case study in the San Bernardino Mountains in
5 southern California. In A Bytnerowicz; MJ Arbaugh; AR Riebau; C Anderson (Eds.), Wildland fires and
6 air pollution; Section III: Ecological Impacts of Forest Fires and Air Pollution (pp. 365-403). Amsterdam,
7 The Netherlands: Elsevier Ltd. http://dx.doi.org/10.1016/S1474-8177(08)00017-X
8 Heagle, AS. (1989). Ozone and crop yield*. Annu Rev Phytopathol 27: 397-423.
9 http://dx.doi.org/10.1146/annurev.py.27.090189.002145
10 Heck, WW; Cowling, EB. (1997).The need for a long term cumulative secondary ozone standard - An ecological
11 perspective. EM January: 23-33.
12 Henderson, R. (2006) Letter from CAS AC Chairman Rogene Henderson to EPA Administrator Stephen Johnson.
13 October 24, 2006, EPA-CASAC-07-001.
14 Henderson, R. (2008) Letter from CAS AC Chairman Rogene Henderson to EPA Administrator Stephen Johnson.
15 April 7,2008, EPA-CASAC-08-009.
16 Hogsett, WE; Weber, JE; Tingey, D; Herstrom, A; Lee, EH; Laurence, JA. (1997). Environmental auditing: An
17 approach for characterizing tropospheric ozone risk to forests. J Environ Manage 21: 105-120.
18 http://dx.doi.org/10.1007/s002679900010
19 IPCC (Intergovernmental Panel on Climate Change). (2007). Climate change 2007: Impacts, adaptation and
20 vulnerability. Cambridge, UK: Cambridge University Press.
21 Kline, LJ; Davis, DD; Skelly, JM; Savage, JE; Ferdinand, J. (2008). Ozone sensitivity of 28 plant selections exposed
22 to ozone under controlled conditions. Northeast Nat 15: 57-66. http://dx.doi.org/10.1656/1092-
23 6194(2008)15[57:OSOPSE]2.0.CO;2
24 Kubiske, ME; Quinn, VS; Heilman, WE; McDonald, EP; Marquardt, PE; Teclaw, RM; Friend, AL; Karnoskey, DF.
25 (2006). Interannual climatic variation mediates elevated CO2 and O3 effects on forest growth. Global
26 Change Biol 12: 1054-1068. http://dx.doi.0rg/10.llll/j.1365-2486.2006.01152.x
27 Kubiske, ME; Quinn, VS; Marquardt, PE; Karnosky, DF. (2007). Effects of elevated atmospheric CO2 and/or O3 on
28 intra- and interspecific competitive ability of aspen. Plant Biol (Stuttg) 9: 342-355.
29 http://dx.doi.org/10.1055/S-2006-924760
30 Lee, EH; Hogsett, WE. (1996). Methodology for calculating inputs for ozone secondary standard benefits anaylsis:
31 Part II. Research Triangle Park, NC: U.S. Environmental Protection Agency.
32 Lefohn, AS; Laurence, JA; Kohut, RJ. (1988). A comparison of indices that describe the relationship between
33 exposure to ozone and reduction in the yield of agricultural crops. Atmos Environ 22: 1229-1240.
34 http://dx.doi.org/10.1016/0004-6981(88)90353-8
3 5 Lefohn, AS; Runeckles, VC. (1987). Establishing standards to protect vegetation - ozone exposure/dose
36 considerations. Atmos Environ 21: 561-568. http://dx.doi.org/10.1016/0004-6981(87)90038-2
37 Low, M; Herbinger, K; Nunn, AJ; Haberle, KH; Leuchner, M; Heerdt, C; Werner, H; Wipfler, P; Pretzsch, H;
3 8 Tausz, M; Matyssek, R. (2006). Extraordinary drought of 2003 overrules ozone impact on adult beech trees
39 (Fagus sylvatica). Trees Struct Funct 20: 539-548. http://dx.doi.org/10.1007/s00468-006-0069-z
40 Maier-Maercker, U. (1998). Predisposition of trees to drought stress by ozone. Tree Physiol 19: 71-78.
5-67
-------
1 McLaughlin, SB; Nosal, M; Wullschleger, SD; Sun, G. (2007a). Interactive effects of ozone and climate on tree
2 growth and water use in a southern Appalachian forest in the USA. New Phytol 174: 109-124.
3 http://dx.doi.0rg/10.llll/j.1469-8137.2007.02018.x
4 McLaughlin, SB; Wullschleger, SD; Sun, G; Nosal, M. (2007b). Interactive effects of ozone and climate on water
5 use, soil moisture content and streamflow in a southern Appalachian forest in the USA. New Phytol 174:
6 125-136. http://dx.doi.0rg/10.llll/j.1469-8137.2007.01970.x
7 Millennium Ecosystem Assessment (MEA). (2005). Ecosystems and Human Well-being: CurrentState and Trends,
8 Volume 1. R. Hassan, R. Scholes, and N. Ash, eds. Washington,DC:Island Press. Available at
9 http://www.millenniumassessment.0rg/documents/document.766.aspx.pdfhttp://www.millenniumassessme
10 nt.org/documents/document.766.aspx.pdf.
11 Paoletti, E; Manning, WJ. (2007). Toward a biologically significant and usable standard for ozone that will also
12 protect plants. Environ Pollut 150: 85-95. http://dx.doi.0rg/10.1016/i.envpol.2007.06.037
13 Pollastrini, M; Desotgiu, R; Cascio, C; Bussotti, F; Cherubini, P; Saurer, M; Gerosa, G; Marzuoli, R. (2010).
14 Growth and physiological responses to ozone and mild drought stress of tree species with different
15 ecological requirements. Trees Struct Funct 24: 695-704. http://dx.doi.org/10.1007/s00468-010-0439-4
16 Richards, B.L., Middleton, J.T., and Hewitt, W.B. (1958). Air pollution with relation to agronomic crops. V.
17 Oxidant stipple on grape. Agron. J. 50:559-561.
18 Rose, A and Coulston, J. W. (2009). Ozone injury across the Southern United States, 2002 - 2006. Gen. Tech. Rep.
19 SRS-118. Asheville, N.C.: US Department of Agriculture, Forest Service, Southern Research Station, pp.
20 25.
21 Smith, G; Coulston, J; Jepsen, E; Prichard, T. (2003). A national ozone biomonitoring program: Results from field
22 surveys of ozone sensitive plants in northeastern forests (1994-2000). EnvironMonit Assess 87: 271-291.
23 Smith, G. (2012). Ambient ozone injury to forest plants in Northeast and North Central USA: 16 years of
24 biomonitoring. Environ Monit Assess 184: 4049-4065. http://dx.doi.org/10.1007/sl0661-011-2243-z
25 Stevenson, DS; Dentener, FJ; Schultz, MG; Ellingsen, K; Van Noije, TPC; Wild, O; Zeng, G; Amann, M; Atherton,
26 CS; Bell, N; Bergmann, DJ; Bey, I; Butler, T; Cofala, J; Collins, WJ; Derwent, RG; Doherty, RM; Drevet,
27 J; Eskes, HJ; Fiore, AM; Gauss, M; Hauglustaine, DA; Horowitz, LW; Isaksen, ISA; Krol, MC; Lamarque,
28 JF; Lawrence, MG; Montanaro, V; Muller, JF; Pitari, G; Prather, MJ; Pyle, JA; Rast, S; Rodriguez, JM;
29 Sanderson, MG; Savage, NH; Shindell, DT; Strahan, SE; Sudo, K; Szopa, S. (2006). Multimodel ensemble
3 0 simulations of present-day and near-future tropospheric ozone. J Geophys Res 111: D08301.
31 http://dx.doi.org/10.1029/2005JD006338
32 UNEP (United Nations Environment Programme). (2003). Millennium Ecosystem Assessment: Ecosystems and
33 human well-being: A framework for assessment. Washington, DC: Island Press.
34 US EPA (2014). Welfare Risk and Exposure Assessment for Ozone, Second External Review Draft. Office of Air
3 5 Quality Planning and Standards, Research Triangle Park, NC, 27711. EPA-452/P-14-003a
36 US EPA (2013). Integrated Science Assessment of Ozone and Related Photochemical Oxidants (Final). U.S.
3 7 Environmental Protection Agency, Washington, DC. EPA/600/R-10/076F
38 US EPA (2012). Regulatory Impact Analysis for the Final Revisions to the National Ambient Air Quality Standards
39 for Paniculate Matter U.S. Environmental Protection Agency, Office of Air Quality Planning and
40 Standards, Health and Environmental Impacts Division,Research Triangle Park, NC 27711. EPA-452/R-
41 12-005. Available at: http://www.epa.gov/ttn/naaqs/standards/pm/s_pm_2007_ria.html
5-68
-------
1 US EPA (2009). Risk and Exposure Assessment for the Review of the Secondary national Ambient Air Quality
2 Standards for Oxides of Nitrogen and Oxides of Sulfur. EP A/452/R/09/008a. U. S environmental Protection
3 Agency, Office of Air Quality planning and Standards, Health and Environmental Impacts Division,
4 Research Triangle Park, NC.
5 US EPA (2006a). Air Quality Criteria for Ozone and Related Photochemical Oxidants (2006 Final). U.S.
6 Environmental Protection Agency, Washington, DC. EPA/600/R-05/004aF-cF. March 2006. Available at:
7 http://www.epa.gOv/ttn/naaqs/standards/ozone/s o3crcd.html
8 US Environmental Protection Agency (U.S. EPA). 2006b. Ecological Benefits Assessment Strategic Plan. EPA-240-
9 R-06-001. Office of the Administrator. Washington, DC. October. Available on the Internet at
10 .
11 US EPA (1996). Air quality criteria for ozone and related photochemical oxidants [EPA Report]. (EPA/600/P-
12 93/004AF). U.S. Environmental Protection Agency, Research Triangle Park, NC.
13 US EPA (1984). Air quality criteria for ozone and other photochemical oxidants, volume III of V (review draft)
14 [EPA Report]. (EPA-600/8-84-020A3). Research Triangle Park, NC.
15 http://www.ntis.gov/search/product.aspx? ABBR=PB85126050
16 USEPA(1978). Air quality criteria for ozone and other photochemical oxidants [EPAReport]. (EPA/600/8-
17 78/004). Washington, DC.
18 West, JJ; Fiore, AM; Horowitz, LW; Mauzerall, DL. (2006). Global health benefits of mitigating ozone pollution
19 with methane emission controls. PNAS 103: 3988-3993. http://dx.doi.org/10.1073/pnas.0600201103
20 West, JJ; Fiore, AM; Naik, V; Horowitz, LW; Schwarzkopf, MD; Mauzerall, DL. (2007). Ozone air quality and
21 radiative forcing consequences of changes in ozone precursor emissions. Geophys Res Lett 34: L06806.
22 http://dx.doi.org/10.1029/2006GL029173
23 Wittig, VE; Ainsworth, EA; Long, SP. (2007). To what extent do current and projected increases in surface ozone
24 affect photosynthesis and stomatal conductance of trees? A meta-analytic review of the last 3 decades of
25 experiments [Review]. Plant Cell Environ 30: 1150-1162. http://dx.doi.0rg/10.llll/j.1365-
26 3040.2007.01717.x
27 Wittig, VE; Ainsworth, EA; Naidu, SL; Karnosky, DF; Long, SP. (2009). Quantifying the impact of current and
28 future tropospheric ozone on tree biomass, growth, physiology and biochemistry: A quantitative meta-
29 analysis. Global Change Biol 15: 396-424. http://dx.doi.0rg/10.llll/j.1365-2486.2008.01774.x
30 Young, T. F.; Sanzone, S., eds. (2002). A framework for assessing and reporting on ecological condition: an SAB
31 report. Washington, DC: U.S. Environmental Protection Agency, Science Advisory Board; report no.
32 EPASAB- EPEC-02-009. Available: http://www.epa.gov/sab/pdf/epec02009.pdf [9 December, 2003].
5-69
-------
1 6 CONSIDERATION OF ALTERNATIVE SECONDARY
2 STANDARDS
3 To the extent that the information available in this review suggests that revision of the
4 current secondary standard is appropriate to consider, as discussed in chapter 5, staff has
5 evaluated the available body of evidence, and exposure, risk and air quality information with
6 regard to the support for consideration of alternative standards, as articulated by the following
7 overarching question:
8 • What alternative secondary standards are supported by the currently available
9 scientific evidence, exposure/risk information and air quality analyses?
10 To assist us in interpreting the currently available scientific evidence and the results of
11 recent quantitative exposure/risk analyses to address this question, we have focused on a series
12 of more specific questions in sections 6.1, 6.2 and 6.3 below. We consider both the scientific
13 and technical information available at the time of the last review and information newly available
14 since the last review which has been critically analyzed and characterized in the ISA.
15 Specifically, we consider the currently available scientific evidence and technical information in
16 the context of decisions regarding the basic elements of the NAAQS: indicator, (section 6.1);
17 averaging time and form (section 6.2); and level (section 6.3). CASAC advice on potential
18 alternative standards is described in section 6.4 and preliminary staff conclusions on potential
19 alternative standards are discussed in section 6.5. Section 6.6 summarizes preliminary staff
20 conclusions on adequacy of the current standard and potential alternative standards appropriate
21 to consider. Key uncertainties in this review and areas in which future research and data
22 collection would better inform the next review are identified in section 6.7.
23 6.1 INDICATOR
24 With regard to the indicator for potential alternative secondary standards, we consider the
25 following question.
26 • Does the information available in this review continue to support Os as the
27 indicator for ambient air photochemical oxidants?
28 In the last review of the air quality for Oj, and other photochemical oxidants and of the 63
29 standard, as in other prior reviews, the EPA focused on a standard for Os as the most appropriate
30 surrogate for ambient photochemical oxidants. Ozone is a long-established surrogate for
31 ambient photochemical oxidants, among which it is by far the most widely studied with regard to
32 effects on welfare and specifically on vegetation. The information available in this review adds
33 to our understanding of the atmospheric chemistry for photochemical oxidants and Os in
6-1
-------
1 particular (as described in the ISA, sections 3.2 and 3.6, and summarized in section 2.2 in this
2 document). The 1996 Staff Paper noted that the database on vegetation effects is generally
3 considered to raise concern at levels found in the ambient air for Os and, therefore, control of
4 ambient O?, levels has previously been concluded to provide the best means of controlling other
5 photochemical oxidants of potential welfare concern (U.S. EPA, 1996b, p. 277). In the current
6 review, while the complex atmospheric chemistry in which Os plays a key role has been
7 highlighted, no alternatives to 63 have been advanced as being a more appropriate surrogate for
8 ambient photochemical oxidants. Ozone continues to be the only photochemical oxidant other
9 than nitrogen dioxide that is routinely monitored and for which a comprehensive database exists
10 (U.S. EPA, 2013, section 3.6). Thus, staff concludes that 63 remains the appropriate pollutant
11 indicator for use in a secondary NAAQS that provides protection for public welfare from
12 exposure to all photochemical oxidants.
13 6.2 FORM AND AVERAGING TIME
14 In considering potential forms and averaging times alternative to that of the current
15 secondary standard (i.e. 4* high daily maximum 8-hour average, averaged over 3 years) we
16 address several specific questions.
17 • To what extent does the currently available information provide support for
18 considering forms different from that of the current secondary standard?
19 In characterizing the current evidence, the ISA states that "[n]o recent information is
20 available since 2006 that alters the basic conclusions put forth in the 2006 and 1996 Oi AQCDs"
21 with regard to biologically relevant exposure indices (U.S. EPA, 2013, section 2.6.6.1). Based
22 on the current state of knowledge and the best available data assessed in this review, the ISA
23 therefore concludes that exposure indices that cumulate and differentially weight the higher
24 hourly average concentrations over a season and also include the mid-level values continue to
25 offer the most defensible approach for use in developing response functions and in defining
26 indices for vegetation protection (U.S. EPA, 2013, section 2.6.6.1). In particular, the available
27 body of evidence provides a wealth of information, compiled over several decades, on the
28 aspects of 63 exposure that are most important in influencing plant response. As discussed in the
29 ISA, the importance of the duration of the exposure and the relatively greater importance of
30 higher concentrations (over lower concentrations) in determining plant response to Os have been
31 well documented (U.S. EPA, 2013, section 9.5.3). Building on this research, other work has
32 been focused on developing "mathematical approaches for summarizing ambient air quality
33 information in biologically meaningful forms for 63 vegetation effects assessment purposes ..."
34 (U.S. EPA, 2013, section 9.5.3), including those known as cumulative, concentration weighted
6-2
-------
1 forms (i.e. SUM06, W126). Much of this work was completed by the mid-1990s, and was
2 summarized in the 1996 Criteria Document (CD) (U.S. EPA, 1996a, section 5.5).
3 On the basis of this longstanding and extensive evidence demonstrating that the risk to
4 vegetation comes from cumulative seasonal exposures, the EPA in the 1997 and 2008 reviews, as
5 well as in the 2010 proposed rulemaking to reconsider the 2008 decision, recognized the
6 importance of cumulative, seasonal exposures as a primary determinant of plant responses to Os
7 in ambient air (61 FR 65741-42; 62 FR 38878; 72 FR 37893, 37896, 37900, 37904; 73 FR
8 16488-90, 16493-94; 75 FR 3000, 3010, 3012). For example, in the 1996 notice of proposed
9 rulemaking, the Administrator recognized that the scientific evidence supported the conclusion
10 that "a cumulative seasonal exposure index is more biologically relevant than a single event or
11 mean index" (61 FR 65742). In the 2008 review, CAS AC recognized that an important
12 difference between the effects of short-term exposures to O?, on human health and the effects of
13 63 exposures on welfare is that "vegetation effects are more dependent on the cumulative
14 exposure to, and uptake of, Os over the course of the entire growing season" (Henderson, 2006).
15 In that review, the CAS AC 63 Panel members were unanimous in supporting the final Staff
16 Paper recommendation that "protection of managed agricultural crops and natural terrestrial
17 ecosystems requires a secondary Ozone NAAQS that is substantially different from the primary
18 ozone standard in averaging time, level and form" (Henderson, 2007). Accordingly, in both the
19 1997 and 2008 reviews as well as the 2010 reconsideration, the Administrator proposed a
20 secondary standard with a cumulative seasonal form as an appropriate policy option (61 FR
21 65742-44; 72 FR 37899-905; 75 FR 3012-3027).
22 Over the past two decades, several indices have been considered with regard to their
23 representation of Os exposure in terms best describing its effects on vegetation. The 1996 CD
24 and 1996 Staff Paper evaluated different types of forms to determine which performed the best in
25 describing the available data on plant response to Os exposures. These documents noted that a
26 number of forms (e.g., the one event, mean and unweighted cumulative SUMOO) are unable to
27 reliably predict plant response because they either ignore the role of duration or ignore the
28 disproportionate impact of higher concentrations by weighting all concentrations equally (U.S.
29 EPA, 1996b, p. 224). Other forms that were considered include multicomponent forms.
30 Because these forms take into account many other relevant factors (e.g., plant growth stage,
31 predisposition from earlier exposures), they consistently predict plant response best of all the
32 different exposure forms. However, due to being species-specific and highly complex, they were
33 not considered suitable for more general application in the context of standard setting (U.S. EPA,
34 1996b, pp. 224-225). On the other hand, it was also found that concentration-weighted forms
35 that take into account the role of duration and concentration perform almost as well as the
36 multicomponent forms. These include several threshold forms (e.g., SUM06, AOT60) and
6-3
-------
1 sigmoidally weighted cumulative indices (e.g., W126) (U.S. EPA, 1996a, pp. 5-84 to 5-136; U.S.
2 EPA, 1996b, pp. 223-227). Given that these cumulative concentration-weighted forms all were
3 able to similarly predict plant response on the datasets for which they were evaluated (i.e.
4 NCLAN), it was not possible to distinguish between them on this basis. Partly as a result,
5 CAS AC deliberations in 1995 did not produce a consensus on which form would be best suited
6 for a secondary NAAQS. As discussed further in 6.3 below, a workshop held in January of 1996
7 provided a consensus recommendation on the SUM06 form as appropriate for use in secondary
8 standards, while also recognizing that a W126 form could also be appropriate (Heck and
9 Cowling, 1997). Subsequent to this, the final 1996 Staff Paper and 1996 proposal notice both
10 identified the SUM06 form as appropriate to consider and propose, respectively (U.S. EPA,
11 1996b, p. 285, 61 FR 65716). In selecting the SUM06 form, it was noted that while it imposed a
12 threshold even while evidence was lacking for a discernible threshold for Os-related vegetation
13 effects, it had the benefit of not including concentrations considered to be from background
14 sources, which was an important feature (U.S. EPA, 1996b, pp. 223-227).
15 In the subsequent review, the form of the standard was revisited in light of continued
16 evidence that there remained a lack of discernible threshold for vegetation effects in general, and
17 newer estimates of Os concentrations associated with background sources that were lower than in
18 the previous review so that their inclusion was less of a concern. On these bases, the 2007 Staff
19 Paper recommended consideration of the W126 index1 as the basis for the form of a distinct
20 secondary standard (U.S. EPA, 2006, pp. 9-11 to 9-15 and pp. AX9-159 to AX9-187; U.S. EPA,
21 2007, pp. 7-15/16). The EPA then proposed two options for the secondary standard, one of
22 which was to adopt a cumulative, seasonal standard based on the W126 index, while the other
23 option was a secondary standard identical to the proposed revised primary standard (72 FR
24 37818). The CASAC Panel in that review expressed preference for the W126 index (Henderson,
25 2006). In EPA's proposed reconsideration of the 2008 decision, the Agency proposed only the
26 option of a cumulative, seasonal standard based on the W126 function (75 FR 2938).
27 In this review, we conclude that specific features associated with the W126 index still
28 make it the most appropriate cumulative concentration-weighted form for use in the context of
29 the secondary 63 NAAQS review. In particular, the W126 index does not apply an arbitrary
30 exposure threshold below which concentrations are not included. Given the acknowledged range
31 in vegetation sensitivity and the continued lack of evidence of an exposure threshold for effects
32 above a W126 index of zero, such a feature is desirable. In addition, we conclude the W126
1 The W126 is a non-threshold approach described as the sigmoidally weighted sum of all hourly O3
concentrations observed during a specified diurnal and seasonal exposure period, where each hourly O3
concentration is given a weight that increases from 0 to 1 with increasing concentration (Lefohn et al, 1988; Lefohn
and Runeckles, 1987; U.S. EPA, 2013, section 9.5.2).
6-4
-------
1 sigmoidal weighting function provides the best way to weight concentrations associated with
2 background sources. Thus, we conclude that the W126 form is best matched to the evidence
3 associated with vegetation effects, as well as addressing the policy-relevant issue of how to
4 weight exposures associated with background sources.
5 • To what extent does the currently available information provide support for
6 consideration of a cumulative seasonal form derived as a sum of weighted Os
7 concentrations over daylight hours (8:00 am to 8:00 pm) for the 3 month period
8 having the highest sum within the Os season?
9 For a standard with a cumulative seasonal form, including one defined in terms of the
10 W126 exposure index, the NAAQS element of "averaging time" is appropriately considered in
11 terms of exposure periods - diurnal and seasonal — over which the index would be summed in
12 any given year. In the EPA's consideration of such a standard in both the 1997 and 2008
13 reviews, the diurnal period of interest was identified to be the daylight hours, which were defined
14 for these purposes as spanning the 12-hour period from 8 am to 8 pm during the growing season.
15 The seasonal period of interest was identified as the consecutive 3 month period with the
16 maximum W126 index value that occurs within the monitored Os season. Since many species of
17 vegetation have growing seasons longer than 3-months in some locations within the U.S., this
18 seasonal period is viewed as the minimum duration that can appropriately serve as a surrogate
19 for the range of growing seasons associated with U.S. vegetation (72 FR 37883; U.S. 2013,
20 section 9.5.3, p. 9-112/13. As discussed below, the evidence available in this review continues
21 to provide support for use of these periods in considering a cumulative seasonal secondary
22 standard (U.S. EPA, 2013, section 9.5.3).
23 For the majority of plants, the diurnal conditions of maximum 63 uptake occur mainly
24 during the daytime hours (U.S. EPA, 2013, section 9.5.3.2). This is because, in general, (1)
25 plants have the highest stomatal conductance during the daytime; (2) atmospheric turbulent
26 mixing is greatest during the day in many areas; (3) the high temperature and high light
27 conditions that typically promote the formation of tropospheric Os also promote physiological
28 activity in vegetation (U.S. EPA, 2013, section 9.5.3.2). In consideration of such a cumulative
29 metric in the prior review, the EPA has identified the 12-hour period from 8 am to 8 pm as
30 appropriately capturing the diurnal window with most relevance to the photosynthetic process
31 (72 FR 37900; 75 FR 3013). In so doing, the EPA recognized, as did CASAC that in some parts
32 of the country this period may not include all daytime hours or exposures of importance to
33 vegetation, thus potentially underestimating the impact of 63 at those sites (72 FR 37900-01; 75
34 FR 3013-14; Henderson, 2007, p. 3, pp. C-22-23).
35 In addition to daytime uptake, a number of studies have also reported O?, uptake at night
36 in some species (U.S. EPA, 2013, section 9.5.3.2). Typically the rate of stomatal conductance at
6-5
-------
1 night is much lower than during the day. Several field studies have attempted to quantify night-
2 time 63 uptake with a variety of methods. Across the studies discussed in the ISA, nocturnal
3 conductance ranged from negligible to 25% of daytime values (U.S. EPA, 2013, section 9.5.3.2).
4 In some studies the percent of nocturnal uptake varied by season and drought conditions.
5 However, many of these studies did not link the night-time flux to measured effects on plants,
6 making it difficult to know whether the impacts on the plant from nocturnal exposures are
7 greater or less than those from similar daytime exposures, and whether or not they should be
8 considered as separate impacts or as additive or synergistic with impacts from the preceding
9 daytime exposure. In addition to the uncertainties associated with understanding the plant
10 response to nocturnal uptake, there are also uncertainties associated with the extent of its
11 occurrence. For significant nocturnal stomatal flux and Os effects to occur, a susceptible plant
12 with nocturnal stomatal conductance and low defenses must be growing in an area with relatively
13 high night-time 63 concentrations (often high elevation sites) and appreciable nocturnal
14 atmospheric turbulence. It is unclear how many areas there are in the U.S. where these conditions
15 occur. It may be possible that these conditions exist in mountainous areas of southern California,
16 front-range of Colorado and the Great Smoky Mountains of North Carolina and Tennessee.
17 More information is needed in locations with high night-time O^ to assess the local O^ patterns,
18 micrometeorology and responses of potentially vulnerable plant species (U.S. EPA, 2013,
19 section 9.5.3.2).
20 In consideration of such uncertainties regarding the importance and extent of nocturnal
21 exposures associated with plant uptake, and whether and how they might be incorporated into a
22 national index, staff continues to focus on the 12-hour daytime exposure period of 8:00 am to
23 8:00 pm, consistent with CASAC advice in the last review (Henderson, 2007, p. 3). In so doing,
24 we recognize the variability in daylight hours across the country and across the 63 seasons, such
25 that this period will not always include all the daylight hours in all areas. Thus, a focus on this
26 set period for all locations has the potential to underestimate the impact of 63 exposures in some
27 areas, although, given that available monitoring data indicates that the daily increase in Os
28 concentrations generally does not begin until after 8 am (U.S. EPA, 2013, section 3.6.3.2), as
29 well as our additional focus on the 3-month period with the highest index (as discussed below),
30 the extent of such underestimation is unclear. We note, however, the variability in daylight
31 hours nationwide in areas to which a W126-based secondary standard would apply as well as in
32 areas where studies have been conducted that form the basis for consideration of such a standard
33 is an important consideration in identifying the appropriate elements for a W126-based
34 secondary standard.
35 With regard to identification of the seasonal period over which to cumulate exposures,
36 EPA notes that a plant is vulnerable to 03 pollution as long as it has foliage and is
6-6
-------
1 physiologically active (U.S. EPA, 2013, section 9.5.3, p. 9-112), i.e. during its growing season.
2 The length of vegetative growing seasons varies depending on the type or species of vegetation
3 and where it grows. For example, as discussed in the ISA, annual crops are typically grown for
4 periods of two to three months while perennial species may be photosynthetically active longer,
5 and up to 12 months each year for some species (U.S. EPA 2013, section 9.5.3, p. 9-112). In
6 general, the period of maximum physiological activity and thus, potential Os uptake for
7 vegetation coincides with some or all of the intra-annual period defined as the 63 season, which
8 varies on a state-by-state basis (U.S. EPA, 2013, Figure 3-24, p. 3-83). This is because the high
9 temperature and high light conditions that typically promote the formation of tropospheric O?,
10 also promote physiological activity in vegetation (U.S. EPA, 2013, section 9.5.3, p. 9-112).
11 The exposure periods used in studies of Os effects on vegetation reflect this
12 understanding, with crop studies typically using shorter seasonal exposure periods, while studies
13 of longer lived trees and other perennial vegetation often extend for the entire annual growing
14 season or in some cases over multiple growing seasons. Specifically, the ISA notes that "[m]ost
15 of the crop studies done through NCLAN had exposures less than three months with an average
16 of 77 days. Open-top chamber studies of tree seedlings, compiled by the EPA, had an average
17 exposure of just over three months or 99 days. In more recent FACE experiments, SoyFACE
18 exposed soybeans for an average of approximately 120 days per year and the Aspen FACE
19 experiment exposed trees to an average of approximately 145 days per year of elevated Os,
20 which included the entire growing season at those particular sites. Further, the U.S. Forest
21 Service and federal land managers have typically used the 6 months from April through
22 September as the accumulation period (U.S, EPA, 2013, section 9.5.3.2, p. 9-112). However,
23 despite the possibility that plants may be exposed to ambient 63 longer than 3 months in some
24 locations, the ISA notes that "[t]he exposure period in the vast majority of 63 exposure studies
25 conducted in the U.S. has been much shorter than 6 months..." and "there is generally a lack of
26 exposure experiments conducted for longer than 3 months" (U.S. EPA, 2013, section 9.5.3.2, p.
27 9-112)
28 As a result, analyses of effects in terms of the W126 exposure index have typically
29 defined the index in terms of a 3-month exposure period or at least in terms of periods shorter
30 than 6 months (e.g. SoyFACE, AspenFACE) (U.S, EPA, 2013, p. 112). In addition, the O3
31 season within which Os monitoring is required is shorter than 6 months in many areas in the
32 country, yet longer than 3 months in all locations (U.S. EPA, 2013, section 3.6.2, p. 3-83). Thus,
33 in the last review, the EPA proposed use of the consecutive 3-month period within the O^ season
34 having the maximum W126 index value. A 3-month exposure period was also supported by
35 CAS AC in advice provided both during the last review and on the 2010 proposed reconsideration
36 (Henderson, 2006; Samet, 2010). In revisiting this issue in the current review, the EPA
6-7
-------
1 conducted a new analysis to further inform the consideration of the most appropriate seasonal
2 accumulation period (U.S. EPA, 2013, section 9.5.3). This analysis calculated and compared the
3 3- and 6-month maximum W126 values for over 1,200 AQS and CASTNET EPA monitoring
4 sites for the years 2008-2009. The two accumulation periods were found to have highly
5 correlated metrics (U.S. EPA, 2013, Figure 9-13). The two accumulation periods were centered
6 on the yearly maximum for each monitoring site, and it is possible that this correlation would be
7 weaker if the two periods were not temporally aligned (U.S. EPA, 2013, section 9.5.3). The
8 analysis indicates that in the U.S., W126 cumulated over 3 months and W126 cumulated over 6
9 months are proxies of one another, as long as the period in which daily W126 is accumulated
10 corresponds to the seasonal maximum. Therefore, it is expected that either statistic will predict
11 vegetation response equally well.
12 • Are there other aspects of the form that are appropriate to consider that
13 contribute to welfare protection provided by a cumulative seasonal standard?
14 In considering a secondary standard in terms of the W126 exposure index with daylight
15 12-hour averaging time and maximum consecutive 3-month cumulative exposure period, we find
16 it appropriate to evaluate the protection that might be afforded by a form limited to a single year
17 or one that is based on evaluation of W126 index values across multiple years. In so doing, we
18 are mindful that the protection provided by the secondary standard derives from the combination
19 of all elements of the standard. Accordingly, the discussion below explores the information
20 relevant to consider in conjunction with identification of standard level, as well as whether there
21 is a lack of support in the current information for either single or multi-year options.
22 In considering an annual form of a standard, we particularly take note of Cb-induced
23 vegetation effects that can occur as a result of a single year's exposure. These include visible
24 foliar injury symptoms, growth reduction in annual species, and crop yield loss in annual crops.
25 The following discussion considers these effects, in the context of their potential public welfare
26 significance, with regard to the extent to which a W126-based standard with an annual form or
27 one based on evaluation across multiple years may be able to provide appropriate protection.
28 Visible foliar injury that occurs on public lands that have been afforded special
29 protections (e.g., Class I areas) is an annual effect with potential public welfare significance,
30 particularly when the injury is of some severity. However, we recognize the complexity
31 associated with defining Os exposure conditions that might consistently provide appropriate
32 protection from this endpoint. As summarized in section 5.4.2 and described more completely in
33 the WREA, over 80% of USFS biosites show no foliar injury in a wide variety of tree species
34 under air quality conditions (2006-2010) that include a broad range of W126 index values (U.S.
35 EPA 2014, section 7.2). The WREA USFS biosite analysis additionally suggested a reduction in
36 prevalence of sites with visible foliar injury with W126 index values below approximately 10
6-8
-------
1 ppm-hrs. However, the full body of evidence indicates wide variability in this endpoint, such
2 that although evidence shows visible foliar injury can occur under very low cumulative Os
3 concentrations, ".. .the degree and extent of visible foliar injury development varies from year to
4 year and site to site..., even among co-members of a population exposed to similar 03 levels, due
5 to the influence of co-occurring environmental and genetic factors" (U.S. EPA 2013, section
6 9.4.2, p. 9-38). In particular, drought conditions which can frequently occur with high seasonal
7 Oj, concentrations can potentially offer some protection so that the visible foliar injury that might
8 be expected based only on 63 exposure conditions does not develop. Accordingly, staff
9 recognizes the challenges associated with characterizing the role of relative soil moisture, and its
10 appreciable spatial and temporal variability. For example, the WREA analysis of the
11 relationship between foliar injury at USFS biosites and soil moisture data found foliar injury to
12 be less prevalent in years when sites were drier (U.S. EPA, 2014, section 7.2 and Figure 7-10).
13 The uncertainties associated with describing the potential for the occurrence of foliar injury and
14 its severity or extent of occurrence for any given set of Os exposure conditions, limit our ability
15 to identify annual ambient 63 exposure conditions that might be expected to provide an
16 appropriate degree of protection. Thus, the evidence for this endpoint does not appear to provide
17 support for the necessity of consideration of a single year form.
18 Another annual effect with the potential for public welfare significance is that of reduced
19 yields in annual commodity crops. It has been well documented that certain important
20 commodity crops are sensitive to O?, and that achieving the optimum level of crop yields is of
21 special importance to the public welfare. Although the impact of this effect can be monetized,
22 determining what the optimum yields are is less straightforward. This is because optimizing the
23 public welfare associated with annual crop production is not necessarily the same as maximizing
24 the annual yield but instead by achieving the maximum levels of producer and consumer
25 surpluses. These latter objectives depend less on the magnitude of the annual 03 exposure and
26 more on the management strategies employed by the agricultural community during that year,
27 strategic planting choices that are made from year to year, and environmental factors (e.g.,
28 drought) which can in large part offset these concerns (U.S. EPA, 2014, Chapter 6, section 6.5).
29 Thus, since the optimum annual yield is not necessarily equivalent to the greatest public welfare
30 value, it is difficult to know what degree of protection from Os would be requisite in any given
31 year. Thus, as in the 2008 review, we do not believe that a public welfare benefit of optimizing
32 crop yields by protection of commodity crops from annual 63 impacts is necessarily achievable
33 through the secondary NAAQS (73 FR 16492, 16497). Further, depending on the level of
34 protection being considered for other vegetation effects (i.e. seedling biomass loss), crops may
35 indirectly receive protection. Thus, similar to the annual foliar injury effect, the uncertainties
36 associated with identifying at what point annual yield loss in commodity crops becomes adverse
6-9
-------
1 to the public welfare make it difficult to identify an appropriate degree of protection, particularly
2 when considering the effect of a single year.
3 In contrast to impacts on annual species that accrue in the single growing season in which
4 the Os exposures occur, annual effects in perennial species can be "carried over" into the
5 subsequent year where they affect growth and reproduction (U.S. EPA, 2013, pp. 9-43 to 9-44
6 and p. 9-86). In addition, when these effects occur over multiple years due to elevated Os
7 exposures across several years, they are compounded, increasing the potential for effects at the
8 ecosystem level and associated ecosystem services that may be of significance to the public
9 welfare. For example, in perennial species, Os-induced reduction in photosynthesis decreases
10 the amount of carbohydrate produced by and available to the plant for growth and storage. The
11 plant often responds to decreases in carbohydrate production by reallocating the remaining
12 carbohydrate to above ground growth at the expense of storage in and growth of the roots (U.S.
13 EPA, 2013, section 9.4.3.1). When fewer carbohydrates are available in the roots, less energy is
14 available for root-related functions such as water and nutrient acquisition. By inhibiting the
15 amount of carbohydrate transfer to the roots, 63 also can disrupt the symbiotic relationship that
16 exists between some plants and mycorrhizal fungi in the soil. Mycorrhizal fungi have been
17 shown to improve the uptake of nutrients, protect host plant roots against pathogens, and produce
18 plant growth hormones (U.S. EPA, 2007, 7.3.3.2). Thus, such effects of elevated 63 in a given
19 year may affect the plant's resistance in a subsequent year, with multiple such years having the
20 potential for larger impacts. However, where there is variation across years such that lower 03
21 years with lesser impact occur between elevated 63 years, it may be that the longer-term impacts
22 are tempered.
23 Effects of elevated 63 years on perennial plants, when they occur over several years, may
24 also contribute to effects on ecosystem services, i.e. alteration of below-ground biogeochemical
25 cycles, and alteration of both above- and below- ground terrestrial community composition and
26 terrestrial ecosystem water cycling (U.S. EPA, 2013, Table 9-19). Ozone has also been shown to
27 affect plant reproduction in numerous ways (U.S. EPA, 2007, 7.3.3.3; U.S. EPA, 2013, 9.4.3.1).
28 These effects, when they occur at sufficient magnitude for a single species, may result in
29 impaired recruitment and loss of the species from the stand or community. This has the potential
30 to change the community composition and biodiversity. If these effects occur in multiple plant
31 species and/or over multiple years, they can result in a reduction in the productivity and carbon
32 sequestration of terrestrial ecosystems. Such ecosystem-related effects and others discussed in
33 the ISA may be considered to reflect impacts of critical 03 exposures over the longer term. We
34 additionally note that as compared to intermittent (or single year) critical 63 exposures, multiple
35 years of such exposures might be expected to result in larger impacts on forested areas, i.e.
6-10
-------
1 increased susceptibility to other stressors such as insect pests, disease, co-occurring pollutants
2 and harsh weather, due to the compounding or carry-over effects on tree growth.
3 Given the above, we find it reasonable to conclude that the public welfare significance of
4 the effects that can occur as a result of multiple-year O^ exposures are greater than those
5 associated with a single year. Thus, to the extent that the Administrator's focus for public
6 welfare protection to be afforded by the secondary Os standard is on long-term effects that occur
7 in sensitive tree species in natural forested ecosystems including in federally protected areas such
8 as Class I areas or on lands set aside by States, Tribes and public interest groups to provide
9 similar benefits to the public welfare, a standard with a form that evaluates the cumulative
10 seasonal index across multiple years (in combination with an appropriate level) might be
11 considered to provide an appropriate match to the nature of Os-related effects on vegetation most
12 at risk of being adverse to the public welfare and thus, upon which the secondary 03 standard is
13 focused.
14 Although cumulative, seasonal exposure values of interest for vegetation effects have
15 been expressed in terms of a single season (e.g., W126 index values for a season's exposure of
16 tree seedlings), we recognize that it is also appropriate to consider a form that is evaluated over a
17 multiple-year period, such as three years (U.S. EPA, 2007; 72 FR 37901; 75 FR 3021). Such an
18 aspect to the form specifies the number of years for which monitoring data are considered in
19 judging attainment with a standard. For example, under a standard with a single year form, a
20 monitor may be judged to meet the standard based on a single year of data, while under a
21 standard with a form requiring evaluation over a multi-year period, a monitor is not judged to
22 have met the standard until a complete multi-year record is available. For metrics with
23 substantial year-to-year variability, a multi-year evaluation period can provide stability in air
24 quality management programs, thus facilitating achievement of the protection intended by a
25 standard, while a standard with an annual form has the potential to contribute to challenges to
26 implementation of air pollution control programs in areas identified as not meeting the standard
27 that, through variability in the metric, meet the standard level in other years.
28 For a W126-based potential standard, the multi-year form identified for consideration in
29 the last review was the average cumulative seasonal metric over three consecutive years (75 FR
30 3027). Such a multi-year form remains appropriate to consider to provide stability to an
31 alternative secondary standard, just as the multi-year form provides for the current standard
32 (average over three years of annual fourth-highest daily maximum 8-hour average 63
r\
33 concentrations). In considering the issue of stability in the context of such a form, we first note
34 the inter-annual variability of seasonal W126 index, which is not unexpected given the logistic
2 See ATA III, 283 F. 3d at 374-75 (recognizing programmatic stability as a legitimate consideration in the
NAAQS standard-setting process).
6-11
-------
1 weighting function and also inter-annual variability in meteorological conditions which
2 contribute to 63 formation (see Appendix 2C). The staff analysis in Appendix 2C describes the
3 variability in annual W126 values in relation to variability in the 3-year average, which indicates
4 that a standard based on an annual W126 index would be expected to have a lower degree of
5 year-to-year stability relative to a standard based on a form that averages seasonal indices across
6 three consecutive years. A more stable standard can be expected to contribute to greater public
7 welfare protection by limiting year-to-year disruptions in ongoing control programs that would
8 occur if an area was frequently shifting in and out of attainment due to extreme year-to-year
9 variations in meteorological conditions. 283 F. 3d at 374. Thus, we recognize the public welfare
10 benefit of having a standard of a form with more year-to-year stability.
11 In considering the relative value of a multi-year form for an alternative secondary
12 standard in affording the desired protection, we also take note of the influence such a form might
13 be expected to have on air quality conditions and the associated public welfare protection. For
14 example a W126-based standard with an annual form would have a level defining the maximum
15 cumulative seasonal index. Where air quality is attained that reliably meets such a standard, the
16 cumulative seasonal index values would be at or lower than the level of the standard in all years
17 and, given the significant inter-annual variability in seasonal W126 values (described above),
18 would be appreciably below the standard level in many years. Alternatively, for a standard with
19 a form that averages the cumulative seasonal index values across three consecutive years, the
20 seasonal index would be above the level in some years, but would have to be below it in others
21 within the same 3-year period, restricting there to no more than two years out of three to have
22 indices above the level, and depending on magnitude of each year's index, potentially no more
23 than one.
24 In our consideration of this aspect of form for a cumulative seasonal secondary standard,
25 we additionally take note of advice from CASAC on this topic in prior reviews. For example, in
26 comments provided on the final Staff Paper in 2007, CASAC expressed the view that "[i]f multi-
27 year averaging is employed to increase the stability of the secondary standard, the level of the
28 standard should be revised downward to assure that the desired threshold is not exceeded in
29 individual years" (Henderson, 2007). Accordingly, in considering all elements for a revised
30 standard, including level, as well as form, we note the importance of considering the potential for
31 single-year 63 exposures that would result in adverse effects to public welfare. As noted above,
32 a standard with a form that averages across three years can also control year-to-year variability
33 and individual year concentrations. In so noting, we recognize the availability of air quality
34 analyses that can inform our consideration of the likely extent of such control (e.g., Appendix
35 2C). Additionally, we note the potential that, depending on type of effects and the magnitude
36 which may be judged adverse, low 03 years within each three-year period might play a
6-12
-------
1 countervailing role with regard to the extent of Os-attributable impacts that might pose public
2 welfare significance. The WREA analyses on the potential for biomass loss estimates based on a
3 3-year average to underestimate the cumulative impact on growth inform this point. Those
4 limited analyses indicate the potential for underestimation, although of relatively small
5 magnitude (U.S. EPA, 2014, section 6.2.1.4). We further recognize the role of public policy
6 judgments in drawing conclusions regarding adversity of effects to public welfare.
7 In light of the relationships described above, the appropriate level and form combination
8 will depend on the Administrator's objectives for the target level of protection. In articulating
9 these objectives it will be appropriate to evaluate the nature of the O^ induced effects and their
10 significance or importance to the public welfare, as well as the role that year-to-year exposure
11 variability can play in public welfare impacts. For example, to the extent that the
12 Administrator's priority for public welfare protection is against long-term effects in natural
13 forested ecosystems, and given uncertainty with regard to the extent to which effects from
14 single-year exposures may be judged adverse, articulation of a target level of protection in terms
15 of a longer-term condition, such as through a form that averages the index across multiple years,
16 may be appropriate.
17 • What does the available information indicate with regard to protection of welfare
18 from cumulative Os exposures that might be afforded by alternative secondary
19 standards based on the form of the current standard (a 3-year average of 4th high
20 8-hour average concentrations)?
21 In staff consideration of the primary standard in chapter 4, staff preliminarily concludes it
22 is appropriate to consider alternative primary standards of the same form and averaging time as
23 the current primary standard. Thus, although the discussion in this chapter, with regard to the
24 secondary standard, indicates the appropriateness of considering an alternative secondary
25 standard with a cumulative seasonal form, we also recognize that, to the extent that the
26 Administrator may find it more effective to control air quality using the same form for both the
27 primary and secondary standards, it may be practical to consider the extent to which a standard
28 in the form of the primary standard might be expected to also reduce and provide protection from
29 cumulative seasonal exposures of concern. For example, if a clear and robust relationship was
30 found to exist between 8-hour daily peak 03 concentrations and cumulative, seasonal exposures,
31 the averaging time and form of the current standard might be concluded to have the potential to
32 be effective as a surrogate. Addressing this point, the ISA describes the results of a recent focus
33 study that examined the diel variability in Oj, concentrations in six rural areas between 2007 and
34 2009 (U.S. EPA, 2013, pp. 3-131 to 3-133). The ISA reported that "[tjhere was considerable
35 variability in the diel patterns observed in the six rural focus areas" with the three mountainous
36 eastern sites exhibiting a "generally flat profile with little hourly variability in the median
6-13
-------
1 concentration and the upper percentiles", while the three western rural areas demonstrated a
2 "clear diel pattern to the hourly 63 data with a peak in concentration in the afternoon similar to
3 those seen in the urban areas", which was especially obvious at the San Bernardino National
4 Forest site, 90 km east of Los Angeles at an elevation of 1,384 meters (U.S. EPA, 2013, p. 3-
5 132). Thus, while the western sites that are influenced by upwind urban plumes may have
6 increased cumulative seasonal values coincident with increased daily 8-hour peak Os
7 concentrations, this analysis indicates that, in sites without such an urban influence (the eastern
8 sites in this analysis), such a relationship does not occur (U.S. EPA, 2013, section 3.6.3.2).
9 Thus, the lack of such a relationship indicates that in some locations, Os air quality patterns can
10 lead to elevated cumulative, seasonal 63 exposures without the occurrence of elevated daily
11 maximum 8-hour average Os concentrations (U.S. EPA, 2013, section 3.6.3.2). Further staff
12 notes that the prevalence and geographic extent of such locations is unclear, since as in the last
13 review, there continue to be relatively fewer monitors in the West, including in high elevation
14 remote sites. In considering the findings of this analysis, we additionally recognize, however,
15 that in general, the cumulative seasonal values for the eastern rural sites, where cumulative
16 seasonal 63 concentrations appear to be relatively less related to daily maximum 8-hour
17 concentrations, are lower than those of the western, urban-influenced sites.
18 In addition to the focus study described in the ISA (U.S. EPA, 2013, section 3.6.3.2), we
19 considered analyses of air quality monitoring data and air quality modeling analyses. Chapter 2
20 of this document characterizes recent monitoring data on O?, air quality in rural areas. While
21 approximately 80% of the 63 monitoring network is urban focused, about 120 rural monitors are
22 divided among CASTNET, NCore, and POMs sites (Chapter 2, pp. 2-2 to 2-3, Figure 2.1).
23 Specifically, as stated in chapter 2 "[although rural monitoring sites tend to be less directly
24 affected by anthropogenic pollution sources than urban sites, rural sites can be affected by
25 transport of O?, or 03 precursors from upwind urban areas and by local anthropogenic sources
26 such as motor vehicles, power generation, biomass combustion, or oil and gas operations" (U.S.
27 EPA, 2013, section 3.6.2.2). In addition, OT, tends to persist longer in rural than in urban areas
28 due to lower rates of chemical scavenging in non-urban environments. At higher elevations,
29 increased 63 concentrations can also result from stratospheric intrusions (U.S. EPA, 2013,
30 sections 3.4, 3.6.2.2). As a result, Os concentrations measured in some rural sites can be higher
31 than those measured in nearby urban areas (U.S. EPA, 2013, section 3.6.2.2) and the ISA
32 concludes that "cumulative exposures for humans and vegetation in rural areas can be
33 substantial, and often higher than cumulative exposures in urban areas" (U.S. EPA, 2013, p. 3-
34 120). These known differences between urban and rural sites suggest that there is the potential
35 for an inconsistent relationship between 8-hour daily peak OT, concentrations and cumulative,
6-14
-------
1 seasonal exposures in those areas where protection of Os-sensitive vegetation may be most
2 needed.
3 In addition, as was done in both the 1997 and 2008 reviews, staff has analyzed
4 relationships between Os levels in terms of the current averaging time and form and a W126
5 cumulative form, based on recent air quality data. One analysis describes the W126 index values
6 and current standard design values at each monitor for two periods: 2001-2003 and 2009-2011
7 (e.g., Appendix 2B, Figures 2B-2 and 2B-3). This shows that between the two periods, during
8 which broad scale 63 precursor emission reductions occurred, 63 concentrations in terms of both
9 metrics were reduced. There is a fairly strong, positive degree of correlation between the two
10 metrics (Appendix 2B).3 Focusing only on the latter dataset (2009-2011), it can be seen that at
11 monitors just meeting the current standard (3-year average fourth-highest daily maximum 8-hour
12 average concentration equal to 0.075 ppm), W126 values (in this case 3-year averages) varied
13 from less than 9 ppm-hrs to approximately 20 ppm-hrs (Appendix 2B, Figure 2B-3b). At sites
14 with a 3-year average fourth-highest daily maximum 8-hour average concentration at or below a
15 potential alternative primary standard level of 70 ppb, 3-year W126 index values range from 5 to
16 18 ppm-hrs, with very few above 15 ppm-hrs. An alternative presentation of such data for recent
17 3-year periods back to 2006 - 2008 indicate that among the counties with 03 concentrations that
18 met the current standard, the number with 3-year W126 index values above 15 ppm-hrs ranges
19 from fewer than 10 to 24 (Appendix 2B, Figure 2B-9). In general during this longer period,
20 W126 index values above 15 ppm-hrs and meeting the current standard were pre-dominantly in
21 Southwest region. As the first analysis in Appendix 2B (for the 2001-2003 and 2009-2011
22 periods) indicates, monitors in the West and Southwest tend to have higher W126 values relative
23 to their design values than do monitors in other regions. This pattern is noteworthy because the
24 Southwest region has a less dense monitoring network than regions in the Eastern US (see Figure
25 2-1), so that the extent to which this pattern occurs throughout these regions is uncertain.
26 Although single-year W126 index values were not separately analyzed in this analysis of the
27 monitor data, it indicates appreciable variation in cumulative, seasonal Os concentrations among
28 monitor locations meeting different levels of a standard of the current form.
29 Analyses of the WREA air quality scenarios indicate the potential for O^ precursor
30 emission reductions achieving 63 concentrations that just meet different 8-hour standards to
31 produce a significant reduction in 3-year W126 index values. For example, for the current
32 standard scenario, nearly all model-adjusted monitors are at or below an estimated 3-year
33 average W126 value of 15 ppm-hrs (as summarized in section 5.2.2 above and described in U.S.
3 Appendix 2B additionally observes that the program implemented for reducing precursor emissions,
especially NOx, appears to have been an effective strategy for lowering both design values and W126 values.
6-15
-------
1 EPA, 2014, Table 4-1). Those monitors above 15 ppm-hrs would be limited to large urban areas
2 in the southwestern U.S. (i.e., Phoenix, Los Angeles and Denver). When meeting a 4th high 8-
3 hour average scenario of 70 ppb, averaged across 3 years, nearly all monitors in the U.S. would
4 meet a 3-year W126 of 11 ppm-hrs, though some monitors in the southwest would remain
5 between 11 and 15 ppm-hrs. At 65 ppb, all locations are at or below 11 ppm-hrs. Thus, similar
6 to the monitoring analysis, the modeling analysis generally indicates reductions in W126 levels
7 with reduced Os concentrations in terms of the current standard averaging time and form. This
8 suggests that depending on the level for a standard of the current averaging time and form, a
9 degree of welfare protection may be afforded. The extent to which such protection provides
10 adequate public welfare protection additionally depends on the level of protection identified by
11 the Administrator for the public welfare in terms of the W126 index. In so noting, however, we
12 recognize the importance of also considering uncertainties in both the model-adjustment analyses
13 and those based on monitoring data. These uncertainties, including those related to monitor
14 coverage, the extent to which recent data can be expected to describe future relationships, and
15 modeling approaches, among others, should be kept in mind when assessing the strength of this
16 apparent relationship.
17 6.3 LEVEL
18 In considering potential levels for alternative secondary standards, we have taken into
19 account scientific evidence characterized in the ISA and exposure and risk estimates for different
20 air quality scenarios analyzed in the WREA, as well as the uncertainties and limitations in this
21 information. We consider this information together with regard to support for alternative
22 standards that might be appropriate to consider to provide the requisite protection from adverse
23 effects to public welfare. Drawing from section 6.2 above, our discussion here is primarily in
24 terms of the W126 sigmoidally-weighted metric, cumulated over the 12 daylight hour period
25 (8:00 am to 8:00 pm) for the consecutive three month period with the maximum index value
26 within the growing season. In addition to considering the information in the context of a single
27 growing season, we also consider it in the context of a form for this W126 metric averaged
28 across 3 consecutive growing seasons for reasons discussed in section 6.2 above.
29 In the discussion below, we turn first to consideration of the currently available scientific
30 evidence as assessed and characterized in the ISA. We then consider the WREA findings with
31 regard to vegetation, ecosystem and ecosystem services effects or risks estimated for different air
32 quality scenarios. We additionally take note of important uncertainties and limitations in the
33 evidence and exposure/risk analyses. Lastly, we take note of judgments to be made by the
34 Administrator in drawing conclusions regarding effects and risks that represent adverse
35 environmental effects to public welfare. In so doing, we identify key considerations with regard
6-16
-------
1 to the currently available evidence, exposure/risk information and associated uncertainties in
2 identifying levels that may be appropriate to consider for a cumulative seasonal secondary
3 standard. Such levels are described in section 6.5 below which describes preliminary staff
4 conclusions regarding alternative secondary standards appropriate to consider in this review.
5 • What does the currently available evidence indicate with regard to the range of
6 W126-based index values that may provide protection from vegetation effects of
7 03?
8 In considering this question in terms of a cumulative seasonal W126-based index, we
9 consider first quantitative evidence for O^ exposure effects on plant growth, productivity and
10 related endpoints. In so doing, we draw on the C-R functions derived from the evidence on tree
11 seedlings and crops assessed in the ISA which are further described in the WREA and
12 summarized in sections 5.2.1 and 5.3.1 (U.S. EPA 2013, section 9.6; U.S. EPA, 2014, section
13 6.2). It is important to note that these functions are used to provide estimates of growth and yield
14 reduction in tree seedlings and crops that might be expected to result from exposure over a single
15 growing season to various 63 concentrations expressed in terms of a W126 index (Figure 5-1
16 above).
17 Table 6-1 below presents estimates of relative yield and biomass loss for the studied
18 species of crops and tree seedlings, respectively, for a growing season exposure to a number of
19 W126 levels. In this table, we have included observations related to 2% relative biomass loss in
20 tree seedlings and 5% relative crop yield loss. These values are consistent with advice from
21 CASAC thus far in this review (Frey and Samet, 2012) on factors on which to base consideration
22 of levels for a secondary standard (see section 6.4 below), and with values given focus in the
23 1996 consensus-building workshop, described below. The observations in Table 6-1 for each
24 W126 index value include relative biomass or crop yield loss estimates for the median across the
25 studied species, as well as the proportion of species with estimates below 2%, 5% or 10%, as
26 examples of benchmarks that may be of interest in different contexts.
27 Staff recognizes Table 6-1 as a useful way to consider the magnitude of biomass and crop
28 yield loss estimates across the 12 and 10 studied species, respectively, for the purposes of
29 informing consideration of potential levels for alternative secondary standards based on the
30 W126 index. The different index levels can be considered with regard to the magnitude of
31 median species response and the proportion of species with responses expected to be below the
32 various benchmarks. For example, at an 63 exposure equivalent to a W126 index value of 7
33 ppm-hrs, the median tree seedling response across the 12 studied species is projected to be near
34 or below 2% reduction in growth compared to that for the control plants exposed to W126 index
35 values of zero (Table 6-1; U.S. EPA, 2014, section 6.2, Appendix 6A). Seven of the 12 species
36 are projected to have smaller than 2% reduction in growth, while three other species are
6-17
-------
1 projected to have a reduction between 2 and 5%, for a total of 10 of 12 species projected to have
2 less than 5% reduction in growth, at a W126 index value of 7 ppm-hrs. The remaining two
3 species, black cherry and cottonwood, are projected to have reductions just over 15% and 40%,
4 respectively, at 7 ppm-hrs and above (Table 6-1). The median growth reduction response
5 projected for a W126 index value of 15 ppm-hrs is approximately 5%, with responses less than
6 2% projected for five of the twelve studied species and less than 5% for six, while less than 10%
7 reduction in growth is projected for ten of the twelve species. Somewhat similar estimates are
8 projected for a W126 index value of 17 ppm-hrs, with the median growth reduction response
9 projected to be 6%, and also with five of the twelve species less than 2%, nine less than 10% and
10 a tenth just over 10% (Table 6-1). Crop yield loss estimates are also described in Table 6-1.
11
6-18
-------
1 Table 6-1. Seedling and crop growth or yield reductions estimated for
2 season.
exposure over a
W126 value
for exposure
period
21 ppm-hrs
19 ppm-hrs
17 ppm-hrs
15 ppm-hrs
13 ppm-hrs
11 ppm-hrs
9 ppm-hrs
7 ppm-hrs
Tree seedling growth reduction (biomass loss)A
BMedian species w. 7.7 % loss (varying lower)
3/12 species w. < ~2%c loss; 5/12 w. < 5% loss
7/12 species w. <10% loss; 10/12 <15% loss
BMedian species w. 6.8 % loss (varying lower)
3/12 species w. < -2% loss; 5/12 w. <5% loss
7/12 w. <10% loss; 10/12 w. <15% loss
BMedian species w. 6.0 % loss (varying lower)
5/12 species w. < -2% loss; 5/12 w. <5% loss
9/12 w. <10% loss; 10/12 w. <15% loss
BMedian species w. 5.2 % loss (varying lower)
5/12 species w. < -2% loss; 6/12 w. <5% loss
10/12 species w. <10% loss
BMedian species w. 4.4% loss (varying lower)
5/12 species w. < -2% loss; 7/12 w. <5% loss
10/12 species w.<10% loss
BMedian species w. 3.5% loss (varying lower)
5/12 species w. < -2% loss; 8/12 w. <5% loss
10/12 species w. <10% loss
BMedian species w. 2.5% loss (varying lower)
5/12 species w. < -2% loss; 10/12 w. <5% loss
BMedian species w. <2% loss (varying lower)
7/12 species w. < -2% loss; 10/12 w. <5% loss
Note: Most sensitive
species are
cottonwood (relative
biomass loss
estimates > 40% for
W126>7) and black
cherry (relative
biomass loss
estimates > 15% for
W126>7).
Crop yield lossA
Median species0 = 7.7 % loss
4/10 species w. < 5% loss
3/10 species w. >5,<10% loss
3/10 species w. >10,<20% lossE
Median species0 = 6.4 % loss
5/10 species w. < 5% loss
3/10 species w. >5, <10% loss
2/10 species w. >10,<20% loss
Median species0 = 5.1 % loss
5/10 species w. < 5% loss
3/10 species w. >5, <10% loss
2/10 species w. >10,<20% loss
Median species0 <5% loss
6/10 species w. < 5% loss
4/10 species w. >5, <10% loss F
Median species0 <5% loss
7/10 species w. < 5% loss
3/10 species w. >5, <10% loss
Median species0 <5% loss
9/10 species w. < 5% loss
1/10 species w. >5, <10% loss a
Median species0 <5% loss
All species < 5% loss
Median species0 <5% loss
All species < 5% loss
A Estimates here are based on the C-R functions described in WREA, section 6.2 and discussed in section 5.2.1
B This median value is the composite median from WREA, Figure 6-4 (also discussed in section 5.2.1). The note in parentheses
refers to the median of the stochastic sampling estimate shown in WREA Figure 6-4 (and Figure 5-2 in this document).
c In making comparisons, biomass loss estimates described as < 2 refer to values < 2.04.
D This median is the composite median from WREA, Figure 6-5 (also discussed in section 5.2.1).
E Kidney bean , potato, and winter wheat >15%.
F Cotton, kidney bean, potato, & soybean >5%.
G Kidney bean at 5.7% loss.
4 With respect to W126 index values associated with different degrees of protection for
5 visible foliar injury, we first recognize the value of this endpoint in its long-standing use as a
6 well established bioindicator of Os exposure, as described in the ISA (U.S. EPA 2013, section
7 9.4.2).
6-19
-------
1 In addition to the role of visible foliar injury as an indicator, we note that the aesthetic
2 aspects of visible foliar injury itself have the potential to be important to public welfare (as
3 described in section 5.4 above). Analyses such as those presented in the WREA, and
4 summarized in section 5.4.2 above, provide approaches by which foliar injury observations
5 under existing air quality may be analyzed. However, there is very little information on which to
6 base judgments of the prevalence and/or severity of injury that might be considered adverse. We
7 further recognize that there is substantial variability such that "the degree and extent of visible
8 foliar injury development varies from year to year and site to site ... even among co-members of
9 a population exposed to similar Os levels, due to the influence of co-occurring environmental
10 and genetic factors" (U.S. EPA, 2013, p. 9-38). Thus, based on consideration of the evidence,
11 staff recognizes the lack of a consistent or generally predictable relationship between particular
12 W126 exposures and visible foliar injury incidence, given the influence of and variability in
13 relative soil moisture and other environmental factors. We additionally note uncertainty in what
14 can be concluded from foliar injury in relation to plant health, productivity and ecological
15 function as "it is not presently possible to determine, with consistency across species and
16 environments, what degree of injury at the leaf level has significance to the vigor of the whole
17 plant" (U.S. EPA, 2013, p. 9-39). Thus, on the basis of the evidence, we are not able to identify
18 a range of appropriate W126 index values.
19 In further considering the available information pertaining to the question above, we
20 additionally recognize conclusions that have been drawn by expert committees with regard to
21 these endpoints (i.e., tree seedling growth, crop yields and visible foliar injury). For example, in
22 their review of staff documents during the Os NAAQS review completed in 1997 review, the
23 CAS AC 63 panel members expressed a wide range of opinions on aspects of the evidence
24 important to consider in judging the adequacy of the 63 secondary standard and in considering
25 the form and level that would be appropriate for a secondary 03 standard (Wolff, 1996).
26 Subsequent to CAS AC meetings in 1995 on this topic, a consensus-building workshop sponsored
27 by the Southern Oxidant Study group was held on the topic of the Os secondary standard in
28 January 1996 (Heck and Cowling, 1997). This workshop was attended by 16 scientists with
29 backgrounds in agricultural, managed forest, natural systems, and air quality, all of whom were
30 leaders in their fields and whose research formed the basis of much of the research examined in
31 the 1996 Criteria Document. These scientists expressed their judgments on what standard
32 level(s) would provide vegetation with protection from (Vrelated adverse effects that would be
33 adequate, in their view.4'5 As the 1997 workshop publication indicates, the scientists at the 1996
4 At the time of the workshop, the secondary O3 standard being reviewed by EPA was a 1-hour average of
0.08 ppm (identical to the primary standard at that time). In 1997, EPA concluded the review by revising both
standards to a longer averaging time of 8 hours with a level of 0.08 ppm (62 FR 38856).
6-20
-------
1 workshop also reached consensus views regarding the types of exposures that were important in
2 eliciting plant response and the types of metrics that were best at predicting these responses
3 (Heck and Cowling, 1997). Before coming to agreement on daily and seasonal durations and
4 forms pertinent to a distinct secondary standard, the participants discussed and identified
5 endpoints to consider for natural, forest and agricultural ecosystems.6 With regard to form of the
6 standard, participants concurred with either the SUM06 or W126 metrics, with consensus finally
7 reached for SUM06, with some qualification regarding implications for a threshold. The
8 participants identified the ranges they felt should be considered for each of three endpoints.
9 Overall, the SUM06 values ranged from 8 to 20 ppm-hr (corresponding to W126 values ranging
10 from 5 to 17 ppm-hr, based on EPA analysis focused on conditions in NCLAN studies).7 This
11 overall range reflected ranges for each of the three endpoints, with the following considerations
12 (Heck and Cowling, 1997).
13 - Crops (yield reductions): SUM06 of 15-20 ppm-hrs (13-17 ppm-hrs, W126). This
14 range was recognized to generally consider <10% yield loss in more than 75% of
15 species.
16 - Trees (growth effects): SUM06 of 10-16 ppm-hrs (7 to 14 ppm-hrs, W126). This
17 range was recognized to generally consider 1-2% per year growth reduction; in so
18 doing,the group identified a need to consider the potential for year-to-year
19 compounding of impacts in long-lived perennial species.
20 - Visible Foliar Injury: SUM06 of 8 to 12 ppm-hrs (5 to 9 ppm-hrs, W126).
21 Since the publication of 1996 workshop report and conclusion of the 1997 NAAQS
22 review, the evidence base has continued to expand as described in the 2006 CD and 2013 ISA
23 (U.S. EPA, 2006; U.S. EPA, 2013). With regard to tree growth effects and crop yield reductions,
24 results of additional studies conducted in the field have confirmed the tree seedling biomass loss
25 and crop yield loss concentration-response relationships derived from earlier studies that used
26 open top chambers (U.S. EPA 2013, section 9.6). Further, similar data, although from an
5 The workshop publication describes the primary objective for the workshop as having been to assemble
knowledgeable scientists to develop a group consensus on "various critical components associated with a possible
revised secondary ozone standard" (Heck and Cowling, 1997).
6 For natural ecosystems, they focused on foliar injury as an indicator. For forest ecosystems, concluded
the data did not support selection of an indicator of effects on forest structure or function. As a result, they
identified two indicators pertinent to the systems: growth effects on seedlings from species of natural forest stands
(1-2% per year reduction), and growth effects on seedlings and saplings from tree plantations (1-2% per year
reduction). For agricultural systems, the participants focused on protection against crop yield reductions, with their
acknowledgment of high uncertainties at 5% leading them to a crop yield endpoint of 10% yield reduction (Heck
and Cowling, 1997).
'During the last review, W126 index values corresponding to the SUM06 values cited in the report were
estimated using the NCLAN crop loss data, a key dataset considered by workshop participants (see Appendix 7B of
2007 Staff Paper; Appendix 6A of this document).
6-21
-------
1 ambient gradient study (outside New York City) rather than a controlled experiment, are
2 available for an additional, apparently quite sensitive species, the cottonwood (Gregg, 2003).
3 In the 2008 review, CASAC provided comments related to a cumulative seasonal
4 secondary standard in the context of their comments on the draft and final Staff Papers and on
5 the final decision (Henderson, 2006; Henderson, 2007; Henderson, 2008). In all instances, they
6 conveyed support for establishment of a distinct secondary standard with a cumulative seasonal
7 form. While the EPA, in the 2007 Staff Paper and 2007 notice of proposed rulemaking,
8 recognized a broader range of W126 values as appropriate for consideration with regard to a
9 distinct secondary standard, the CASAC Panel focused on a range they described as
10 approximately equivalent to that identified by the 1996 workshop participants (Henderson, 2007,
11 pp. 3, C-27).8 In the CASAC Panel 2006-2007 advice on levels for such a standard, their
12 suggestion was a focus on levels for a W126 index approximately equivalent to a SUM06 range
13 of 10 to 20 ppm-hr (Henderson, 2006, 2007, 2008), which they estimated in 2007 to be a range
14 from 7 (or 7.5) to 15 ppm-hrs. Based on their consideration of the information available in that
15 review (e.g., with regard to potential magnitude of effects across multiple years), the CASAC
16 Panel further advised that "[i]f multi-year averaging is employed to increase the stability of the
17 secondary standard, the level of the standard should be revised downward to assure that the
18 desired threshold is not exceeded in individual years" (Henderson, 2007, p. 3). The CASAC
19 advice provided on the 2010 proposed reconsideration and thus far in this review is summarized
20 in section 6.4 below.
21 In considering the evidence briefly summarized above in the context of levels for a
22 W126-based standard, we recognize that given the different types of (Vinduced effects, genetic
23 variability within and between species, and environmental modifiers of effects that also
24 contribute to variability, it is not feasible to identify a range of cumulative seasonal exposures
25 from the vegetation effects evidence which would provide a consistent degree of protection for
26 all species. In so doing, we note the importance of considering this evidence in several
27 dimensions that pertain to judgments required by the Administrator regarding public welfare
28 significance. For example, we take note of the usefulness of considering the cumulative seasonal
29 exposure at which the median species response or the majority of the species' responses are
30 expected to be below minimal response benchmarks of interest and at which only a very few
31 species' responses are expected to exceed more substantial response benchmarks. Before
32 articulating such considerations with regard to specific benchmarks and index values, we first
33 consider the WREA findings in the context of the following question.
8 Appendix C of the March 26, 2007 CASAC letter used a 2001 ambient concentration dataset and other
factors, rather than study data considered in the 1996 workshop, in estimating an "equivalency" between the two
indices.
6-22
-------
1 • What are the nature and magnitude of risks to vegetation estimated for the
2 average W126 index scenarios evaluated in the WREA, and what is the
3 magnitude of risk reduction from risks estimated for air quality conditions
4 estimated for the current standard?
5 In considering the WREA estimates here we take note of uncertainties in the extent to
6 which the results for each modeled air quality scenario represent cumulative seasonal Os
7 exposures that would be expected to occur across the three years represented in each scenario. In
8 general, each scenario is represented by a dataset of 3-year average W126 index values across
9 the national modeling area. Thus, the results estimated for the various analyses performed do not
10 reflect any year-to-year variability that would be expected in single year results. Rather, they
11 reflect average estimates for the three year period modeled. Such estimates do not account for
12 the potential for compounding of effects in perennial species that has been discussed previously.
13 Limited analyses in the WREA describe the potential for the WREA estimates to underestimate
14 cumulative biomass-related effects in perennial species (as noted in sections 6.2 and 5.2.2 above
15 and described in detail in U.S. EPA, 2014, chapter 6, 6.2.1.4). This potential for underestimation
16 is recognized in the context of the uncertainties associated with other aspects of the different
17 analyses in section 6.9 of the WREA (e.g., U.S. EPA, 2014, Table 6-27). We additionally note
18 that the limited WREA compounding analyses do not take into account other variables which can
19 affect the magnitude of these effects in the field.
20 In considering the question posed above, we focus particularly on WREA estimates
21 related to 63 effects on plant biomass and associated ecosystem services effects. In so doing, we
22 note the relationships among effects on individual plants to other ecosystem components and
23 functions, such as carbon sequestration and air pollutant removal (U.S. EPA 2013, section
24 9.4.3.4; U.S. EPA, 2014, sections 6.6 and 6.7), as well as market responses to changes in timber
25 and agricultural production (U.S. EPA, 2014, sections 6.3 and 6.5). We additionally recognize
26 other biomass-related responses, such as non-timber forest products, and other Os-attributable
27 ecosystem responses for which we have primarily qualitative characterizations of impacts (U.S.
28 EPA, 2014, chapters).
29 We turn first to the WREA estimates for a range of effects related to biomass loss, which
30 are based on application of the C-R functions for seedlings of 12 tree species described in the
31 ISA (U.S. EPA, 2013, section 9.6.2) and the WREA (U.S. EPA, 2014, section 6.2). First we
32 note (as considered above) the range of responses for the individual species for which C-R
33 functions have been developed. These twelve species vary appreciably in sensitivity of growth
34 reduction (in terms of relative biomass loss, or RBL) in response to O?, exposure. For example,
35 2% seedling biomass loss is estimated to occur following exposure represented by W126 index
36 values below 10 ppm-hours for seven of the studied species and at or above approximately 20
6-23
-------
1 ppm-hours in the other five studied species (Figure 5-1 above). The WREA presents three
2 approaches for characterizing a median response across the species (Figure 5-2; U.S. EPA 2014,
3 section 6.2.1.2). Across the three approaches, the median seasonal W126 index value for which
4 a 2% biomass loss is estimated in seedlings for the studied species ranges between approximately
5 7 and 14 ppm-hrs (see description in section 5.2.1 above).
6 We additionally consider the WREA estimates of overall ecosystem-level effects from
7 biomass loss considering the 12 studied species together (U.S. EPA 2014, section 6.8). The
8 WREA analysis used the species-specific biomass loss C-R functions, information on prevalence
9 of the studied species across the U.S., and a weighting approach based on proportion of the basal
10 area within each grid cell that each species contributes. A weighted RBL value for each grid cell
11 is generated by weighting the RBL value for each studied tree species found within that grid cell
12 by the proportion of basal area it contributes to the total basal area of tree species within the grid
13 cell, and then summing those individual weighted RBLs. Based on the average W126 values
14 estimated for the air quality scenario just meeting the current standard across the contiguous
15 U.S., the WREA estimates 0.8 percent of the total geographic area to have a weighted relative
16 biomass loss above 2% (Table 6-2 below; U.S. EPA 2014, Table 6-24). In the W126 air quality
17 scenarios for 15, 11, and 7 ppm-hrs (average across three years), the percent of total area having
18 weighted relative biomass loss greater than two percent was 0.7 percent, 0.5 percent and 0.2
19 percent, respectively (Table 6-2 below; U.S. EPA 2014, Table 6-25). In considering these
20 estimates, we note that the values for percentages of basal area include many grid cells in which
21 none of the 12 studied species are found. Thus, since only 64 percent of grid cells contained one
22 or more of the 12 species, these values are likely to be low and illustrate the potential
23 uncertainties associated with the limited number studied tree species for which C-R functions
24 exist (U.S. EPA, 2014, section 6.8).
25
6-24
-------
1 Table 6-2. Percent of assessed geographic area exceeding 2% weighted relative biomass
2 loss in WREA air quality scenarios.
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
Percent of total area with
wRBL>2%
Air Quality Scenarios
Conditions just
meeting the current
standard*
0.8 %
W126 index scenarios6
15ppm-hrs
0.7 %
11 ppm-hrs
0.5 %
7 ppm-hrs
0.2 %
A This analysis uses air quality values that are estimated per model grid cell using the W126 value assigned to the grid cell
based on application of the VNA method to the monitor-location W126 values that are the average at that location across
the 3 years of W126 values for the model-adjusted dataset that just meets the current standard (fourth-highest daily
maximum 8-hour concentration, averaged over 3 consecutive years of 75 ppb).
B The national distribution of W126 values within model grid-cells for each scenario reflects model adjustment of 2006-
2008 Os concentrations at monitoring sites such that the average W126 index at the controlling location in each of the
modeling regions just meets the scenario target index value, followed by application of the VNA interpolation methodology
(see U.S. EPA, 2014, section 4.3.4.1 and Appendix 4A).
We next consider the wRBL estimates from the WREA analysis of 145 (of the 155)
federally-designated Class I areas for which there was sufficient information regarding 63-
sensitive species (U.S. EPA, 2014, section 6.8.1, Table 6-26, Appendix 6E). These 145 parks
had at least one Os-sensitive tree species for which a C-R function for RBL was available. Using
the C-R functions for the species found within each park, the WREA calculated an average
wRBL value for each park for the 3-year average W126 index values estimated in those locations
for the current standard and three W126 air quality scenarios. Under conditions model-adjusted
to just meet the current standard, the average wRBL in 2 of the 145 parks is estimated to be
above 2%, as presented in Table 6-3 below. We compare this estimate to those for the W126
scenarios. For the W126 scenarios of 15 and 11 ppm-hrs, the estimated weighted RBL is greater
than 2% in two of the 145 parks, while it is great than 2% in only 1 park for the 7 ppm-hrs
scenario.
6-25
-------
1 Table 6-3. Number of Class I areas (of 145 assessed) with weighted relative biomass loss
2 greater than 2%.
Number of Class 1 areas with
wRBL>2%
Air Quality Scenarios
Conditions just meeting
the current standard*
2
3-Year Average W126 index
15 ppm-hrs
2
11 ppm-hrs
2
scenarios6
7 ppm-hrs
1
A The wRBL is estimated per model grid cell (in which there are any of the 12 studied species) from W126 value assigned to the
grid cell based on application of the VNA method to the monitor-location W126 values that are the average at that location across
the 3 years of W126 values for the model-adjusted dataset that just meets the current standard (fourth-highest daily maximum 8-
hour concentration, averaged over 3 consecutive years of 75 ppb).
B The national distribution of W126 values within model grid-cells for each scenario reflects model adjustment of 2006-2008 Os
concentrations at monitoring sites such that the average W126 index at the controlling location in each of the modeling regions
just meets the scenario target index value, followed by application of the VNA interpolation methodology (see U.S. EPA, 2014,
section 4.3.4.1 and Appendix 4A).
4 With respect to crops, based on the C-R functions described in the ISA and additionally
5 analyzed in the WREA, a 5% median relative yield loss (RYL) for all studied crop cases occurs
6 within the range of W126 index values from 12-17 ppm-hrs (U.S. EPA 2014, Figure 6-5). The
7 species-specific C-R functions project less than 5% yield loss for 6 out of 10 species for O^
8 exposure equivalent to a seasonal W126 index value of 15 ppm-hrs and for all of the 10 studied
9 species after an exposure equivalent to a seasonal W126 index of 7 ppm-hrs (Table 6-1 above,
10 U.S. EPA 2014, section 6.2 and Appendix 6A). For soybeans, less than 5% yield loss was
11 estimated for the W126 index value of 12 ppm-hrs (U.S. EPA 2014, Figure 6-3). The WREA
12 estimates of crop yield loss for the modeled air quality scenarios are summarized in Table 6-4
13 below (details are provided in U.S. EPA 2014, section 6.5.1 and Appendix 6B). For the recent
14 air quality conditions scenario, the means for all crops were less than 5% loss across all states.
15 Crop yield loss estimates for all states were also less than 5% in the air quality scenario
16 representing conditions just meeting the current standard (U.S. EPA, 2014, section 6.5.1 and
17 Appendix 6B).
18
6-26
-------
1 Table 6-4. Estimated mean yield loss (and range across states) due to
2 important crops.
exposure for two
Crop
Corn
Soybean
Air r^iolitw ^f*onorirto
rMl \J(UQII Ly WwGI IQI 1 WO ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Recent
Conditions
(2006-2008)
<5%c
(0.01-0.88)
<5%
(0.69-8.30)
Conditions just
meeting the current
standard*
<5%
(0.0-0.01)
<5%
(0.01-1.39)
Average W126 index scenarios6
15ppm-hrs
<5%
(0.0-0.01)
<5%
(0.01-1.13)
11 ppm-hrs
<5%
(0.0-0.0)
<5%
(0.01 - 0.75)
7 ppm-hrs
<5%
(0.0-0.0)
<5%
(0.01 - 0.59)
A The crop yield loss is estimated per grid cell (and per FASOMGHG region) from W126 value assigned to the cell based on
application of the VNA method to the monitor-location W126 values that are the average at that location across the 3 years of
W126 values for the model-adjusted dataset that just meets the current standard (fourth-highest daily maximum 8-hour
concentration, averaged over 3 consecutive years of 75 ppb).
B The national distribution of W126 values within grid cells for each scenario reflects model adjustment of 2006-2008 Os
concentrations at monitoring sites such that the average W126 index at the controlling location in each of the modeling regions
just meets the scenario target index value, followed by application of the VNA interpolation methodology (see U.S. EPA 2014
section 4.3.4.1 and Appendix 4A).
c Mean yield loss is the mean across modeling units. The range presented in parentheses below the mean represents the
minimum and maximum estimates across modeling units (U.S. EPA 2014, Appendix 6B).
3
4 As explained above, however, comparisons of the WREA's air quality scenarios for the
5 national-scale estimates of timber production and consumer and producer surpluses are not
6 straightforward to interpret due to market dynamics. Estimates for the recent conditions and
7 current standard scenarios are compared to the three W126 scenarios. In general, substantially
8 greater economic surpluses (approximately 51 billion in terms of 2010 dollars) are estimated
9 from the comparison of the recent conditions (2006-2008) scenario to the current standard
10 scenario. The vast majority of these economic surpluses are estimated for agricultural
11 production. Differences of the average W126 scenarios from the current standard scenario are
12 much smaller (U.S. EPA 2014, Appendix 6B).
13 Because increases in timber production represent increased tree growth and concurrent
14 carbon sequestration, we also consider WREA estimates of the potential increase in carbon
15 storage that potentially could occur for different air quality scenarios (U.S. EPA 2014, section
16 6.6.1). Comparisons of the W126 scenarios to the current standard scenario with regard to
17 carbon sequestration estimates do not indicate an appreciable difference for the W126 scenario
18 of 15 ppm-hrs. The majority of the enhanced carbon sequestration potential in the forest biomass
19 increases over time, for alternatives, is predicted to occur for the W126 scenarios of 11 and 7
20 ppm-hrs. Over 30 years, the current standard scenario projection is 89,184 million metric tons of
21 CC>2 equivalents (MMtCC^e). The WREA estimates additional sequestration potential of 13, 593
6-27
-------
1 and 1,600 MMtCC^e, for the W126 scenarios of 15, 11 and 7 ppm-hrs, respectively, as compared
2 to the current standard baseline (U.S. EPA 2014, Table 6-18)9. We also take note of the
3 relatively smaller estimates for carbon sequestration associated with improved crop yields (over
4 30 years) in the agricultural sector, which indicate little difference among the different W126
5 scenarios.
6 We additionally consider the WREA estimates for five urban areas of how reduced
7 growth of Os-sensitive trees in urban forests may affect the ecosystem services of air pollutant
8 removal and carbon sequestration (U.S. EPA, 2014, sections 6.6.2 and 6.7 and Appendix 6D).
9 With regard to air pollutant removal, the WREA estimated metric tons of carbon monoxide,
10 nitrogen dioxide, ozone and sulfur dioxide removed under the modeled air quality scenarios. In
11 considering these estimates we note the general assumptions made to estimate order of
12 magnitude effects of O^ removal by trees on 03 concentrations in the five urban areas and the
13 associated uncertainties (U.S. EPA 2014, sections 6.7 and 6.9 and Appendix 6D). Estimates for
14 all five areas indicate increased pollutant removal for air quality model-adjusted from recent
15 conditions to just meet the current standard, with much smaller difference where they exist
16 between the current standard and three W126 scenarios (Table 6-5 below). With respect to
17 carbon sequestration, relative to the scenario representing just meeting the current standard,
18 again the largest difference is generally observed with the existing conditions scenario (Table 6-
19 5). This is because of the largest difference in W126 estimates occurring between these two
20 scenarios. In addition to the small differences in W126 index values among the current standard
21 and W126 scenarios for these five areas, which contribute to the similarities among scenarios, we
22 also note that as only 2 or 3 tree species were able to be assessed in each city, these results may
23 underestimate the overall impacts nationally, although other areas of uncertainty (recognized
24 below) may tend to contribute to the opposite potential (U.S. EPA 2014, Table 6-27).
25
91 million metric tons of carbon dioxide equivalents (MMtCO2e) is equivalent to 208,000 passenger
vehicles or the electricity to run 138,000 homes for 1 year as calculated by the EPA Greenhouse Gas Equivalencies
Calculator (updated September 2013and available at http://www.epa.gov/cleanenergy/energy-
resources/calculator. html).
6-28
-------
1 Table 6-5. Estimated effect of Os-sensitive tree growth-related impacts on the ecosystem
2 services of air pollutant removal and carbon sequestration in five urban case
3 study areas.
Case Study Area
Atlanta
Baltimore
Chicago
Syracuse
Tennessee urban
Atlanta
Baltimore
Chicago
Syracuse
Tennessee urban
Air Oii^ilitw Qf*onarinc
^^11 \XUCIII LV WwGI ICil 1 vw
Recent
Conditions
(2006-2008)
Conditions just
meeting the
current
standard*
Average W126 index scenarios6
15ppm-hrs
15 ppm-hrs
15 ppm-hrs
Air Pollutant Removal (metric tons, CO, N02, 03, S02)
33,000
8,500
355,000
1,500
474,000
35,800
9,200
359,000
1,700
511,000
35,800
9,200
359,000
1,700
511,000
36,000
9,200
361,000
1,700
516,000
36,300
9,200
365,000
1,700
522,000
Million Metric Tons of C02 Equivalents, Carbon Storage (cumulative over 25 years)
1.2
0.5
16.9
0.14
18.0
1.32
0.57
17.05
0.17
19.67
1.32
0.57
17.05
0.17
19.67
1.32
0.57
17.10
0.17
19.89
1.34
0.57
17.21
0.17
20.16
A Results are derived from estimates per model grid cell (in which there are any of the 12 studied species) from W126 value
assigned to the grid cell based on application of the VNA method to the monitor-location W126 values that are the average at
that location across the 3 years of W126 values for the model-adjusted dataset that just meets the current standard (fourth-
highest daily maximum 8-hour concentration, averaged over 3 consecutive years of 75 ppb).
B The national distribution of W126 values within model grid-cells for each scenario reflects model adjustment of 2006-2008 Os
concentrations at monitoring sites such that the average W126 index at the controlling location in each of the modeling regions
just meets the scenario target index value, followed by application of the VNA interpolation methodology (see U.S. EPA, 2014,
section 4.3.4.1 and Appendix 4A).
4
5
6
1
8
9
10
11
12
13
14
15
With regard to foliar injury, we take note of the WREA analyses of the nationwide
dataset (2006- 2010) for U.S. Forest Service biosites described in section 5.4.2 above, including
the presentation indicating that the proportion of biosites with injury varies with soil moisture
conditions and O?, W126 index values, and that the proportion of biosites with injury severity
greater than 5 also varied with soil moisture (U.S. EPA 2014, Chapter 7, Figure 7-11). The
evidence of Cb-attributable visible foliar injury incidence occurring in USFS biosites shows that
the proportion of biosites showing foliar injury incidence increases steeply with W126 index
values up to approximately 10 ppm-hrs. At W126 index levels greater than approximately 10
ppm-hrs the proportion of sites showing foliar injury incidence is relatively constant.
In reflecting across the range of W126 estimates identified in various WREA analyses,
we first note the substantial benefits and reductions in biomass-related risks estimated for air
6-29
-------
1 quality adjusted to just meet the current standard. Additional incremental benefits or risk
2 reduction, generally of relatively smaller magnitude, is estimated across the W126 scenarios. In
3 considering this information discussed above in the context of identifying levels appropriate to
4 consider for a W126-based standard, we first take note of associated uncertainties in the context
5 of the following question.
6 • What are important uncertainties and limitations in the evidence and
7 exposure/risk analyses?
8 In considering the evidence and exposure/risk information summarized above and the
9 weight to place on this information, we are mindful of the uncertainties and limitations
10 associated with several key aspects of this information. We first consider the uncertainties
11 associated with the evidence underlying the tree seedling and crop C-R functions, given the
12 importance of these functions for many of the ecosystem service analyses described in the
13 WREA. Several key uncertainties associated with this information are listed here.
14 • Uncertainty regarding the extent to which the subset of studied tree and crop species
15 encompass the total number of 63 sensitive species in the nation and to what extent it
16 is representative of U.S. vegetation as a whole, given that information is available for
17 only a small fraction of the number of total species of trees and crops grown in the
18 U.S. (U.S. EPA, 2013, section 9.6, U.S. EPA, 2014, Table 6-27).
19 • Uncertainties regarding intra-species variability due to the different numbers of studies
20 that exist for different species so that the weight of evidence is not the same for each
21 species. Those species with more than one study show variability in response and C-R
22 functions. The potential variability in less well studied species is therefore unknown
23 (U.S. EPA, 2013, pp. 9-123/125, U.S. EPA, 2014, section 6.2.1.2, and Table 6-27).
24 • Uncertainty regarding the extent to which tree seedling C-R functions can be used to
25 represent mature trees since seedling sensitivity has been shown in some cases to not
26 reflect mature tree O?, sensitivity in the same species (U.S. EPA, 2013, section 9.6,
27 U.S. EPA, 2014, section 6.2.1.1 and Tables 6-5 and 6-27).
28 • Uncertainty in the relationship of Os effects on tree seedlings (e.g., relative biomass
29 loss) in one or a few growing seasons to effects that might be expected to accrue over
30 the life of the trees extending into adulthood (U.S. EPA, 2013, pp. 9-52/53, U.S.
31 EPA, 2014, section 6.2.1.4 and Table 6-27).
32 • Uncertainties associated with estimating the national scale ecosystem-level impacts
33 using weighted relative biomass loss (U.S. EPA, 2014, section 6.8, and Table 6-27)
34 • Uncertainties associated with potential biomass loss in federally designated Class I
35 areas (U.S. EPA, 2014, section 6.8. and Table 6-27)
36 Turning to consideration of the air quality conditions estimated for the various air quality
37 scenarios, we take note of the following uncertainties associated particularly with estimates of 63
38 exposures in rural areas nationally. These are described more completely in chapter 4 of the
6-30
-------
1 WREA (see for example, U.S. EPA, 2014, section 4.4) and summarized in chapter 8 of the REA
2 (U.S. EPA, 2014, section 8.5).
3 • Uncertainties in O?, exposures due to a lack of rural monitors, especially in the western
4 U.S. and at high elevation sites.
5 • Uncertainties associated with the method (VNA) used to interpolate monitor values to
6 estimate W126 index values in locations without monitors.
7 • Uncertainties in model-adjusted estimates of 63 concentrations associated with
8 meeting the current standard and potential alternative W126-based standards.
9 Numerous ecosystem services assessments were described in the WREA. These
10 assessments relied heavily on models, which also relied on the inputs of the seedling and crop C-
11 R functions and model-adjusted air quality estimates. Thus, including the uncertainties from the
12 first two categories discussed above, additional uncertainties associated with the ecosystem
13 services models include the following.
14 • Uncertainties associated with use of the iTree model to estimate pollution removal and
15 carbon storage in 5 urban area case studies, including uncertainties in the base
16 inventory of city trees, the functions used for air pollution removal and carbon storage
17 (U.S. EPA, 2014, sections 6.6.2, 6.7, and Table 6-27).
18 • Uncertainties associated with use of the FASOMGHG model for national timber and
19 crop production, including use of median C-R functions for crops in FASOM and
20 crop proxy and forest type assumptions to fill in where there was insufficient data
21 (U.S. EPA, 2014, sections 6.3, 6.5, 6.6.1, and Table 6-27).
22 • Uncertainties associated with use of the FASOMGHG model to estimate national scale
23 carbon sequestration, including those associated with the functions for carbon
24 sequestration (U.S. EPA, 2014, sections 6.2.1.1, 6.6.1, and Table 6-27).
25 In addition to biomass loss and crop yield loss, the WREA estimates the incidence and/or
26 severity of C^-induced visible foliar injury, both at the national and national park scales.
27 Numerous uncertainties are associated with these assessments and include the following.
28 • Uncertainties associated with our understanding of the number and sensitivity of 63
29 sensitive species (U.S. EPA, 2014, sections 7.2.1, 7.5 and Table 7-22).
30 • Uncertainties associated with spatial assignment of foliar injury biosite data to 12x12
31 km grids (U.S. EPA, 2014, sections 7.2.1, 7.5 and Table 7-22).
32 • Uncertainties associated with availability of biosite sampling data in some locations in
33 the western U.S. (U.S. EPA, 2014, sections 7.2.1, 7.5 and Table 7-22).
34 • Uncertainties associated with soil moisture threshold for foliar injury (U.S. EPA,
35 2014, sections 7.2.2, 7.2.3, 7.5 and Table 7-22).
6-31
-------
1 • Uncertainties associated with spatial resolution of soil moisture data, time period for
2 soil moisture data, drought categories and the combination of soil moisture and
3 biosite data (U.S. EPA, 2014, sections 7.3.3.2, 7.5 and Table 7-22).
4 • Uncertainties associated with 63 exposure data of vegetation and recreational areas
5 within parks (U.S. EPA, 2014, sections 7.4, 7.5 and Table 7-22).
6 • Uncertainties associated with surveys of recreational activities (U.S. EPA, 2014,
7 sections 7.1.1.2, 7.5 and Table 7-22).
8 Additionally, there is uncertainty associated with the extent to which the endpoints and
9 associated risk estimates considered above represent effects reasonably judged adverse in the
10 context of public welfare. All of these uncertainties are important to considerations below in the
11 context of target levels of protection with regard to weight to be placed on various lines of
12 evidence and assessment results.
13 • What considerations may be important to the Administrator's judgments on the
14 public welfare significance of Os associated vegetation effects that may be
15 expected under air quality conditions associated with different levels for a
16 seasonal cumulative standard?
17 Our consideration of this question is intended to provide a public welfare context for
18 consideration of the evidence and exposure/risk information discussed above, which includes the
19 nature and magnitude of observed and predicted effects at various levels of cumulative seasonal
20 exposures. We also note the importance of considering information in an integrated manner,
21 rather than focusing only on results from any one analysis. For example, we find it appropriate,
22 in considering the evidence with regard to seedling growth reduction (or biomass loss), to
23 consider the WREA estimates of affected area based on tree basal area together with estimates of
24 individual species responses based simply on the evidence-based C-R functions, and in light of
25 other potential impacts summarized above. In so doing in section 6.5 below, we take into
26 account considerations relevant to public welfare policy judgments required of the
27 Administrator, such as those described here.
28 As recognized in sections 1.3.2 and 5.1, the Clean Air Act specifies that secondary
29 standards protect against known or anticipated adverse effects to public welfare. In the
30 Administrator's judgment as to the standards that would be requisite (i.e., neither more nor less
31 stringent than necessary) to protect the public welfare under the Act, she will consider a number
32 of factors including 1) what can be considered to constitute an adverse effect to the public
33 welfare; 2) the nature and magnitude of the effects and the risks that remain after meeting the
34 level of the current standard; and, 3) what is necessary to achieve the requisite (no more and no
35 less) degree of protection. In the 2008 decision by which the current standard was established,
36 the Administrator considered these factors in judging the previously-existing standard to not
37 provide the requisite public welfare protection. At that time the Administrator found that the
6-32
-------
1 exposure- and risk-based analyses available in that review indicated that adverse effects to
2 vegetation would be predicted to occur under air quality conditions associated with just meeting
3 the then-current standard. The effects identified were "visible foliar injury and seedling and
4 mature tree biomass loss in Os-sensitive vegetation" (73 FR 16496). In so noting, the
5 Administrator indicated that he believed that "the degree to which such effects should be
6 considered to be adverse depends on the intended use of the vegetation and its significance to
7 public welfare" (73 FR 16496). With regard to consideration of intended use, the Administrator
8 took note of the specific uses of public lands set aside by Congress and intended to provide
9 benefits to the public welfare, "including lands that are to be protected so as to conserve the
10 scenic value and the natural vegetation and wildlife within such areas, and to leave them
11 unimpaired for the enjoyment for future generations" such as Class I areas (73 FR 16496). The
12 Administrator also recognized areas set aside by States, Tribes and public interest groups with
13 the intent "to provide similar benefits to the public welfare, for residents on State and Tribal
14 lands, as well as for visitors to those areas" (73 FR 16496).10
15 In the Administrator's judgments in the 2008 review, he did not identify specific criteria
16 or benchmarks or an overall level of protection from adverse environmental effects to public
17 welfare judged to be requisite under the Act.n As noted above, the scientists at the 1996
18 workshop identified ranges of cumulative seasonal index values (e.g., in terms of SUM06 or
19 W126) in the context of considering a degree of protection for vegetation effects defined in terms
20 of relative yield loss in crops and relative biomass loss in tree seedlings. In considering this
21 information in the context of a secondary standard, judgments are required by the Administrator
22 with regard to the degree that these or other benchmarks and other effects should be judged
23 adverse to the public welfare. In considering levels for a W126-based secondary standard that
24 may be appropriate to consider, we recognize that the statute requires that a secondary standard
25 be protective against only those known or anticipated Cb effects that are "adverse" to the public
26 welfare, not all identifiable Cb-induced effects. Thus, we recognize both the importance of
27 scientific consensus statements that have been made regarding vegetation-related endpoints and
28 Os exposure levels that might protect against such key endpoints and the importance of placing
29 such conclusions in the context of consideration of the public welfare more broadly.
10 In considering areas that have not been afforded such special protection, ranging from vegetation used
for residential or commercial ornamental purposes, such as land use categories that are heavily managed for
commercial production of commodities such as agricultural crops, timer and ornamental vegetation, the
Administrator indicated his expectation that protection of sensitive natural vegetation and ecosystems might be
expected to also provide some degree of additional protection for heavily managed commercial vegetation (73 FR
16496).
11 In remanding the 2008 decision on the secondary standard back to EPA, the Court indicated this
omission, as described in section 1.2.2 above.
6-33
-------
1 As discussed in section 5.1 and recognized by the EPA in prior reviews, staff recognizes
2 the importance of a more expansive construct or paradigm that addresses what constitutes
3 adverse effects of Cb to public welfare. In so doing, we also recognize several aspects or
4 dimensions of vegetation effects for consideration within this paradigm. These include the
5 likelihood, type, magnitude, and spatial scale of the effect, as well as the potential for recovery
6 and any uncertainties relating to these conditions (77 FR 20231). As in the last review, we also
7 continue to recognize that the public welfare significance of Cb-induced effects on sensitive
8 vegetation growing within the U.S. can vary, depending on the nature of the effect, the intended
9 use of the sensitive plants or ecosystems, and the types of environments in which the sensitive
10 vegetation and ecosystems are located. Any given Cb-related effect on vegetation and
11 ecosystems (e.g., biomass loss, foliar injury), therefore, may be judged to have a different degree
12 of impact on the public welfare depending, for example, on whether that effect occurs in a Class
13 I area, a city park, or in commercial cropland. In the 2008 review, the Administrator judged it
14 appropriate that this variation in the significance of Cb-related vegetation effects should be taken
15 into consideration in judging the level of ambient Cb that is requisite to protect the public welfare
16 from any known or anticipated adverse effects (73 FR 16496). For example, in considering
17 visible foliar injury and seedling and mature tree biomass loss in Cb-sensitive vegetation
18 expected under alternative air quality scenarios, the Administrator noted that "the degree to
19 which such effects should be considered to be adverse depends on the intended use of the
20 vegetation and its significance to the public welfare" (73 FR 16496). Further, the rulemaking
21 notice stated that "[i]n considering what constitutes a vegetation effect that is adverse from a
22 public welfare perspective, the Administrator believes it is appropriate to continue to rely on the
23 definition of 'adverse,' ... that imbeds the concept of "intended use" of the ecological receptors
24 and resources that are affected, and applies that concept beyond the species level to the
25 ecosystem level." The notice went on to state that "[i]n so doing, the Administrator has taken
26 note of a number of actions taken by Congress to establish public lands that are set aside for
27 specific uses that are intended to provide benefits to the public welfare, including lands that are
28 to be protected so as to conserve the scenic value and the natural vegetation and wildlife within
29 such areas, and to leave them unimpaired for the enjoyment of future generations" (73 FR
30 16496). Such public lands that are protected areas of national interest include national parks and
31 forests, wildlife refuges, and wilderness areas.
32 We also consider effects on ecosystem services in considering adversity to public
33 welfare. For example, the WREA has evaluated the economic value of ecosystem services
34 affected by 63 and how those services might be expected to change under different air quality
35 scenarios representing the current and potential alternative standards (U.S. EPA, 2014, chapter
36 6).
6-34
-------
1 Thus, we recognize several important considerations in evaluating levels of protection
2 and levels for a cumulative seasonal W126-based standard including: the extent of areas
3 expected to be affected nationwide and the magnitude of those effects; the extent of effects in
4 areas of national significance; the extent to which these impacts might be judged significant from
5 a public welfare perspective and associated uncertainties in the information. Accordingly, we
6 recognize that the range of alternative standard levels that may be appropriate to consider differs
7 based on the weight placed on different aspects of the evidence and on different aspects of the
8 quantitative exposure/risk information, and the associated uncertainties, as well as on public
9 welfare policy decisions regarding the public welfare significance of the effects considered and
10 the approaches for considering benchmarks for growth or biomass loss and other vegetation
11 effects of Os. As described in chapter 1, our objective is to identify the range of policy options
12 supported by the current evidence- and exposure/risk-based information and with consideration
13 of the role of the Administrator's public welfare judgments. In so doing, we recognize support
14 for consideration of a broad range of W126 index values, which we discuss in section 6.5, with
15 recognition of the different judgments that might provide support for different parts of such a
16 range.
17 6.4 CASACADVICE
18 In our consideration of potential alternative standards, in addition to the evidence-based,
19 risk/exposure-based, and air quality information discussed above, we also consider the advice
20 and recommendations of CASAC in EPA's proposed 2010 reconsideration of the 2008 decision,
21 as well as comments received thus far in the current review, in the context of its review of the
22 ISA, and earlier drafts of this document and the WREA. We have additionally considered public
23 comments received to date, some of which have suggested a lack of new information for support
24 of a distinct secondary standard and others that urge the consideration of a secondary standard
25 with a cumulative seasonal form using the W126 metric and a level within the range of 7 to 15
26 ppm-hrs.12
27 In response to the EPA's solicitation of CASAC's advice on the Agency's proposed
28 rulemaking as part of the reconsideration,13 CASAC conveyed their support for a secondary
29 standard distinct from the primary standard, noting that "vegetation effects are more dependent
12 Public comment received thus far in this review are in the docket EPA-HQ-OAR-2008-0699, accessible
at www.regulations.gov.
13 The reconsideration proposal included a proposed new cumulative, seasonal secondary standard,
expressed as an index of the annual sum of weighted hourly concentrations (the W126 index), cumulated over 12
hours per day during the consecutive 3-month period within the Os season with the maximum index value, averaged
over three years, set within a range of 7 to 15 ppm-hour (75 FR 3027).
6-35
-------
1 on the cumulative exposure to, and uptake of, ozone over the course of the entire growing
2 season" (Samet, 2010).
3 CASAC also supports EPA 's secondary ozone standard as proposed: a new
4 cumulative, seasonal standard expressed as an annual index of the sum of
5 weighted hourly concentrations (i.e., the Wl26form), cumulated over 12 hours
6 per day (Sam to 8pm) during the consecutive 3-month period within the ozone
1 season with the maximum index value, set as a level within the range of 7 to [1]5
8 ppm-hours. This W126 metric can be supported as an appropriate option for
9 relating ozone exposure to vegetation responses, such as visible foliar injury and
10 reductions in plant growth. We found the Agency's reasoning ... to be supported
11 by the extensive scientific evidence considered in the last review cycle.
12 In advice offered so far in the current review, which considers an updated scientific and
13 technical record since the 2008 rulemaking, the CASAC stated that "the focus [of the first draft
14 PA] on the W126 form is appropriate" (Frey and Samet, 2012, p. 2). They further commented
15 with regard to the support provided in the first draft PA for the consideration of such a form for a
16 secondary standard (Frey and Samet, 2012, p. 2).
17 There is a strong justification made for using a cumulative and weighted exposure
18 standard for welfare effects (i.e. the W126), and for the utility of using a 3-month
19 daylight exposure metric. Averaging across years is not recommended because a
20 single high exposure year could have lasting effects because of the perennial
21 nature of many plants and the lag times associated with propagating effects
22 through ecosystem trophic levels. Aver aging would obscure such critical impacts
23 and lead to inadequate protection against welfare effects.
24 Additionally, in advice regarding consideration of levels, their advice indicated that "[o]ptions
25 for levels [of a secondary standard] should be based on factors including predicted 5% loss of
26 crop yield and predicted 1-2% loss for trees" (Frey and Samet, 2012, p. 2).
27 6.5 PRELIMINARY STAFF CONCLUSIONS ON ALTERNATIVE STANDARD
28 Staffs consideration of alternative secondary Os standards builds on our conclusion from
29 section 5.7 above that the body of evidence, in combination with the results of the WREA, calls
30 into question the adequacy of the current secondary standard and provides support for
31 consideration of alternative standards. In sections 6.1 to 6.3 above, we consider how the
32 currently available scientific evidence and exposure/risk information inform decisions regarding
33 the basic elements of the NAAQS: indicator (6.1), form and averaging time (6.2), and level (6.3).
34 In so doing, we consider both the information available at the time of the last review and
35 information newly available since the last review which has been critically analyzed and
36 characterized in the 2013 ISA. As an initial matter, with regard to the indicator, we conclude
37 that based on the available science it is still appropriate to continue to use measurements of Os in
6-36
-------
1 accordance with federal reference methods as the indicator to address effects associated with
2 exposure to ambient 63 alone or in combination with related photochemical oxidants.
3 In considering alternative standards, staff has considered the available body of evidence
4 as comprehensively assessed in the ISA, the risk and exposure information presented in the
5 second draft REA and CASAC advice and public comment thus far in this review with regard to
6 support for consideration of options that are different from the current standard, as articulated by
7 the following overarching question
8 • To what extent does the currently available scientific evidence- and exposure/risk
9 based information, as reflected in the ISA and WREA, support consideration of
10 alternatives to the current Os standard to provide increased protection from
11 ambient Os exposures?
12 In considering potential forms alternative to that of the current standard, we note that the
13 form for the current secondary standard is the 4th highest daily maximum 8-hour average,
14 averaged over three years. As discussed in chapter 5 and section 6.2 above, the longstanding
15 evidence regarding the fundamental aspects of Os exposure that are directly responsible for
16 inducing vegetation response indicates that plant response to 63 is driven by the cumulative
17 exposure to Os during the growing season, rather than by a single event (U.S. EPA, 2013, section
18 2.6.6.1). This cumulative exposure depends on both the total duration of the exposure (from
19 repeated 63 episodes) and the concentrations of those exposures (higher concentrations having a
20 disproportionate impact over lower concentrations). On the basis of this longstanding and
21 extensive evidence, the ISA concludes that exposure indices that cumulate and differentially
22 weight the higher hourly average concentrations over a season and also include the mid-level
23 values offer the most defensible approach for use in developing response functions and in
24 defining indices for vegetation protection (U.S. EPA, 2013, section 2.6.6.1).
25 CASAC advice in the 2008 review and on the 2010 proposed reconsideration has
26 additionally recognized that the nature of the exposures relevant to vegetation response is well
27 described by a cumulative seasonal form and has supported the use of such a form for a
28 secondary O3 standard (Henderson, 2006; Samet, 2010). The current CASAC O3 Panel has
29 expressed similar views, stating "[tjhere is a strong justification made for using a cumulative and
30 weighted exposure standard for welfare effects (i.e. the W126)..." (Frey and Samet, 2012, p. 2).
31 We also note that on the basis of the evidence and exposure/risk information available in the two
32 previous reviews, and in consideration of CASAC advice, the Administrator has recognized the
33 importance of protecting vegetation from cumulative, seasonal exposures and proposed such a
34 form as an appropriate reasonable policy option (61 FR 65741-44; 62 FR 37899-905; 75 FR
35 3012-3027).
6-37
-------
1 Thus, in considering alternative forms of the standard we conclude that it is reasonable
2 and appropriate to consider a cumulative, concentration-weighted form to provide protection
3 against cumulative, seasonal exposures to Os that are known or anticipated to harm sensitive
4 vegetation or ecosystems. Such a form is specifically designed to directly measure the kind of
5 63 exposures that can cause harm to vegetation and would have a distinct advantage over the
6 form of the current standard in characterizing air quality conditions potentially of concern for
7 vegetation and demonstrating that the desired degree of protection against those conditions was
8 being achieved.
9 In considering the appropriate index for a cumulative seasonal form, we recognize that a
10 number of different cumulative concentration weighted indices have been developed and have
11 been evaluated in the scientific literature and in past NAAQS reviews in terms of their ability to
12 predict vegetation response and their usefulness in the NAAQS context (U.S. EPA, 2006, pp. 9-
13 11 to 9-15 and pp. AX9-159 to AX9-187; U.S. EPA, 2007, pp. 7-15/16). While these various
14 forms have different strengths and limitations, as noted in the ISA (U.S. EPA, 2013, section 9.5),
15 the W126 index14 has some important aspects not shared by other non-sigmoidally weighted
16 cumulative indices. For example, given the lack of discernible threshold for vegetation effects in
17 general, we recognize the fact that the W126 metric does not have a cut-off in its weighting
18 scheme (down to about 30 ppb below which the weighting factor is effectively zero), such that it
19 includes consideration of potentially damaging lower Os concentrations. Additionally, the W126
20 metric also adds increasing weight to hourly concentrations from about 40 ppb to about 100 ppb,
21 an important feature because "as hourly concentrations become higher, they become increasingly
22 likely to overwhelm plant defenses and are known to be more detrimental to vegetation" (U.S.
23 EPA, 2013, p. 9-104). We additionally take note of CASAC advice in the 2008 review and on
24 the 2010 proposed reconsideration recommending the use of the W126 index for a cumulative
25 seasonal form for a secondary O?, standard (Henderson, 2006; Samet, 2010). Similarly, the
26 current CASAC 63 Panel has indicated that a focus on a W126 form is appropriate (Frey and
27 Samet, 2012) Therefore, on the basis of the strength of the evidence and advice from CASAC,
28 we conclude that the W126 index is the most appropriate cumulative seasonal form to consider
29 in the context of the secondary Oi NAAQS review.
30 We next turn to the exposure periods - diurnal and seasonal - over which the W126
31 index would be summed in any given year. As discussed in section 6.2 above, the currently
32 available information continues to provide support for a definition of the diurnal period of
14 The W126 is a non-threshold approach described as the sigmoidally weighted sum of all hourly O3
concentrations observed during a specified diurnal and seasonal exposure period, where each hourly O3
concentration is given a weight that increases from 0 to 1 with increasing concentration (Lefohn et al, 1988; Lefohn
and Runeckles, 1987; U.S. EPA, 2013, section 9.5.2).
6-38
-------
1 interest as the 12-hour period from 8 am to 8 pm (U.S. EPA, 2013, section 9.5.3). In prior
2 reviews, the EPA has identified the 12-hour period from 8 am to 8 pm as appropriately capturing
3 the diurnal window with most relevance to the photosynthetic process (72 FR 37900; 75 FR
4 3013) and CASAC has generally supported the 12 hour daylight period (Henderson, 2006, 2007).
5 In light of the continued support in the evidence base, and no evidence on this issue differing
6 from that in previous reviews, we again conclude that it is appropriate to use the 12-hour period
7 from 8 am to 8 pm to cumulate daily 63 exposures. On this basis, we conclude that the 12-hour
8 diurnal window (8:00 am to 8:00 pm) represents the portion of the diurnal exposure period that is
9 most relevant to predicting or inducing plant effects related to photosynthesis and growth and
10 thus is an appropriate diurnal period to use in conjunction with a W126 cumulative metric.
11 With regard to a seasonal period of interest, the current evidence base continues to
12 provide support for a seasonal period with a minimum duration of three months (U.S. EPA,
13 2013, section 9.5.3). Included in the currently available evidence is a new analysis that
14 compared 3- and 6-month maximum W126 values for over 1,200 AQS and CASTNET EPA
15 monitoring sites for the years 2008-2009 that found that the two accumulation periods were
16 highly correlated (U.S. EPA, 2013, section 9.5.3, Figure 9-13). Thus, although we recognize that
17 the selection of a single seasonal time period over which to cumulate Os exposures for a national
18 standard necessarily represents a balance of factors, given the significant variability in growth
19 patterns and lengths of growing season among vegetative species growing within the U.S., we
20 conclude it is appropriate to identify the seasonal W126 index value as that derived from the
21 consecutive 3-month period within the Os season with the highest W126 index value. We note
22 that such a 3-month exposure period was also supported by CASAC in advice provided during
23 the last review and on the 2010 proposed reconsideration (Henderson, 2006; Samet, 2010).
24 With regard to form, we additionally consider the period of time over which a cumulative
25 seasonal W126-based standard should be evaluated. In so doing, we have considered the support
26 for both a single year standard and one with a form averaged over three years (section 6.2). We
27 recognize that there are a number of Cb-induced effects that have the potential for public welfare
28 significance within the annual timeframe (i.e. reduced crop yields and visible foliar injury).
29 However, as noted in section 6.2 above, there are uncertainties associated with these effects that
30 make it difficult to determine the degree of annual protection needed to protect the public
31 welfare from any known or anticipated adverse effects. On the other hand, annual effects in
32 perennial species can be "carried over" into the subsequent year where they affect growth and
33 reproduction (U.S. EPA, 2013, pp. 9-43 to 9-44 and p. 9-86). When these annual effects occur
34 over multiple years due to elevated Os exposures across several years, they have the further
35 potential to be compounded, increasing the potential for effects at the population and ecosystem
36 level, including effects on associated ecosystem services that may be of greater significance to
6-39
-------
1 the public welfare. These impacted services can include alteration of below-ground
2 biogeochemical cycles, and alteration of both above- and below- ground terrestrial community
3 composition and terrestrial ecosystem water cycling (U.S. EPA, 2013, Table 9-19) and
4 reductions in productivity and carbon sequestration in terrestrial ecosystems. We additionally
5 note that multiple years of critical 63 exposures might be expected to result in larger impacts on
6 forested areas, i.e. increased susceptibility to other stressors such as insect pests, disease, co-
7 occurring pollutants and harsh weather, than intermittent occurrences of such exposures due to
8 the compounding or carry-over effects on tree growth.
9 Given the above, we conclude that the public welfare significance of the effects that can
10 occur as a result of multiple year 63 exposures are greater than those associated with a single
11 year. Thus, to the extent that the Administrator's priority for public welfare protection to be
12 afforded by the secondary 03 standard is on long-term effects that occur in sensitive tree species
13 in natural forested ecosystems including federally protected areas such as Class I areas or on
14 lands set aside by States, Tribes and public interest groups to provide similar benefits to the
15 public welfare, a standard with a form that evaluates the cumulative seasonal index across
16 multiple years might be considered to provide an appropriate match to the nature of (Vrelated
17 effects on vegetation upon which the secondary O^ standard is focused. In considering such
18 forms, we focus on one that averages the W126 index values across three years, as discussed in
19 section 6.2 above.
20 We take note, however, of comments from CASAC on this matter, in particular their
21 comment in the current review that"...[a]veraging across years is not recommended because a
22 single high exposure year could have lasting effects because of the perennial nature of many
23 plants and the lag times associated with propagating effects through ecosystem trophic levels.
24 Averaging would obscure such critical impacts and lead to inadequate protection against welfare
25 effects" (Frey and Samet, 2012, p. 2). We recognize that annual effects on perennials can
26 propagate into subsequent years, and thus first consider the available analyses of year-to-year
27 variability in W126 index values. For example, based on an analysis of the inter-annual
28 variability of seasonal W126 index values (using 2008-2010 data from the AQS database), the
29 W126 index values can vary significantly from year to year (see Appendix 2C). This is not
30 unexpected given the logistic weighting function and also inter-annual variability in
31 meteorological conditions which contribute to 63 formation (see Appendix 2C). As a result,
32 areas that meet a three-year average standard form for which there is substantial inter-annual
33 variability of seasonal W126 index, are likely to have some years that are below the level of the
34 standard and others that are at or above. Given this fact, several important implications should
35 be noted. First, in regard to implications for potential long-term compounding of vegetation or
36 ecosystem effects, it would be expected that annual impacts in years with cumulative air quality
6-40
-------
1 below that of the three year average would allow species to do better than the target level of
2 protection expected to be achieved on average across the three year period. Thus, staff note the
3 importance, when considering an appropriate level for a form that averages W126 index values
4 across three years, of considering the extent to which the cumulative effect of different average
5 W126 exposures across the three-year period would be judged adverse.
6 Additionally, in regard to implications for standard stability, a standard based on an
7 annual W126 index would be expected to have a lower degree of year-to-year stability relative to
8 a standard based on a form that averages seasonal indices across three consecutive years, given
9 the potential for large year-to-year variability in annual W126 values in areas across the country.
10 Thus, a three-year evaluation period can contribute to greater public welfare protection by
11 limiting year-to-year disruptions in ongoing control programs that would occur if an area was
12 frequently shifting in and out of attainment due to extreme year-to-year variations in
13 meteorological conditions. This greater stability in air quality management programs thus
14 facilitates achievement of the protection intended by a standard.
15 Thus, to the extent that the Administrator puts greater weight on protecting those effects
16 associated with multi-year exposures and given the described public welfare benefit of having a
17 standard of a form with more year-to-year stability, we conclude that it is appropriate to consider
18 a secondary standard form that averages the seasonal W126 index values across three
19 consecutive years to achieve the appropriate target level of protection for longer-term effects,
20 including compounding, and to achieve greater stability in air quality management programs, and
21 thus, public welfare protection, than might result from an annual standard.
22 Turning to consideration of an appropriate range of levels for a W126 based standard, we
23 first note that the available Ch-related vegetation effects evidence reflects a continuum from
24 relatively higher 63 concentrations, at which scientists generally agree that vegetation effects are
25 likely to occur, through lower concentrations at which the likelihood and magnitude of a
26 response become increasingly uncertain. Further, we recognize the different types of Cb-induced
27 effects and genetic variability within and between species which contribute variability to
28 observed responses across species. In light of this, we recognize the role of the Administrator's
29 judgments regarding the adversity of the known and anticipated effects to the public welfare.
30 Thus, the EPA has developed a paradigm to assist the Administrator in putting the available
31 science and exposure/risk information into the context of public welfare (as discussed in section
32 5.1 above). This paradigm has evolved over the course of the Os NAAQS reviews and has also
33 been informed by similar constructs developed for other secondary NAAQS reviews. Most
34 recently, this paradigm has expanded to include consideration of adversity in terms of effects on
35 the ecosystem services associated with identified Cb-induced effects (see discussion in 6.3
36 above).
6-41
-------
1 In considering the range of W126 index values most recently cited by CAS AC, which
2 generally correspond to conclusions from the 1996 workshop, we note that the tree seedling
3 biomass loss percentages were per-year percentages, with consideration of longer-term impacts.
4 Staff is mindful of this in light of our consideration of a potential alternative standard with a
5 multi-year form. We additionally take note of the workshop objectives to identify values that
6 might provide protection of vegetative components of natural ecosystems (recognizing foliar
7 injury as the indicator), protection of some aspect of the integrity of forest ecosystems (using
8 growth effects on seedlings as surrogate) and protection against crop yield reductions (while
9 acknowledging uncertainties). In considering the range that derives, at least in part, from this
10 workshop, we additionally recognize the need for our purposes with regard to a secondary Oj,
11 standard, to consider the public welfare significance of identified effects, as discussed in section
12 6.3 above. We find that this need to consider public welfare significance may lead to
13 identification of a somewhat different range of W126 index values as appropriate to consider for
14 levels for a W126-based standard.15
15 In considering potential levels for an alternative standard based on the W126 metric, we
16 find it useful to consider the observations of biomass loss and crop yield loss in Table 6-1 above.
17 In so doing, we take note of the different index value estimates with regard to number of studied
18 species below different response benchmarks, as well as the median response. For example, we
19 note that the tree seedling relative biomass loss estimates for 15-17 ppm-hrs include five of the
20 twelve studied species below 2%, five to six species below 5% and nine to ten species below
21 10%, as well as a median species response of 5 to 6%. At these index values, the median crop
22 yield loss estimate across studied crops is just at or below 5% (and eight often are below 10%).
23 At the lower end of the index values in Table 6-1, the tree seedling estimates for 7 ppm-hrs
24 include ten species below 5%, seven below 2% and a median response just at or below 2%. We
25 additionally consider the WREA estimates which indicate 143 or 144 of the 145 assessed Class I
26 areas with tree seedling weighted relative biomass loss estimates below 2% for air quality
27 scenarios representing W126 values of 15 and below. Such estimates are of particular relevance
28 to judgments required of the Administrator regarding effects of public welfare importance.
29 Further we note other WREA estimates indicating potential benefits for effects related to public
30 welfare, such as carbon sequestration and pollutant removal. In so doing, however, we also take
31 note of the appreciable uncertainty in these quantitative estimates, and the policy judgements
32 required of the Administrator with regard to consideration of such uncertainties.
33 On the basis of all the considerations described above, including the evidence and
34 exposure/risk analyses, and advice from CASAC, we conclude that an appropriate range of
15 We additionally recognize variability in derivation of W126 estimates generally equivalent to SUM06
estimates from the 1996 workshop, as noted in section 6.3 above.
6-42
-------
1 W126 index values for the Administrator to consider in identifying a target degree of public
2 welfare protection, extends from 7 ppm-hrs to 15 ppm-hrs or somewhat higher (as further
3 described below). In so doing, we primarily consider the evidence- and exposure/risk-based
4 information for cumulative seasonal O^ exposures represented by W126 index values (including
5 those represented by the WREA average W126 scenarios) associated with biomass loss in
6 studied tree species, both in and outside areas that have been afforded special protections. We
7 additionally recognize foliar injury as an important 63 effect which, depending somewhat on
8 severity and spatial extent, may reasonably be concluded to be of public welfare significance
9 when occurring in nationally protected areas. However, we additionally take note of the
10 appreciable variability in this endpoint, as summarized in chapter 5 and section 6.3 above, which
11 poses challenges to giving it primary emphasis in identifying potential alternative standard
12 levels. Similarly, we give less emphasis to consideration of crop yield loss in our consideration
13 of potential standard levels here and in section 6.3 above, noting the median estimates of
14 approximately 5% or lower for W126 index levels at and below 17 ppm-hrs. We also note the
15 range of factors affecting annual crop yields, including those related to the role of management
16 strategies as recognized in sections 5.3 and 6.2 above which complicate the identification of a
17 degree of impact that can be considered adverse to the public welfare. On the other hand, tree
18 biomass loss can be an indicator of more significant ecosystem-wide effects which might
19 reasonably be concluded to be significant to public welfare. For example, when it occurs over
20 multiple years at a sufficient magnitude, it is linked to an array of effects on other ecosystem-
21 level processes, such as nutrient and water cycles, changes in above and below ground
22 communities, carbon storage and air pollution removal (U.S. EPA, 2014, Figure 5-1), that have
23 the potential to be adverse to the public welfare.
24 In focusing on trees and their associated ecosystem services, we first note that the studied
25 tree species vary widely in their sensitivity to Os-induced relative biomass loss. For example,
26 2% seedling biomass loss is estimated to occur with cumulative seasonal 63 exposure in terms of
27 W126 index values below 10 ppm-hrs for seven of the studied species and at or above
28 approximately 20 ppm-hrs in the other five studied species (Figure 5-1 above). The median
29 W126 index value (across studied species) for which a 2% biomass loss is estimated ranges
30 between approximately 7 and 14 ppm-hrs, among the three approaches presented in the WREA
31 (see description in section 5.2.1). In considering the potential magnitude of the ecosystem
32 impact of tree species, we focus on the WREA estimates of weighted relative biomass loss for
33 the W126 air quality scenarios (U.S. EPA, 2014, section 6.8). For the current standard and the
34 three W126 scenarios, these estimates indicate weighted relative biomass loss less than or equal
35 to 2% in 143-144 of 145 assessed nationally protected Class I areas. To the extent that emphasis
36 is given to such estimates for nationally protected Class I areas and for appreciable percentages
6-43
-------
1 of forested areas nationwide, a W126 index value extending up to 15 ppm (or perhaps somewhat
2 higher) may be appropriate to consider.
3 In considering the evidence-based information regarding tree seedling growth effects and
4 specifically biomass loss, we recognize an array of W126 index values that may be appropriate
5 to consider depending on the weight placed on different policy-related objectives with regard to
6 proportions of the 12 studied tree species at or below different growth response benchmarks and
7 on associated uncertainties. For example, conclusions may be reached regarding the higher index
8 values (e.g., up to 17 ppm-hrs) to the extent weight is placed on W126 index values for which
9 tree seedling C-R functions project less than 2% biomass loss in approximately half of the
10 studied species and less than 10% biomass loss in the large majority of studied species and on
11 index values for which weighted RBL is below 2% in nearly all assessed Class I areas, and to the
12 extent that estimates associated with the remaining species are judged too variable and/or
13 uncertain. For index values of 15 or 17 ppm-hrs, the species-specific composite C-R functions
14 indicate relative biomass loss less than 2% in at least five of the 12 studied species and less than
15 10% in at least nine or ten, in addition to an estimated median across studied species of
16 approximately 5-6% . Additionally, the WREA estimates for all three W126 scenarios include
17 weighted RBL below 2% in 143 to 144 of 145 assessed Class I areas. Alternatively, to the
18 extent weight is given to median biomass across tree species of no more than 2% based on the
19 evidence-based C-R functions, and to the potential, while uncertain, for appreciable gains in
20 carbon sequestration, a focus on the lower end of the range (e.g., down to 7 ppm-hr) may be
21 appropriate.
22 Thus, in staffs view, the evidence- and exposure/risk-based information relevant to tree
23 biomass loss and the associated ecosystem services important to the public welfare support
24 consideration of a W126-based secondary standard with index values within the assessed range
25 of 7-15 ppm-hrs or somewhat higher (e.g., 17 ppm-hrs). We consider such a range for a
26 potential alternative cumulative seasonal W126-based standard, averaged over three years. In so
27 doing, we take note of CAS AC's advice regarding the importance of considering the lasting or
28 carry-over effects that can derive from single year exposures of perennial plants, and recognize
29 the importance of considering the available evidence and exposure/risk based information related
30 to such effects, as well as associated uncertainties. We additionally recognize uncertainty
31 associated with any characterization of a relationship between the level of protection afforded for
32 cumulative growth-related effects by potential alternative W126-based standards of single year
33 or three-year average form. Lastly, we are mindful of the public welfare judgments required of
34 the Administrator with regard to the public welfare significance of identified effects and the
35 requisite level of protection, as well as the appropriate weight to assign the range of uncertainties
36 inherent in the evidence and analyses.
6-44
-------
1 In also noting the potential ecosystem services benefits related to tree biomass loss for
2 potential alternative W126-based standards in light of the qualitative and semi-quantitative
3 information in the WREA regarding the types and potential magnitude of impacts on associated
4 services, we recognize, in particular that impacts on climate can reasonably be concluded
5 significant from a public welfare perspective and CC>2 sequestration has been identified as a
6 potentially important tool for managing anthropogenic impacts on climate. However, we
7 additionally take note of significant uncertainties and limitations associated with WREA
8 estimates related to carbon sequestration. Thus, in selecting a target level of protection for forest
9 trees and their associated ecosystem services, the Administrator will need to exercise judgments
10 regarding the appropriate weight to place on the potential for benefits to the public welfare with
11 respect to ecosystem services of carbon storage and urban air pollution removal associated with
12 tree growth, as well as the large uncertainties associated with this information.
13 We further recognize that public welfare considerations taken into account by the
14 Administrator may have the potential to affect the target protection judged requisite by the
15 Administrator and the associated range of W126 index values. For example, to the extent the
16 Administrator chooses to put more weight on effects associated with longer-term conditions, it
17 may be appropriate to evaluate the significance of these longer term effects to the public welfare,
18 as well as the role that year-to-year exposure variability can play in realizing the potential public
19 welfare impacts. In so doing, the Administrator may put weight on a range of percentages of
20 biomass loss (either greater or less than 2%) with an objective to achieve substantial protection
21 of a large proportion of the studied species. Thus, in considering a range of 3-year average
22 W126 index values appropriate to provide the target protection, we recognize that the
23 Administrator may consider a range somewhat beyond the staff identified range.
24 In considering the lower end of the staff-identified range, the Administrator would need
25 to put more weight on the uncertainties that suggest the analyses may be underestimating the
26 public welfare impacts associated with different cumulative exposure levels. These uncertainties
27 can include the relatively small number of Os sensitive species for which we have robust C-R
28 functions included in the analyses, Os-sensitive species for which we have no C-R function, the
29 lack of information regarding the relationship of 63 effects on tree seedlings (e.g., relative
30 biomass loss) that occur in one or a few growing seasons to longer-term effects that might be
31 expected to accrue over the life of the trees extending into adulthood, the paucity of ambient air
32 monitoring data in some areas (e.g., the west) that leads to less than complete national coverage,
33 recognition that co-occurring stressors may in some cases exacerbate predicted Os-induced
34 effects, and the limited number of WREA urban area case studies, uncertainties associated with
35 the model-adjusted air quality exposure surfaces,and the inability to quantify some potentially
6-45
-------
1 important associated ecosystem services and incremental impacts (i.e. for insect damage, fire
2 regimes).
3 In considering a range somewhat above that of the staff-identified range, the
4 Administrator would need to put more weight on the uncertainties suggesting the potential for
5 overestimating the beneficial public welfare impacts that might be achieved at 3-year average
6 levels within the staff-identified range. These uncertainties include the lack of information
7 (particularly quantitative) on the relationship of tree seedling Oj, effects (e.g., relative biomass
8 loss) that occur in one or a few growing seasons to longer-term effects that might be expected to
9 accrue over the life of the trees extending into adulthood, recognition that co-occurring stressors
10 or environmental factors (e.g., drought) may in some cases mitigate predicted Cb-induced effects,
11 the variability in number of C-R functions available for each of the 12 studied tree species, and
12 information concerning the extent to different endpoints and effects might be considered adverse
13 to public welfare . Given these and potentially other uncertainties, the Administrator may choose
14 to select a higher range of levels judged to have less potential to provide overprotection.
15 Lastly, we also conclude that, to the extent the Administrator finds it useful to consider
16 the public welfare protection that might be afforded by a revised primary standard, this is
17 appropriately judged through the use of a cumulative seasonal W126-based exposure metric.
18 Such a use could inform a judgment of whether the primary standard would be expected to
19 achieve the level of public welfare protection concluded to be requisite under the Act in terms of
20 a metric considered appropriate to judging impacts on public welfare. See Mississippi. 723 F. 3d
21 at 272-73. The staff further concludes that the drawing of conclusions with regard to the public
22 welfare protection afforded by such a standard, as well as identification of the requisite level of
23 protection for such a standard, should entail consideration of the air quality conditions likely to
24 be achieved in terms of the cumulative seasonal W126-based metric described above. In such a
25 consideration, such as through the review of overlap analyses discussed in section 6.2 above,
26 staff further concludes it is important to take into account associated uncertainties, including
27 those associated with the limited monitor coverage in many rural areas, including those in the
28 west and southwest and at high elevation sites.
29 6.6 SUMMARY OF PRELIMINARY CONCLUSIONS ON THE SECONDARY
30 STANDARD
31 Staff preliminary conclusions are informed by our consideration of the available
32 scientific evidence as assessed in the ISA, the air quality/exposure/risk information in the second
33 draft WREA, advice from CAS AC thus far in this review and in prior reviews, and public
34 comment thus far in this review. Staff conclusions in the final PA will be further informed by
35 comments from CAS AC and the public on this draft document and by the final WREA.
6-46
-------
1 Staff preliminary conclusions on policy options that are appropriate for the
2 Administrator's consideration in making decisions on the secondary standards for 63, together
3 with supporting conclusions from sections 5.7 and 6.5 above, are briefly summarized below. In
4 reaching conclusions on alternative standards to provide requisite protection for public welfare
5 effects associated with ambient 63 exposures, staff has considered these standards in terms of the
6 basic elements of the NAAQS: indicator, form, averaging time, and level. In drawing these
7 conclusions, we are mindful that the Act requires secondary standards to be set so that, in the
8 Administrator's judgment, they are requisite to protect public welfare from known or anticipated
9 adverse environmental effects, such that the standards are to be neither more nor less stringent
10 than necessary. Thus, the Act does not require that NAAQS be set at zero-risk levels, but rather
11 at levels that reduce risk sufficiently to protect public welfare from adverse effects.
12 (1) Staff preliminarily concludes, based on the combined consideration of the body of
13 evidence and the results from the quantitative exposure/risk assessment, that the
14 available evidence and exposure/risk information call into question the adequacy
15 of the public welfare protection provided by the current standard and it is
16 appropriate to consider revising the standard to provide greater public welfare
17 protection.
18 (2) In considering an appropriate target level of protection for a revised standard,
19 staff additionally preliminarily concludes that it is appropriate to judge Os public
20 welfare impacts using the cumulative seasonal W126-based metric.
21 a. To the extent the Administrator finds it useful to consider the extent of
22 public welfare protection that might be afforded by a revised primary
23 standard, staff preliminarily concludes that public welfare protection is
24 appropriately judged through the use of the cumulative seasonal W126-
25 based metric.
26 (3) With regard to indicator, staff preliminarily concludes that it is appropriate to
27 continue to use 63 as the indicator for a standard that is intended to address
28 welfare effects associated with exposure to Os, alone or in combination with
29 related photochemical oxidants. Based on the available information, staff
30 preliminarily concludes that there is no basis for considering an alternative
31 indicator at this time.
32 (4) With regard to averaging time and form, staff preliminarily concludes that it is
33 appropriate to consider a revised secondary standard in terms of the cumulative,
34 seasonal, concentration-weighted form, the W126 index. With regard to
6-47
-------
1 definition of the W126 index for this purpose, staff makes the additional
2 preliminary conclusions.
3 a. It is appropriate to consider the consecutive 3-month period within the O^
4 season with the maximum index value as the seasonal period over which
5 to cumulate hourly Os exposures. Staff notes that the maximum 3-month
6 period generally coincides with maximum biological activity for most
7 vegetation, making the 3-month duration a suitable surrogate for longer
8 growing seasons.
9 b. It is appropriate to cumulate daily exposures for the 12-hr period from 8
10 am to 8 pm, generally representing the daylight period during the 3-month
11 period identified above.
12 c. It is appropriate to consider a form that averages W126 index values
13 across three consecutive years. Staff concludes it is appropriate to
14 consider this form in conjunction with appropriate levels in order to
15 provide the desired degree of public welfare protection from 03 effects
16 across multiple years.
17 With regard to level for a standard as described above, we preliminarily conclude that it
18 is appropriate to give consideration to a range of levels from somewhat above 15 ppm-hrs to 7
19 ppm-hrs, expressed in terms of the W126 index. Staff additionally notes that, consideration of
20 the support provided by the information available in this review will depend on public welfare
21 policy judgments by the Administrator regarding the protection of public welfare. This range
22 reflects staff judgment that a standard set within this range could provide an appropriate degree
23 of public welfare protection.
24 6.7 KEY UNCERTAINTIES AND AREAS FOR FUTURE RESEARCH AND
25 DATA COLLECTION
26 Staff believes it is important to highlight key uncertainties associated with establishing
27 secondary standards for 63. Such key uncertainties and recommendations for welfare-related
28 research, model development, and data gathering are outlined below. In some cases, research in
29 these areas can go beyond aiding standard setting to aiding in the development of more efficient
30 and effective control strategies. We note, however, that a full set of research recommendations
31 to meet standards implementation and strategy development needs is beyond the scope of this
32 discussion. Based on items highlighted in chapter 9 of the ISA and chapter 5 and 6 herein, we
33 have identified the following key uncertainties, research questions and data gaps that have been
34 highlighted in this review of the welfare-based secondary standard. The first set of key
6-48
-------
1 uncertainties and research recommendations discussed below is that associated with the
2 extrapolation to plant species and environments outside of specific experimental or field study
3 conditions. The second set of key uncertainties and research recommendations pertain to our
4 ability to assess the impact of Os on other welfare effects categories such as climate, ecosystem
5 components such as wildlife, and whole ecosystem structure and function. Third, we identify
6 research areas related to the development of approaches, tools, or methodologies useful in
7 characterizing the relationship between Oj, and public welfare in a policy context. These three
8 areas are described below.
9 There have been five decades of research regarding Os effects on plants and much
10 information has been compiled in previous reviews. One of the most important research results
11 for the review of the secondary Os standard is C-R relationships for plant species. This review
12 uses C-R functions from 22 crop and tree species. However, there are tens of thousands of plant
13 species in the U.S. (USDA, NRCS. 201416) and 66 plant species have been identified as O3
14 sensitive on National Park Service and US Fish and Wildlife Service lands17. Studies using large
15 numbers of native plant species across regions where those species are indigenous, might be
16 expected to reduce uncertainties associated with extrapolating plant response for a given level of
17 Os using composite response functions across differing regions and climates. Research on
18 additional species might additionally improve our understanding of the full range of response of
19 plant species to Os. Studies focused on fruits and vegetables might assist in reducing
20 uncertainties associated with Os effects on agriculture. Particular focus is suggested on
21 organically grown vegetables that may receive less intensive management than conventionally
22 grown crops. Recent studies indicate that watermelons may be particularly sensitive to Os
23 exposure (U.S. EPA, 2013, section 9.4.4.1) and older studies indicate grapes, honeydew melon,
24 lemons and oranges may also be Os sensitive (Abt Associates Inc., 1995).
25 National visible foliar injury surveys can indicate how widespread Os effects may be
26 within the US. However, there remain uncertainties about the nature of the effects indicated by
27 the observed foliar injury. These include uncertainty associated with estimating the risk to
28 vegetation of differing amounts of Os-induced visible foliar injury over the plant's leaf area and
29 the relationship between relative soil moisture and the incidence and severity of foliar injury in
30 sensitive species, as well as the extent to which visible foliar injury impacts ecosystem services
31 (e.g., tourism).
16 USD A, NRCS. 2014. The PLANTS Database (http://plants.usda.gov. 3 January 2014). National Plant
Data Team, Greensboro, NC 27401-4901 USA.
17See http://www2.nature.nps.gov/air/Pubs/pdf/flag/NPSozonesensppFLAG06.pdf
6-49
-------
1 Some new information has emerged linking effects on tree seedlings with larger trees and
2 similarities in results between exposure techniques (U.S. EPA 2013, section 9.6). Uncertainties
3 remain in this area and in relationships between effects on individual plants and ecosystem
4 effects. There are also uncertainties in extrapolating from Os effects on juvenile to mature trees
5 and from trees grown in the open versus those in a closed forest canopy in a competitive
6 environment. Uncertainties in extrapolating individual plant response spatially or to higher
7 levels of biological organization, including ecosystems, could be informed by research that
8 explores and better quantifies the nature of the relationship between 63, plant response and
9 multiple biotic and abiotic stressors, including those associated with climate change. Because
10 these uncertainties are multiple and significant due to the complex interactions involved, new
11 research will likely require a combination of manipulative experiments with model ecosystems,
12 community and ecosystem studies along natural Os gradients, and extensive modeling efforts to
13 project landscape-level, regional, national and international impacts of 63.
14 Uncertainties associated with projections of the effects of Os on the ecosystem processes
15 of water, carbon, and nutrient cycling, particularly at the stand and community levels might be
16 addressed through research on the effects on below ground ecosystem processes in response to
17 Os exposure alone and in combination with other stressors. These below-ground processes
18 include interactions of roots with the soil or microorganisms, effects of 63 on structural or
19 functional components of soil food webs and potential impacts on plant species diversity,
20 changes in the water use of sensitive trees, and if the sensitive tree species is dominant, potential
21 changes to the hydrologic cycle at the watershed and landscape level. Research on competitive
22 interactions under different Os exposures might improve our understanding of how Os may
23 affects biodiversity or genetic diversity. Such research could be strengthened by modern
24 molecular methods to quantify impacts on diversity.
25 Important interactions with biotic and abiotic stressors have been identified in this
26 review. More tools and research would improve our understanding of relationships between Os
27 exposure and stressors such as insect infestations, plant diseases, drought and potential stressors
28 from climate change. It is also important to understand how such interactions may affect
29 ecosystem services such as CO2 sequestration; food and fiber production; wildlife habitat and
30 water resources.
31 One of the most important uncertainties in this review is the characterization of air
32 quality in rural areas where there is limited monitoring. More comprehensive monitoring in
33 these areas would reduce uncertainties associated with Os exposures in many rural areas. Areas
34 of particular uncertainty include protected natural areas in the western U.S, including those at
35 high elevation, as well as those downwind of recently expanded oil and gas development areas.
36 Uncertainties associated with quantifying exposure in areas with and without monitors might be
6-50
-------
1 addressed through additional work on interpolation methods and air quality models that are
2 tailored to estimating cumulative seasonal exposures, as well as on more refined spatial grids and
3 complex terrain.
4 6.8 REFERENCES
5 Abt Associates, Inc. (1995). Ozone NAAQS benefits analysis: California crops. Report to U.S. EPA, July.
6 Federal Register 1996. National Ambient Air Quality Standards for Ozone; Proposed Rule. 40 CFR 50; Federal
7 Register 61: 65716
8 Federal Register 1997. National Ambient Air Quality Standards for Ozone; Final Rule. 40 CFR 50; Federal Register
9 62:38856
10 Federal Register 2007. National Ambient Air Quality Standards for Ozone; Proposed Rule. 40 CFR 50; Federal
11 Register 72: 37818
12 Federal Register 2008. National Ambient Air Quality Standards for Ozone; Final Rule. 40 CFR 50 and 58; Federal
13 Register 73:16436
14 Federal Register 2010. National Ambient Air Quality Standards for Ozone; Proposed Rule. 40 CFR 50 and 58;
15 Federal Register 75 FR: 2938
16 Federal Register 2012. National Ambient Air Quality Standards for Oxides of Nitrogen and Sulfur; Final Rule. 40
17 CFR 50; Federal Register 77 FR 20218
18 Frey, C. and Samet, J.M. (2012) CASAC Review of the EPA's Policy Assessment for the Review of the Ozone
19 National Ambient Air Quality Standards (First External Review Draft-August 2012). EPA-CASAC-13-
20 003. November 26, 2012. Available online at:
21 Gregg, JW; Jones, CG; Dawson, TE. (2003). Urbanization effects on tree growth in the vicinity of New York City
22 [Letter]. Nature 424: 183-187. http://dx.doi.org/10.1038/nature01728
23 Heck, WW; Cowling, EB. (1997).The need for a long term cumulative secondary ozone standard - An ecological
24 perspective. EM January: 23-33.
25 Henderson, R. (2006) Letter from CASAC Chairman Rogene Henderson to EPA Administrator Stephen Johnson.
26 October 24, 2006, EPA-CASAC-07-001.
27 Henderson, R. (2007) Letter from CASAC Chairman Rogene Henderson to EPA Administrator Stephen Johnson.
28 March 26, 2007, EPA-CASAC-07-002.
29 Henderson, R. (2008) Letter from CASAC Chairman Rogene Henderson to EPA Administrator Stephen Johnson.
30 April 7,2008, EPA-CASAC-08-009.
31 Samet, J.M. (2010) Clean Air Scientific Advisory Committee (CASAC) Review of EPA's Proposed Ozone National
32 Ambient Air Quality Standard. EPA-CASAC-10-007. January 19, 2010. Available online at:
3 3 http://yosemite.epa.gov/sab/sabproduct.nsf/264cb 1227d55e02c85257402007446a4/610BB57CFAC8A41C
34 852576CF007076BD/$File/EP A-CASAC-10-007-unsigned.pdf
3 5 USD A, NRCS. 2014. The PLANTS Database (http://plants.usda.gov. 3 January 2014). National Plant Data Team,
36 Greensboro, NC 27401-4901 US A
6-51
-------
1 U.S. EPA (1996a). Air quality criteria for ozone and related photochemical oxidants [EPA Report]. (EPA/600/P-
2 93/004AF). U.S. Environmental Protection Agency, Research Triangle Park, NC.
3 U.S. EPA (1996b). Review of national ambient air quality standards for ozone: Assessment of scientific and
4 technical information: OAQPS staff paper [EPA Report]. (EPA/452/R-96/007). Research Triangle Park,
5 NC. http://www.ntis.gov/search/product.aspx?ABBR=PB96203435
6 U.S. EPA (2006). Air Quality Criteria for Ozone and Related Photochemical Oxidants (2006 Final). U.S.
7 Environmental Protection Agency, Washington, DC. EPA/600/R-05/004aF-cF. March 2006. Available at:
8 http://www.epa.gOv/ttn/naaqs/standards/ozone/s o3crcd.html
9 U.S. EPA (2007). Review of the national ambient air quality standards for ozone: Policy assessment of scientific
10 and technical information: OAQPS staff paper [EPA Report]. (EPA/452/R-07/003). Research Triangle
11 Park, NC. http://www.epa.gov/ttn/naaqs/standards/ozone/data/2007_01_ozone_staff_paper
12 U.S. EPA (2013). Integrated Science Assessment of Ozone and Related Photochemical Oxidants (Final). U.S.
13 Environmental Protection Agency, Washington, DC. EPA/600/R-10/076F
14 U.S. EPA (2014). Welfare Risk and Exposure Assessment for Ozone, Second External Review Draft. Office of Air
15 Quality Planning and Standards, Research Triangle Park, NC, 27711. EPA-452/P-14-003a
16 Wolff, G.T. (1996) Letter to EPA Administrator Carol Browner, RE: Closure by the Clean Air Scientific Advisory
17 Committee (CASAC) on the Secondary Standard Portion of the Staff Paper for Ozone. EPA-SAB-CASAC-
18 LTR-96-002, April 4, 1996.
6-52
-------
APPENDICES
Appendix 2A. Supplemental Air Quality Modeling Analyses of Background O?, 2A-1
Appendix 2B. Monitoring Data Analysis of Relationships Between Current Standard and
W126 Metric 2B-1
Appendix 2C. Inter-annual Variability in W126 Index Values: Annual and 3-Year Average
Metrics (2008-2010) 2C-1
Appendix 3 A. Modes of Action Summary 3A-1
Appendix 3B. Recent Studies of Respiratory-related Emergency Department Visits
and Hospital Admissions 3B-1
Hospital Admissions for All Respiratory Causes 3B-1
Cause-Specific Hospital Admissions 3B-5
Emergency Department Visits for All Respiratory Causes 3B-7
Cause-Specific Emergency Department Visits 3B-7
Appendix 3C. At-risk Populations 3C-1
Appendix 3D. Air Quality Data for Locations of Key Epidemiological Studies 3D-1
Appendix 5A. Ozone-Sensitive Plant Species Used by Some Tribes 5A-1
Appendix 6 A. Calculation of Approximate Equivalent 12-hr SUM06 and 12-hr W126 6A-1
January 2013 Draft - Do Not Quote or Cite
-------
i APPENDIX 2A
2 SUPPLEMENTAL AIR QUALITY MODELING ANALYSES OF
3 BACKGROUND OZONE
4
5 Table of Contents
6 List of Figures 2
7 List of Tables 4
8 1. Introduction 5
9 2. Description of modeling methodologies 6
10 a. 2007 GEOS-Chem/CMAQ zero-out modeling 7
11 b. 2007 GEOS-Chem/CAMx source apportionment modeling 12
12 3. Estimates of seasonal-average background ozone levels 15
13 4. Distributions of background ozone levels 22
14 5. Contribution of various processes and sources to total background ozone 30
15 6. Estimates of the fractional background contribution to total ozone in 12 specific
16 areas 37
17 7. Background ozone and W126 39
18 8. Summary 39
19 9. References 42
20
2A-1
-------
21 List of Figures
22 Figure la. Modeling domain used in 2007 CMAQ and CAMx modeling 10
23
24 Figure Ib. Density scatterplot comparing CMAQ base daily peak 8-hour ozone predictions
25 against observed 8-hour ozone peaks paired in space and time for all sites during April-
26 October 2007 10
27
28 Figure Ic. Bias in seasonal mean (April-October) maximum daily 8-hour ozone predictions in
29 the 2007 CMAQ base simulation 11
30
31 Figure Id. Relationship between CMAQ estimations of MDA8 natural background ozone and
32 daily model biases 11
33
34 Figure 2a. Density scatterplot comparing CAMx base daily peak 8-hour ozone predictions
35 against observed 8-hour ozone peaks paired in space and time for all sites during April-October
36 2007 14
37
38 Figure 2b. Bias in seasonal mean (April-October) maximum daily 8-hour ozone predictions in
39 the 2007 CAMx base simulation 14
40
41 Figure 3a. April-October average MDA8 ozone (ppb) at monitoring locations across the U.S. as
42 estimated by a 2007 CMAQ base simulation 18
43
44 Figure 3b. April-October average natural background MDA8 ozone (ppb) at monitoring
45 locations across the U.S. as estimated by a 2007 CMAQ zero out simulation 18
46
47 Figure 3c. April-October average North American background MDA8 ozone (ppb) at monitoring
48 locations across the U.S. as estimated by a 2007 CMAQ zero out simulation 19
49
50 Figure 3d. April-October average United States background MDA8 ozone (ppb) at monitoring
51 locations across the U.S. as estimated by a 2007 CMAQ zero out simulation 19
52
53 Figure 4a. Ratio of natural background to total April-October average MDA8 ozone at
54 monitoring locations across the U.S. as estimated based on 2007 CMAQ simulations 20
55
56 Figure 4b. Ratio of N. American background to total April-October average MDA8 ozone at
57 monitoring locations across the U.S. as estimated based on 2007 CMAQ simulations 20
58
59 Figure 4c. Ratio of U.S. background to total April-October average MDA8 ozone at
60 monitoring locations across the U.S. as estimated based on 2007 CMAQ simulations 21
61
62 Figure 4d. Ratio of sources other than U.S. anthropogenic emissions to total April-October
63 average MDA8 ozone at monitoring locations across the U.S. as estimated by a 2007 CAMx
64 source apportionment simulation 21
65
66 Figure 5a. Distribution of natural background MDA8 ozone (ppb) at monitoring locations
67 across the U.S. (Apr-Oct), binned by base modeled site-day MDA8, as estimated by 2007
2A-2
-------
68 CMAQ simulations 25
69
70 Figure 5b. Distribution of N. American background MDAS ozone (ppb) at monitoring
71 locations across the U.S. (Apr-Oct), binned by base modeled site-day MDAS, as estimated by
72 2007 CMAQ simulations 25
73
74 Figure 5c. Distribution of U.S. background MDAS ozone (ppb) at monitoring locations across
75 the U.S. (Apr-Oct), binned by base modeled site-day MDAS, as estimated by 2007 CMAQ
76 simulations 26
77
78 Figure 5d. Distribution of MDAS ozone contributions from non-U.S. manmade sources (ppb)
79 at monitoring locations across the U.S. (Apr-Oct), binned by base modeled site-day MDAS, as
80 estimated by 2007 CAMx simulations 26
81
82 Figure 6a. Distribution of natural background MDAS ozone fractions at monitoring locations
83 across the U.S. (Apr-Oct), binned by base modeled site-day MDAS, as estimated by 2007
84 CMAQ simulations 27
85
86 Figure 6b. Distribution of N. American background MDAS ozone fractions at monitoring
87 locations across the U.S. (Apr-Oct), binned by base modeled site-day MDAS, as estimated by
88 2007 CMAQ simulations 27
89
90 Figure 6c. Distribution of U.S. background MDAS ozone fractions at monitoring locations
91 across the U.S. (Apr-Oct), binned by base modeled site-day MDAS, as estimated by 2007
92 CMAQ simulations 28
93
94 Figure 6d. Distribution of MDAS ozone fractions from non-U.S. anthropogenic sources at
95 monitoring locations across the U.S. (Apr-Oct), binned by base modeled site-day MDAS, as
96 estimated by the 2007 CAMx simulation 28
97
98 Figure 7. April-October 95th percentile United States background MDAS ozone (ppb) at
99 monitoring locations across the U.S. as estimated by a 2007 CMAQ base simulation 29
100
101 Figure 8a. Difference in April-October average MDAS ozone (ppb) at monitoring locations
102 across the U.S. between the USB scenario and the NAB scenario 32
103
104 Figure 8b. Difference in April-October average MDAS ozone (ppb) at monitoring locations
105 across the U.S. between the NAB scenario and the NB scenario 32
106
107 Figure 9a. Percentage of April-October average MDAS ozone that is apportioned to
108 boundary conditions as estimated at monitoring locations by a 2007 CAMx simulation 33
109
110 Figure 9b. Percentage of April-October average MDAS ozone that is apportioned to U.S.
Ill anthropogenic sources as estimated at monitoring locations by a 2007 CAMx simulation 33
112
113 Figure 9c. Percentage of April-October average MDAS ozone that is apportioned to purely
114 biogenic emissions as estimated at monitoring locations by a 2007 CAMx simulation 34
115
2A-3
-------
116 Figure 9d. Percentage of April-October average IV1DA8 ozone that is apportioned to
117 climatological fire emissions as estimated at monitoring locations by a 2007 CAMx simulation ... 34
118
119 Figure 9e. Percentage of April-October average IV1DA8 ozone that is apportioned to
120 anthropogenic emissions from in-domain Canadian and Mexican sources as estimated at
121 monitoring locations by a 2007 CAMx simulation 35
122
123 Figure 9f. Percentage of April-October average IV1DA8 ozone that is apportioned to Category 3
124 marine vessel emissions beyond U.S. territorial waters as estimated at monitoring locations by
125 a 2007 CAMx simulation 35
126
127 Figure 9g. Percentage of April-October average MDA8 ozone that is apportioned to Gulf of
128 Mexico point sources as estimated at monitoring locations by a 2007 CAMx simulation 36
129
130
131 List of Tables
132
133 Table la. April-October average MDA8 ozone, average MDA8 ozone from sources other than
134 U.S. manmade emissions, and the fractional contribution of these background sources in the
135 12 REA urban study areas, as estimated by a 2007 CAMx simulation 37
136
137 Table Ib. Average MDA8 ozone, average MDA8 ozone from sources other than U.S.
138 manmade emissions, and the fractional contribution of these background sources in the 12 REA
139 areas, as estimated by a 2007 CAMx simulation using site-days in which base MDA8 ozone
140 exceeded 60 ppb 38
141
142 Table Ic. Fractional contribution of non-U.S. manmade emissions sources in the 12 REA urban
143 study areas, as estimated by a 2007 CAMx simulation using means and medians of daily MDA8
144 fractions 38
145
146 Table Id. April-October average MDA8 ozone, average MDA8 ozone from USB, and the
147 fractional contribution of these background sources in the 12 REA urban study areas, as
148 estimated by two separate 2007 CMAQ simulations 38
149
2A-4
-------
150 1. Introduction
151 One of the aspects of ozone that is unusual relative to the other pollutants with National
152 Ambient Air Quality Standards (NAAQS) is that, periodically, in some locations, an appreciable fraction of
153 the observed ozone results from sources or processes other than local and regional anthropogenic
154 emissions of ozone precursors (Fiore et al., 2002). Any ozone formed by processes other than the
155 chemical conversion of local or regional ozone precursor emissions, such as nitrogen oxides (NOx) or
156 volatile organic emissions (VOC), is generically referred to as "background" ozone. As part of this review
157 of the ozone NAAQS, EPA completed an extensive review of the known aspects of background ozone
158 and summarized the findings in the Integrated Science Assessment (ISA) in March 2013 (USEPA, 2013).
159 The purpose of this appendix is to present the results from supplemental air quality modeling analyses
160 related to background ozone that were completed by EPA subsequent to the ISA. While these updated
161 analyses use a recent base year (2007) and consider an alternative modeling methodology which can
162 better account for non-linear ozone chemistry in some conditions, the results are largely consistent with
163 previous determinations about the magnitude of background ozone contributions across the U.S.
164 Away from the surface, ozone can have an atmospheric lifetime on the order of weeks. As a
165 result, background ozone can be transported long distances at heights above the boundary layer and,
166 when meteorological conditions are favorable, be available to mix down to the surface and add to the
167 total ozone loading from non-background sources. Generically, background ozone can originate from
168 natural sources of ozone and ozone precursors, as well as from far upwind manmade emissions of ozone
169 precursors. Natural sources of ozone precursor emissions such as wildfires, lightning, and vegetation
170 can lead to ozone formation by chemical reactions with other natural sources1. Another important
171 natural component of background is ozone that is naturally formed in the stratosphere through
172 interactions of UV light with atomic oxygen (O2). Stratospheric ozone can periodically mix down to the
173 surface at high concentrations, especially at higher altitude locations. The manmade portion of the
174 background includes any ozone formed due to anthropogenic sources of ozone precursors emitted far
175 away from the local area (e.g., international emissions). Finally, both biogenic and international
176 anthropogenic emissions of methane, which can be chemically converted to ozone over relatively long
177 time scales, can also contribute to global background ozone levels.
178 The precise definition of background ozone can vary depending upon context, but it generally
179 refers to ozone that is formed by sources or processes that cannot be influenced by actions within the
180 jurisdiction of concern. In the first draft policy assessment document (EPA, 2012), EPA presented three
181 specific definitions of background ozone: natural background, North American background, and U.S.
182 background. Natural background (NB) was the narrowest definition of background and it was defined as
183 the ozone that would exist in the absence of any manmade ozone precursor emissions. The other two
184 previously-established definitions of background presume that the U.S. has little influence over
185 anthropogenic emissions outside our continental or domestic borders. North American background
186 (NAB) is defined as that ozone that would exist in the absence of any manmade ozone precursor
Ozone formed through reactions between natural emissions and local anthropogenic emissions (e.g., biogenic
VOC with man-made NOx) is generally not considered to be background ozone.
2A-5
-------
187 emissions inside of North America. U.S. background (USB) is defined as that ozone that would exist in
188 the absence of any manmade emissions inside the United States. It is important to note that each of
189 these three definitions of background ozone requires photochemical modeling simulations to estimate
190 what the residual ozone concentrations would be were the various anthropogenic emissions to be
191 removed.
192 As noted in the first draft policy assessment, EPA has revised several aspects of our
193 methodology for estimating the change in health risk and exposure that would result from a revision to
194 the ozone NAAQS. First, risk estimates are now based on total ozone concentrations as opposed
195 previous reviews which only considered risk above background levels. Second, EPA is now using air
196 quality models to estimate the spatial patterns of ozone that would result from attaining various levels
197 of the NAAQS, as opposed to simplistic rollback techniques that required the estimation of a background
198 ozone "floor" beyond which the rollback would not take place. Both of these revisions have had the
199 indirect effect of obviating the need for estimating background ozone levels as part of the ozone risk
200 and exposure assessment (REA). Regardless, EPA expects that a well-founded understanding of the
201 fractional contribution of background sources and processes to surface ozone levels will be valuable
202 towards informing policy decisions about the NAAQS. Section 2 of this document will describe the
203 supplemental air quality modeling simulations that have recently been completed by EPA to bolster our
204 understanding of background ozone. Section 3 will present the results from the updated analyses and
205 provide estimates of average background ozone levels, and how they can vary in time and space across
206 the U.S. Based on the same modeling, Section 4 will consider the entire spectrum of variable
207 background ozone levels with special emphasis on areas and times in which background can approach or
208 exceed the level of the NAAQS. Section 5 will utilize the supplemental air quality modeling estimates to
209 determine the relative importance of specific components of background ozone. Section 6 will present
210 estimates of the overall fraction of ozone that is estimated to result from background sources or
211 processes in each of the 12 urban case study areas in the epidemiology study based analyses in Chapter
212 7 of the Risk and Exposure Assessment (REA) (EPA, 2014) based on the updated modeling. Finally,
213 Section 7 will conclude with a limited analysis of how background ozone levels impact longer-term
214 ozone metrics that may be important from a welfare perspective (i.e., W126).
215 2. Description of modeling methodologies
216 As noted above, air quality models are typically used to estimate background ozone as it is quite
217 difficult to measure directly. Without special monitoring, it is impossible to determine how much of the
218 ozone measured by a monitor originated from sources that are considered background. Even the most
219 remote monitors within the U.S. can periodically be affected by U.S. anthropogenic emissions. Previous
220 modeling studies have estimated what background levels would be in the absence of certain sets of
221 emissions by simply comparing the ozone differences between a base model simulation and a control
222 simulation in which emissions were removed. This basic approach is often referred to as "zero out"
223 modeling or "emissions perturbation" modeling. Examples of zero out modeling include the three major
224 studies summarized in the ISA (Zhang et al., 2011; Emery et al., 2012, Lin et al., 2013). It is important to
225 note that the specific concepts of NB, NAB, and USB are all explicitly tied to zero-out modeling, as those
226 definitions are based on estimating what remains in the absence o/specific sets of man-made emissions.
2A-6
-------
227 EPA has conducted and will describe updated air quality modeling for a 2007 base year that employs a
228 regional air quality model nested within a coarser-scale global chemical transport model to estimate NB,
229 NAB, and USB levels when the respective manmade emissions are zeroed. This modeling is described in
230 detail in section 2a.
231 While the zero-out approach has traditionally been used to estimate background ozone levels,
232 the methodology has some acknowledged limitations. First, from a policy perspective, the purely
233 hypothetical and ultimately unrealizable zero manmade emissions scenarios have limited application in
234 this regard. Secondly, the assumption that background ozone is what is left after specific emissions have
235 been removed within the model simulation can be misleading in locations where ozone chemistry is
236 highly non-linear. Depending upon the local composition of ozone precursors, NOx emissions
237 reductions can either increase or decrease ozone levels in the immediate vicinity of those reductions.
238 For those specific urban areas in which NOx titration of ozone can be significant, zero-out modeling can
239 result in inflated estimates of background ozone when these NOx emissions are completely and
240 unrealistically removed. Paradoxically, in certain times and locations in a zero-out scenario there can be
241 more background ozone than actual ozone within the model (EPA, 2014).
242 A separate modeling technique attempts to circumvent these limitations by apportioning the
243 total ozone within the model to its contributing source terms. This basic approach is referred to as
244 "source apportionment" modeling. While source apportionment modeling has not been previously used
245 in the context of estimating background ozone levels as part of an ozone NAAQS review, it has
246 frequently been used in other regulatory settings to estimate the "contribution" to ozone of certain sets
247 of emissions (EPA 2005, EPA 2011). The source apportionment technique provides a means of
248 estimating the contributions of user-identified source categories to ozone formation in a single model
249 simulation. This is achieved by using multiple tracer species to track the fate of ozone precursor
250 emissions (VOC and NOx) and the ozone formation caused by these emissions. The methodology is
251 designed so that all ozone and precursor concentrations are attributed to the selected source categories
252 at all times without perturbing the inherent chemistry. The zero out modeling attempts to determine
253 what ozone be in the absence of background sources. The source apportionment modeling attempts to
254 determine how much of the modeled ozone has resulted from background sources. EPA has conducted
255 and will describe new source apportionment modeling that employs a regional air quality model nested
256 within a coarser-scale global chemical transport model to assess the contributions of boundary
257 conditions and other potential background sources (e.g., wildfires, biogenic emissions, and
258 Canadian/Mexican emissions). This modeling is described in detail in section 2b.
259 a. 2007 GEOS-Chem/CMAQ zero-out modeling:
260 In order to provide estimates of the overall fraction of ozone that is estimated to result from
261 background sources in each of the 12 REA urban study areas, EPA conducted new modeling that utilized
262 the same model base year (2007) as was used in the ozone modeling that inform the risk and exposure
263 analyses (EPA, 2014, Appendix 4b). The EPA modeling used a model configuration similar to that of
264 Emery (2012), in that it nested a regional-scale (12 km) air quality model inside a global air quality model
2A-7
-------
265 simulation with a much coarser horizontal grid resolution (2.0 by 2.5 degrees). Figure la shows a map of
266 the model domain.
267 The global scale simulation utilized the GEOS-Chem model, version v8-03-02, except for the
268 chemistry package which was from version v8-02-01. The emissions estimates used in the 2007 base
269 year modeling were aggregated from a variety of sources, starting with the global Emissions Database
270 for Global Atmospheric Research (EDGAR) emission inventory. These initial estimates were then
271 improved by utilizing various area-specific inventories, such as the 2005 National Emissions Inventory
272 (NEI) for the U.S. portions of the domain, and available inventories for Asia, Canada, Europe, and
273 Mexico. In addition to the anthropogenic estimates, emissions were specified for a variety of
274 background sources including: lightning NO, soil NOx, wildfires, and biogenic VOC emissions. The
275 wildfire data is from the Global Fire Emissions Database (GFED). The biogenic VOC estimates were
276 simulated by the Model of Emissions of Gases and Aerosols from Nature (MEGAN) version 2.1. The
277 meteorological data is based on the Goddard Earth Observing System Model, Version 5 (GEOS-5)
278 analysis fields. More information on the global simulation is available within Henderson et al. (2013).
279 This reference also provides a broad evaluation of the ability of this specific GEOS-Chem configuration to
280 provide accurate lateral boundary conditions of ozone to finer-scale regional simulations. Using satellite
281 retrievals from the Tropospheric Emissions Spectrometer (TES), Henderson et al. (2013) concluded that
282 the GEOS-Chem ozone prediction biases and errors are generally within TES uncertainty estimates. For
283 instance, for the ozone season month of August, model predictions are within plus or minus 20 percent
284 of the satellite estimates between nearly 80 percent of the time, with slightly better performance along
285 the southern boundary.
286 The lateral boundary conditions from the global model were then used as inputs for a 12 km
287 horizontal resolution, CMAQ version 4.7.1, model simulation. Four scenarios were modeled: 1) a 2007
288 base case simulation which was the basis of the air quality modeling performed for the 2nd draft ozone
289 REA and is described in more detail in Appendix 4b of EPA (2014), 2) a natural background run with
290 anthropogenic ozone precursor emissions2 removed in both the global and regional models, 3) a North
291 American background run with anthropogenic ozone precursor emissions removed across North
292 America (global and regional model simulations), and 4) a U.S. background run with anthropogenic
293 ozone precursor emissions were removed over the U.S (global and regional model simulations).
294 Detailed analyses of EPA's 2007 zero out modeling results are provided in sections 3 through 6 of this
295 appendix.
296 An operational model performance evaluation was completed for surface ozone in the 2007
297 base simulation as described separately (EPA 2014, Appendix 4b). For the purposes of this analysis, EPA
298 assessed the model ability to reproduce measured daily maximum 8-hour (MDA8) ozone values and
299 seasonal mean MDA8 ozone concentrations for the period April to October 2007. As noted earlier, the
300 base year modeling in this analysis used climatological monthly-average wildfire emissions which are not
2 In the global model all ozone precursor species were removed (i.e., VOC, NOx, CO), except for methane which
was reset to pre-industrial levels to reflect natural contributions. In the regional modeling, the methane levels
were left unchanged.
2A-8
-------
301 intended to capture discrete events from specific fires that occurred in 2007, so perfect correlation
302 between observations and model predictions should not be expected. Figure Ib provides a density
303 scatterplot of the observed and predicted daily 8-hour ozone peaks paired in space and time for the
304 2007 CMAQ base. As can be seen, the majority of pairs line up along the 1:1 line. There is a tendency
305 for the model to overestimate site-days with low 8-hour ozone peaks, and underestimate the site-days
306 with higher peak ozone values. Modeled 8-hour ozone peak concentrations exhibited relatively small
307 bias and error compared to the observations. The average bias in IV1DA8 ozone estimates was 3.5 ppb.
308 Figure Ic depicts the spatial bias patterns in IV1DA8 ozone at all sites that measured valid ozone data for
309 at least 100 days during the April-October period. CMAQ overestimations are greatest along the Gulf
310 Coast region, along the Atlantic coastline, and over the central U.S. The majority of underestimated
311 seasonal mean MDA8 occurs within southern California. The model performance for the 2007 base
312 simulation is equivalent or better than typical state-of-the-science photochemical model performance
313 recently reported in the literature (Simon et al, 2012).
314 Certainly some remote monitoring locations are more affected by background sources than
315 other locations in the network. However, this and numerous other analyses have shown that even the
316 most remote ozone monitoring locations in the U.S. are periodically affected by U.S. manmade
317 emissions. In this analysis we carefully assess model performance to ensure that model error does not
318 influence the characterization of background ozone. As noted in the recent ISA (EPA, 2013), there is
319 greater confidence in the ability of the model to predict mean contributions from background sources
320 rather than individual events. Beyond the statistical analyses summarized in the previous paragraph and
321 in appendix 4b of the 2nd draft ozone REA (EPA, 2014), it is valuable to attempt to diagnose the model
322 ability to account for background ozone within the simulation. EPA assessed whether any correlation
323 existed between daily model biases and daily background ozone estimates. Figure Id shows that at
324 high-elevation sites (i.e., sites more than 1km above sea level) the highest estimates of natural
325 background ozone tend occur on days with greatest overestimation. Conversely, the site-days with the
326 lowest natural background estimates tend to occur when the model underestimates the observed daily
327 peaks at these sites. This relationship between background estimates and simulation bias appears to be
328 constrained to the mountainous portion of the Western U.S. Figure Id also shows that estimates of
329 natural background ozone greater than 60 ppb are associated with large over-predictions. However,
330 based on the relatively low model bias and the general lack of correlation between daily bias values and
331 background estimates, EPA believes that these model estimates can be used to help characterize
332 background ozone levels over the U.S. Although the highest background estimates should be
333 considered with caution.
2A-9
-------
334
335
336
337
I2US2 domain .
>,y origin; .24UOO
-------
341
342
343
344
345
^r CO
*
O *
ft „„„,_, O K L f
mi...
' O&'^fr
^"^r tb , f*S 0°,
"ffi_*_
^ ..&
Figure Ic. Bias in seasonal mean (April-October) maximum daily 8-hour ozone predictions in the 2007
CMAQ base simulation.
60-
*
i
3
i
* •
r
_ q
n u
*
* * t
Mi
EL J
t « *
, * *
•
t
t ' •
I *
!
l
t *
i
Model Bias (ppb)
t^d > 10 ppb under
i— H 6-10 ppb under
HH 0-3 ppb under
Hjjjj 0-3 ppb over
^P 3-6 ppb over
j— H 6-10 ppb over
|— — I > 10 ppb over
346
347
348
Low Med High
Elevation of site (m MSU: Low (< 500): Med (500-1000): High <> 1000)
Figure Id. Relationship between CMAQ estimations of MDA8 natural background ozone and daily
model biases.
2 A-11
-------
349 b. 2007 GEOS-Chem/CAMx source apportionment modeling:
350 The same global modeling described above was used to assign lateral boundary conditions to
351 the regional-scale (12 km) CAMx v5.0 source apportionment simulations. Wherever possible, the
352 emissions and meteorological inputs in the CAMx modeling were chosen to mimic the 2007 base CMAQ
353 simulation described earlier. Figure la shows a map of the model domain.
354 As with the CMAQ base case, a limited operational model performance evaluation was also
355 completed for surface ozone in the 2007 base simulation. For the purposes of this analysis, EPA
356 assessed the model ability to reproduce measured daily maximum 8-hour (MDA8) ozone values and
357 seasonal mean MDA8 ozone concentrations for the period April to October 2007. Figure 2a provides a
358 density scatterplot of the observed and predicted daily 8-hour ozone peaks paired in space and time for
359 the 2007 CAMx base simulation. As can be seen, the majority of pairs line up along the 1:1 line. Again,
360 there is a tendency for the model to overestimate site-days with low 8-hour ozone peaks and
361 underestimate the site-days with higher peak ozone values. Modeled 8-hour ozone peak concentrations
362 exhibited relatively small bias and error compared to the observations. The average bias in MDA8 ozone
363 estimates was 3.5 ppb. Figure 2b depicts the spatial bias patterns in MDA8 ozone at all sites that
364 measured valid ozone data for at least 100 days during the April-October period. CAMx overestimations
365 are greatest along the Gulf Coast region, along the Atlantic and Pacific coastlines, and within the
366 southeastern U.S. The majority of underestimated seasonal mean MDA8 occurs in California away from
367 the coastline.
368 The apportionment tools in CAMx utilized here to estimate the contribution of background
369 sources are well-established and have previously been peer-reviewed (UNC, 2009). EPA used the
370 Anthropogenic Precursor Culpability Assessment (APCA) tool in this analysis. The APCAtool attributes
371 ozone production to manmade sources whenever ozone is determined to result from a combination of
372 anthropogenic and biogenic emissions (Environ, 2011). The APCA methodology defines natural ozone as
373 the production resulting from the interaction of biogenic VOC with biogenic NOx emissions. Eleven
374 separate source categories were tracked in the source apportionment analysis, including five boundary
375 condition terms and six in-domain sectors:
376 • Boundary condition terms:
377 o Northern edge
378 o Eastern edge
379 o Southern edge
380 o Western edge
381 o Top boundary
382
383 • In-domain sectors:
384 o U.S. anthropogenic emissions
385 o Point sources located within the Gulf of Mexico
386 o Category 3 marine vessels outside State boundaries
387 o Climatologically-averaged wildfire emissions
2 A-12
-------
388
389
390
391
392
393
394
395
396
397
398
o Biogenic emissions
o Canada/Mexico emissions (only those sources within the domain)
It should be noted that the source apportionment modeling conducted here does not allow for
replication of natural background because of the construct of boundary conditions. The boundary
conditions for our applications can include ozone and/or ozone precursors that were originally
generated by natural sources, as well as ozone produced from far upstream anthropogenic emissions
(e.g., Asia). It is not possible to disentangle these two terms. Instead, the source apportionment
modeling is primarily used to help estimate background into the U.S., which is assumed to be the
contributions from nine of the modeled sectors; that is, everything except U.S. anthropogenic emissions
and point sources located within the Gulf of Mexico.
399
400
401
402
<
o
50
100
150
MDA8 observations (ppb)
Figure 2a. Density scatterplot comparing CAMx base daily peak 8-hour ozone predictions against
observed 8-hour ozone peaks paired in space and time for all sites during April-October 2007.
2 A-13
-------
403
404
405
Figure 2b. Bias in seasonal mean (April-October) maximum daily 8-hour ozone predictions in the 2007
CAMx base simulation.
406 3. Estimates of seasonal-average background ozone levels
407 This section of the appendix provides estimates of seasonal average background ozone levels
408 over the U.S. As noted in the introduction and as discussed in detail in the ISA, background ozone values
409 can vary significantly in space and time. There can be atypical episodes of higher background ozone
410 concentrations amidst the routine days that drive seasonal average background. The highest
411 background episodic concentrations are typically associated with stratospheric intrusions or wildfires.
412 These background "events" can be difficult to model as they require event-specific model inputs. The
413 primary goal of the EPA modeling was to estimate the seasonal average background concentrations
414 between April and October 2007. Previous analyses have shown that this is the period in which average
415 background levels are highest (Zhang et al., 2011). This section of the appendix focuses on seasonal
416 mean levels of background. (Section 4 will consider the upper range of possible background ozone.)
417 The analysis focus on the maximum daily 8-hour ozone average in ppb. This metric is referred to
418 as MDA8. This section will first present model estimates of seasonal mean ozone levels in the base
419 simulation. This will be followed by estimates of NB, NAB, and USB from the CMAQ zero out modeling.
420 After discussing the magnitudes of background levels, the section switches to a consideration of the
421 relative percentage of background to total ozone across the U.S. This portion of the text will utilize both
422 the CMAQ zero out and CAMx source apportionment modeling.
423 Figure 3a displays the 2007 base case, CMAQ model-predicted, seasonal mean (April-October)
424 MDA8 ozone concentrations in grid cells with active monitoring locations over the U.S. The model
425 results are shown at the monitoring site level as opposed to in the default gridded format to foster
426 subsequent site-level estimates of background magnitudes. Each grid cell containing an Air Quality
427 System (AQS) ozone monitor that was collecting valid data in 2007 was identified and the model
2 A-14
-------
428 background estimates were extracted for those grid cells and displayed accordingly. The base
429 predictions are provided for context to allow easier interpretation of the following plots which isolate
430 specific background levels. As can be seen, most of the U.S. experiences seasonal mean IV1DA8 ozone
431 levels greater than 50 ppb in the base case simulation. The median value over the 1,294 monitoring
432 locations is 52.5 ppb.
433 Figure 3b provides an estimate of what seasonal-average IV1DA8 would be in a natural
434 background scenario, using the 2007 EPA zero out modeling. Again, in this GEOS-Chem/CMAQ
435 simulation, all anthropogenic ozone precursor emissions were removed from both the global and
436 regional simulations, and methane levels were adjusted to pre-industrial levels in the global simulation.
437 As shown, natural background ozone levels range from approximately 15-35 ppb with the highest values
438 occurring over the higher-elevation sites in the western U.S. The median value over these locations is
439 24.2 ppb, and more than 50 percent of the sites have natural background levels of 20-25 ppb. The
440 highest modeled estimate of seasonal average, natural background, IV1DA8 ozone is 34.3 ppb at the
441 high-elevation CASTNET site (Gothic) in Gunnison County, CO.
442 Figures 3c and 3d show the same information for the North American and U.S. background
443 scenarios. In these model runs, all anthropogenic ozone precursor emissions were removed from the
444 U.S., Canada, and Mexico (NAB scenario) and then only the U.S. (USB scenario). The figures show that
445 there is not a large difference between the NAB and USB scenarios. Seasonal mean IV1DA8 NAB and USB
446 ozone levels range from 25-50 ppb, with the most frequent values estimated in the 30-35 ppb bin. The
447 median seasonal mean background levels are 31.5 and 32.7 ppb (NAB and USB, respectively). Again, the
448 highest levels of background are predicted over the intermountain western U.S. Locations with NAB and
449 USB concentrations greater than 40 ppb are confined to Colorado, Nevada, Utah, Wyoming, northern
450 Arizona, eastern California, and parts of New Mexico. Similar to NB, the highest NAB and USB levels
451 were modeled to occur at the Gothic CO site (46.7/47.7). This remote rural site is located 2,926 meters
452 (9,600 feet) above mean sea level and should not be considered representative of background ozone at
453 lower-altitude, more-populated regions. The high USB and NAB values along the Gulf Coast are most
454 likely due to model biases.
455 Absolute model estimates of various background definitions are useful, but they can be
456 influenced by any local biases and errors in the modeling. A separate way to look at the role of
457 background in seasonal mean ozone levels is to consider the fractional contribution of NB, NAB, and USB
458 to total ozone at each location. Considering the proportional role of background allows for an
459 informative comparison between the two modeling approaches without having to account for the
460 differences in base case biases and errors. Figures 4a, 4b, and 4c show the estimated fractional
461 contribution of NB, NAB, and USB to total seasonal average MDA8 ozone levels at the monitoring
462 locations from the CMAQ zero out modeling. The modeling estimates that approximately 35-80 percent
463 of the seasonal average MDA8 ozone at monitoring locations is due to natural background sources. A
464 majority sites have NB fractions between 40 and 60 percent. The mean natural background proportion
465 over all sites is 47 percent. That is, when all global anthropogenic emissions are removed and global
466 methane levels in GEOS-Chem are restored to pre-industrial levels, seasonal average MDA8 levels are
467 reduced by approximately half. The fractional proportions of NAB and USB are very similar. In both
2 A-15
-------
468 cases, most sites have background fractions that range from 50 to 80 percent. The mean NAB fraction
469 (to seasonal mean MDA8) is 63 percent. The mean USB fraction is 66 percent.
470 As noted in the introduction, the advantage of the source apportionment modeling is that all of
471 the modeled ozone is attributed to various source terms and thus this approach is not affected by the
472 confounding occurrences of background ozone values exceeding the base ozone values as can happen in
473 the zero out modeling (i.e., background proportions > 100%). Consequently, one would expect the
474 fractional background levels to be lower in the source apportionment methodology as a result of
475 removing this artifact. It is also important to remember that the terms NB, NAB, and USB are explicitly
476 linked to the zero out modeling approach. (USB is the ozone that would exist in the absence of U.S.
477 anthropogenic emissions.) In contrast, the source apportionment modeling performed here provides
478 estimates the amount of IV1DA8 ozone that is attributable to U.S. anthropogenic emissions relative to
479 total base model ozone. Figure 4d shows the relative contribution from sources other than U.S.
480 anthropogenic emissions to total seasonal mean MDAS ozone based on the 2007 source apportionment
481 modeling. The fractional contribution fields between CMAQ zero out USB estimates and CAMx source
482 apportionment estimates of source other than U.S. anthropogenic emissions are quite similar. The
483 spatial patterns in Figures 4c and 4d are consistent, with the highest fractional contributions from
484 sources other than U.S. anthropogenic emissions occurring along U.S. borders and over the
485 intermountain western States. The source apportionment modeling estimates that approximately 40-
486 80% of the seasonal average MDAS ozone at monitoring locations is due to sources other than
487 manmade ozone precursor emissions from the U.S. A majority of sites have non-U.S. fractions between
488 40 and 70 percent. The mean proportion attributable to international and natural sources over all sites
489 is 59 percent. Despite the differences in the methodologies this is very similar to the mean USB
490 estimate of 66 percent from the zero out modeling.
491
492
493
2 A-16
-------
494
495
496
497
498
Ozone (ppb)
< 20 (0)
20 - 25 (0)
0 25-30 (6)
0 30-35(10)
0 35-40 (38)
O 40-45(112)
O 45-50 (264)
O 50-55 (488)
55 - 60 (300)
> 60 (75)
Figure 3a. April-October average „-„.
estimated by a 2007 CMAQ base simulation.
M ,- > i , n
MDA8 ozone (ppb) at monitoring locations across the U.S. as
499
500
501
f * *
^H •- ^^k iiidi^k
_»
• •**
^ *^^^^^^ ^^ '
- I
• 2
nmwrfrfi
Ozone (ppb)
< 20 (104)
20 - 25 (740)
25-30(331)
30 - 35 (119)
35 - 40 (0)
O 40-45 (0)
O 45-50 (0)
50 - 55 (0)
55 - 60 (0)
> 60 (0)
Figure 3b. April-October average natural background MDA8 ozone (ppb) at monitoring locations
across the U.S. as estimated by a 2007 CMAQ zero out simulation.
2 A-17
-------
502
503
504
505
506
o,
Ozone (ppb)
<20(0)
20-25 (0)
25-30 (396)
30-35 (628
Q 35-40(147)
O 40-45(121)
O 45-50(2)
50-55 (0)
55-60 (0)
> 60 (0)
Figure 3c. April-October average North American background MDA8 ozone (ppb) at monitoring
locations across the U.S. as estimated by a 2007 CMAQ zero out simulation.
507
508
509
Ozone (ppb)
•= 20 (0)
20-25 (0)
25-30(127)
30-35 (842)
35 - 40 (188)
O 40-45(132)
O 45-50 (5)
© 50-55 (0)
0 55-60 (0)
0 > 60 (0)
Figure 3d. April-October average United States background MDA8 ozone (ppb) at monitoring
locations across the U.S. as estimated by a 2007 CMAQ zero out simulation.
2 A-18
-------
510
511
512
513
514
< 40% (154)
40 - 50% (752)
50-60% (371)
60-70% (12)
70-80% (3)
> 30% (1l
Figure 4a. Ratio of natural background to total April-October average MDA8 ozone at monitoring
locations across the U.S. as estimated based on 2007 CMAQ simulations.
515
516
517
<40%<0)
40-50% (21)
50-60% (539)
60-70% (475)
70 - 80% (223)
> 80% (35)
Figure 4b. Ratio of A/. American background to total April-October average MDA8 ozone at
monitoring locations across the U.S. as estimated based on 2007 CMAQ simulations.
2 A-19
-------
518
519
520
521
522
523
524
525
Figure 4c. Ratio of U.S. background to total April-October average MDA8 ozone at monitoring
locations across the U.S. as estimated based on 2007 CMAQ simulations.
< 40% (0)
40 - 50% (343)
50 - 60% (483)
SO - 70% (237)
O 70-80% (178)
> 80% (52)
Figure 4d. Ratio of sources other than U.S. anthropogenic emissions to total April-October average
MDA8 ozone at monitoring locations across the U.S. as estimated by a 2007 CAMx source
apportionment simulation.
2A-20
-------
526 4. Distributions of background ozone levels
527 As a first-order understanding, it is valuable to be able to characterize seasonal mean levels of
528 background ozone. However, it is well established that background levels can vary substantially from
529 day-to-day. From an implementation perspective, the values of background ozone on possible
530 exceedance days is a more meaningful distinction. The first draft policy assessment (EPA, 2012)
531 considered this issue in detail, via summaries of the existing 2006 zero out modeling (Henderson et al.,
532 2012), and concluded that "results suggest that background concentrations on the days with the highest
533 total ozone concentrations are not dramatically higher than typical seasonal average background
534 concentrations." Based on this finding, the 1st draft policy assessment determined that "anthropogenic
535 sources within the U.S. are largely responsible for 4th highest 8-hour daily maximum ozone
536 concentrations." This portion of the appendix will consider the entire spectrum of variable background
537 ozone levels with special emphasis on days in which base model ozone concentrations approach or
538 exceed the level of the NAAQS.
539 The 2007 modeling agrees with the finding from the previous 2006-based modeling analyses
540 that the highest modeled ozone site-days tend to have background ozone levels similar to mid-range
541 ozone days. Figures 5a-5c show the distribution of April-October IV1DA8 background levels (NB, NAB,
542 and USB, respectively) from the CMAQ zero out runs. As noted in section 2, zeroing out emissions can
543 remove the effects of local NOx titration and result in modeled background values that are higher than
544 the base model ozone. The "box and whisker" plots shown in these figures display four key features of
545 the distributions:
546 a. the median concentration (black horizontal line) per bin,
547
548 b. the inter-quartile range (blue colored box) which represents the 25th-75th percentile range in
549 values within the distribution,
550
551 c. the "whiskers" (dark gray vertical lines with top and bottom whiskers) which represent the
552 range of values within 1.5 times the inter-quartile range, and
553
554 d. the "outliers" (gray points) which are any values outside the whiskers.
555 As can be seen in Figure 5a, natural background values do not vary greatly as a function of the
556 base modeled ozone. Recall that the seasonal average natural background IV1DA8 ozone values were
557 modeled to range from 15-35 ppb across the U.S. with a median value of 24 ppb. The highest values
558 were at the high-elevation sites in the western U.S. Based on the distributional analysis, the 75th
559 percentile values are on the order of 30 ppb. Natural background levels exceeding 40-45 ppb are
560 considered to be statistical outliers, due to their infrequency. Figure 5b shows the same type of
561 distributions but for NAB instead of NB. NAB values are generally 6-12 ppb higher than their NB
562 counterparts, due to the affect of higher global methane values and the influence of anthropogenic
563 emissions from Asia. It was previously reported (in section 3) that the median seasonal average NAB
564 IV1DA8 values were 31.5 ppb. Based on the distributions, it can be seen that 75th percentile values are
2A-21
-------
565 approximately 40 ppb; it is rare for NAB MDAS values to exceed 50-55 ppb. NAB values are constant in
566 magnitude once the base ozone exceeds 50 ppb indicating that the higher base ozone values are driven
567 by non-NAB sources (i.e., North American emissions). Finally in Figure 5c, the USB MDAS distributions
568 by base model MDAS are shown. The results are similar to NAB.
569 Figure 5d shows the results from the source apportionment modeling of non-U.S. anthropogenic
570 source contributions to MDAS ozone (i.e., the nine source apportionment categories other than U.S.
571 anthropogenic emissions and Gulf of Mexico point sources). This non-counterfactual approach is
572 expected to give a better indication of background levels at low concentrations. At low levels, almost all
573 of the ozone is determined to be from background origins. The CAMx modeling shows that
574 contributions from non-U.S. anthropogenic emissions peak when base ozone ranges from 45-55 ppb and
575 then drop off slightly at higher base MDAS values. The source apportionment modeling of non-U.S.
576 impacts (similar to USB) indicates slightly lower background levels than the zero out modeling. The 75th
577 percentile values are generally less than 35 ppb, compared to 40 ppb in the zero out modeling. It is rare
578 to have background impacts greater than 55ppb. Interestingly, when base model MDAS ozone exceeds
579 70 ppb, it is rare to have background impacts greater than 45 ppb in the CAMx source apportionment
580 modeling.
581 Figures 6a-6d show the equivalent plots as 5a-5d, but use background fractions (background
582 MDAS / base MDAS) as the dependent variable instead of the absolute background concentrations.
583 These plots show the same effect; that is, the proportional relative contribution of background sources
584 and processes decreases as peak ozone increases. For natural background (Figure 6a), the median
585 fractions drop from 50% background for values between 45-50 ppb to only 35% background for base
586 MDAS values between 70-75 ppb. For NAB and USB (Figures 6b and 6c), the median fractions drop from
587 70% background for values between 45-50 ppb to only 45% background for base MDAS values between
588 70-75 ppb. The source apportionment modeling (Figure 6d) estimates less of a proportional role of non-
589 U.S. anthropogenic emissions. In that modeling, the median fractions drop from 65% background for
590 values between 45-50 ppb to only 35% background for base MDAS values between 70-75 ppb. A key
591 observation, as noted in the first draft policy assessment document, is that the relative importance of
592 background decreases on days most likely to violate the NAAQS. An additional policy-relevant finding
593 from the distributional analyses is that the relative role of background sources would be increased if the
594 level of the NAAQS were lowered. At 60 ppb, the modeling suggests that the median fractional
595 contribution from background is 45-55 percent, but there can be cases where background comprises 80-
596 90 percent of the total ozone.
597 Many of the cases when background ozone is estimated to contribute in large proportions to
598 relatively high ozone days may be eligible for consideration as exceptional events, but again, this
599 modeling is not designed to resolve specific events that occurred in 2007. While there is greater
600 confidence in the model's ability to predict mean contributions from background sources than from
601 individual events, it is also useful to briefly consider the upper end of the background ozone
2A-22
-------
602 distributions. Figure 7 shows the 95th percentile3 USB estimates from the zero out modeling. The 95th
603 percentile MDAS USB ozone levels range from 35-60 ppb, with the most frequent values residing in the
604 35-40 and 40-45 ppb bins. The median 95th percentile background USB ozone level is 42.0 ppb. As with
605 the seasonal mean MDAS USB, the highest levels of high background days (i.e., 95th percentile days) are
606 observed over the intermountain western U.S. At these locations, 95th percentile USB levels can exceed
607 50 ppb. Background values at the 95th percentile end of the distribution are 4-12 ppb higher than the
608 mean background values at the same locations.
609
3 During the April-October period, there were 214 days of modeling results. Thus, the 95th percentile values
represent approximately the 10th highest days from the distribution.
2A-23
-------
610
611
612
613
n
a.
0.50-
CO
i
•o
I-
U)
it
U
a
CD
id
<2E 25-30 30-35 25-40 ^0-^5 45-50 50-55 55-50 50-55 65-70 70-75 75-80 80-86 85-90 90-95 96-100 > 100
Bins of Base Model MDA8 Ozone (ppb)
Figure 5a. Distribution of natural background MDA8 ozone (ppb) at monitoring locations across the
U.S. (Apr-Oct), binned by base modeled site-day MDA8, as estimated by 2007 CMAQ simulations.
i
•o
u
re
m
c
n
u
E
<
614
615
616
<25 25-30 30-35 35-40 40-45 45-50 50-55 55-60 60-65 65-70 70-75 75-30 SO-S5 85-90 90-95 95-100 > 100
Bins of Base Model MDA8 Ozone (ppb)
Figure 5b. Distribution of N. American background MDA8 ozone (ppb) at monitoring locations across
the U.S. (Apr-Oct), binned by base modeled site-day MDA8, as estimated by 2007 CMAQ simulations.
2A-24
-------
a.' :•
Q
•D
i
o
O)
CD
w
-
617
618
619
620
<25 26-30 30-35 35-40 40-45 45-50 50-55 55-60 60-65 65-70 70-75 75-80 80-85 85-90 90-95 95-100 > 130
Bins of Base Model MDA8 Ozone (ppb)
Figure 5c. Distribution of U.S. background MDA8 ozone (ppb) at monitoring locations across the U.S.
(Apr-Oct), binned by base modeled site-day MDA8, as estimated by 2007 CMAQ simulations.
621
622
623
624
0BBB
<25 25-30 30-35 35-40 40-45 45-50 50-55 55-60 60-6-5 65-70 70-75 75-80 80-85 85-90 90-95 95-100 >130
Bins of Base Model MDA8 Ozone (ppb)
Figure 5d. Distribution of MDA8 ozone contributions from non-U.S. manmade sources (ppb) at
monitoring locations across the U.S. (Apr-Oct), binned by base modeled site-day MDA8, as estimated
by 2007 CAMx simulations.
2A-25
-------
1.25-
625
626
627
628
<25 26-30 30-35 36-40 40-45 46-50 50-55 55-60 60-65 65-70 70-75 75-80
Bins of Base Model MDA8 Ozone (ppb)
&3-9E SJ-100 > 100
Figure 6a. Distribution of natural background MDA8 ozone fractions at monitoring locations across
the U.S. (Apr-Oct), binned by base modeled site-day MDA8, as estimated by 2007 CMAQ simulations.
1.25-
629
630
631
632
<25 25-30 30-35 35-40 40-45 45-60 50-55 55-60 60-65 6E-70 70-7E 75-30 80-85 85-90 90-95 96-100 > 100
Bins of Base Model MDA8 Ozone (ppb)
Figure 6b. Distribution of N. American background MDA8 ozone fractions at monitoring locations
across the U.S. (Apr-Oct), binned by base modeled site-day MDA8, as estimated by 2007 CMAQ
simulations.
633
2A-26
-------
634
635
636
1.25-
5>1.00-
m
u
ffi
o.oo-
<25 26-30 30-35 3E-40 40-4E
SE SB-BO &0-9E SE-130 > 100
Bins of Base Model MDA8 Ozone (ppb)
Figure 6c. Distribution of U.S. background MDA8 ozone fractions at monitoring locations across the
U.S. (Apr-Oct), binned by base modeled site-day MDA8, as estimated by 2007 CMAQ simulations.
1.25 -
637
638
639
640
641
o.oo-
e25 25-30 30-35 35-40 40-45 4E-EO EO-EE EE-cO EO-" ?E-73 73-~E 7E-33 80-S5 85-90 90-95 95-100 > 100
Bins of Base Model MDA8 Ozone (ppb)
Figure 6d. Distribution of MDA8 ozone fractions from non-U.S. anthropogenic sources at monitoring
locations across the U.S. (Apr-Oct), binned by base modeled site-day MDA8, as estimated by the 2007
CAMx simulation.
2A-27
-------
642
643
644
645
646
20-25(0)
26-30(01
30-35(0)
36-40(485)
40-45(571)
Q 45-50(116)
50-55(116)
55 - 60 (5)
> 60 (0)
Figure 7. April-October 95th percentile United States background MDA8 ozone (ppb) at monitoring
locations across the U.S. as estimated by a 2007 CMAQ base simulation.
2A-28
-------
647 5. Contribution of various processes and sources to total background ozone
648 This section will utilize the supplemental 2007 air quality modeling estimates to determine the
649 relative importance of specific elements of background ozone. Comparing the differences between the
650 three zero out scenarios can provide some information about the role of certain sets of emissions.
651 Figure 8a compares the NAB (zero out North American manmade emissions) and USB (zero out U.S.
652 manmade emissions) scenarios. The difference between these two runs is the inclusion of
653 anthropogenic emissions within the Canada and Mexico portions of the modeling domain. These
654 emissions contribute less than 2 ppb to the seasonal mean IV1DA8 ozone levels over most of the U.S.
655 There are 70 sites, near an international border, where the modeling estimates Canadian/Mexican
656 seasonal average impacts of 2-4 ppb. While not shown, the modeled peak single day impacts from
657 these specific international emissions sources can approach 25 ppb (e.g., San Diego, Buffalo NY). Figure
658 8b compares the NB (zero out all manmade emissions and reset GEOS-Chem methane values to pre-
659 industrial levels) to the NAB. The difference between these two runs is the inclusion of global methane
660 emissions related to recent human activity as well as anthropogenic emissions outside of North America.
661 These emissions are estimated to contribute 6-15 ppb to seasonal mean ozone levels over the U.S. The
662 most frequent bin is the 8-10 ppb increase. It is not possible via these runs to parse out what fraction of
663 this change is due to international emissions as opposed to methane emissions, but the ISA summarized
664 existing modeling (Zhang et a/., 2011) that suggested that the rise in methane from pre-industrial levels
665 to present-day levels led to increases in seasonal average ozone levels of 4-5 ppb. The greatest impacts
666 from these sources occurs over the western U.S., where international emissions would be expected to
667 have the largest impacts.
668 Figures 9a-9g show the fractional contribution to total seasonal mean IV1DA8 values of
669 individual source sectors that were tracked in the CAMx source apportionment modeling. Figure 9a
670 shows the impact from the regional model boundary conditions. The ozone entering the model domain
671 via the boundary conditions could have a variety of origins including: a) natural sources of ozone and
672 ozone precursors (including methane) emanating from outside the domain, b) anthropogenic sources of
673 ozone precursors (including methane) from international emitters, and c) some fraction of U.S.
674 emissions (natural and anthropogenic) which are exported and then re-imported into the domain via
675 synoptic-scale recirculation. Thus, one should not presume that the boundary condition contribution is
676 directly tied to any particular background definition. At most locations, boundary conditions
677 contributed 40-60 percent of the total IV1DA8 seasonal mean at sites across the U.S. The highest
678 proportional impacts from the boundary conditions (the top boundary contributes negligibly) are along
679 the coastlines and the intermountain West.
680 Figure 9b shows the source apportionment contribution (to seasonal mean MDA8) from the
681 most significant sector that was tracked: U.S. anthropogenic ozone precursor emissions. Again the most
682 common outcome at an individual site was that 40-60% of the seasonal mean ozone values originated
683 from U.S. anthropogenic emissions. The locations with smaller fractional contributions (e.g., 10-20
684 percent) from U.S. sources are generally located in places where ozone values are typically low such as
685 the Pacific Northwest. Figures 9c-9g display the fractional contributions from the other five in-domain
686 sectors listed in section 2. The impacts from these sectors are briefly summarized below:
2A-29
-------
687 • Biogenic emissions:
688 o Most frequent bin: 3-5 percent
689 o Highest site-specific contribution: 10-20 percent
690 o Region with greatest impacts: Great Plains states where soil NOx emissions are large
691 • Climatologically-average fire emissions:
692 o Most frequent bin: 0-1 percent
693 o Highest site-specific contribution: 3-5 percent
694 o Region with greatest impacts: California, Kansas/Oklahoma region
695 • Within-domain Canadian/Mexican manmade emissions:
696 o Most frequent bin: 0-1 percent
697 o Highest site-specific contribution: 10-20 percent
698 o Region with greatest impacts: Sites along international borders (NY, VT, CA, AZ, TX)
699 • Category 3 marine vessels outside U.S. territorial waters:
700 o Most frequent bin: 0-1 percent
701 o Highest site-specific contribution: 10-20 percent
702 o Region with greatest impacts: Coastal sites (especially southern CA)
703 • Gulf of Mexico point sources4:
704 o Most frequent bin: 0-1 percent
705 o Highest site-specific contribution: 1-3 percent
706 o Region with greatest impacts: Sites in southeast TX and southern LA
707
708
709
710
This sector was also included as part of U.S. anthropogenic source impacts in Figure 9b, but is broken out
separately in Figure 9g.
2A-30
-------
711
712
713
714
715
716
717
718
719
Figure 8a. Difference in April-October average MDA8 ozone (ppb) at monitoring locations across the
U.S. between the USB scenario and the NAB scenario. The difference between these two runs isolates
the impact of within-the-domain anthropogenic emissions from Canada and Mexico.
Ozone (ppb)
< 1 (0)
1 - 2 (0)
2-3(0)
3-4(0)
4-5(0)
O 6-8(141)
O 8-10(821)
O 10-12(267)
12-15(65)
Figure 8b. Difference in April-October average MDA8 ozone (ppb) at monitoring locations across the
U.S. between the NAB scenario and the NB scenario. The difference between these two runs isolates
the impact of the rise in global methane emissions from the pre-industrial and anthropogenic
emissions from outside North America.
2A-31
-------
720
721
722
723
724
725
726
727
^ Norlk
G R E A
^^^^
9\ «o
c&Q
U n i I e^T^
Stale
Mexico
3-5%|0)
5-10%(0)
10-20%(0)
20-40% (212)
40-60% (849)
> 60% (232)
Figure 9a. Percentage of April-October average MDA8 ozone that is apportioned to boundary
conditions as estimated at monitoring locations by a 2007 CAMx simulation.
O 10-20% (59)
20-40% (441)
40-60% (791)
> 60% (0)
Figure 9b. Percentage of April-October average MDA8 ozone that is apportioned to U.S.
anthropogenic sources as estimated at monitoring locations by a 2007 CAMx simulation.
2A-32
-------
728
729
730
731
732
733
734
735
0-1% (42)
1 - 3% (332)
3 - 5% (527)
O 5-10% (362)
O 10-20% (30)
20-40%(0)
40-60%(0)
> 60% (0)
Figure 9c. Percentage of April-October average MDA8 ozone that is apportioned to purely biogenic
emissions as estimated at monitoring locations by a 2007 CAMx simulation.
0-1% (1,104)
1-3% (181)
3 - 5% (8)
5-10%(0)
O 10-20%(0|
20-40%(0)
40-60%(0)
> 60% 10)
Mexico
Somces: USG.S. ESRI'TAIIA. AND, Soul
Figure 9d. Percentage of April-October average MDA8 ozone that is apportioned to climatological fire
emissions as estimated at monitoring locations by a 2007 CAMx simulation.
2A-33
-------
736
737
738
739
740
741
742
743
744
745
0-1% (704)
1 - 3% (407)
3-5% (124)
5-10% (49)
O 10-20% (9)
20-40%(0)
40-60%(0)
> 60% (0)
Figure 9e. Percentage of April-October average MDA8 ozone that is apportioned to anthropogenic
emissions from in-domain Canadian and Mexican sources as estimated at monitoring locations by a
2007 CAMx simulation.
0-1% (974)
1-3% (193)
3-5% (66)
5-10% (56)
O 10-20% (4)
20-40%(0)
40-60%(0)
> 60% (0)
Figure 9f. Percentage of April-October average MDA8 ozone that is apportioned to Category 3 marine
vessel emissions beyond U.S. territorial waters as estimated at monitoring locations by a 2007 CAMx
simulation.
2A-34
-------
746
°- 1% (1.275)
1-3% (18)
3-5% (01
O 5-10% 10)
O 10-20% (0)
20-40%(0)
40 -60% (0)
> 60% (0)
747 Figure 9g. Percentage of April-October average MDA8 ozone that is apportioned to Gulf of Mexico
748 point sources as estimated at monitoring locations by a 2007 CAMx simulation.
749
2A-35
-------
750 6. Estimates of the fractional background contribution to total ozone in 12 specific areas
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
This penultimate section of the appendix presents estimates of the overall fraction of ozone that
is estimated to result from background sources or processes based on the updated modeling in each of
the 12 urban case study areas in the epidemiology study based analyses in Chapter 7 of the Risk and
Exposure Assessment (REA). Tables la-lc summarize the CAMx-estimated fractional contributions of
sources other than U.S. anthropogenic emissions to total ozone in each of the 12 areas. Table la shows
that the fractional contributions from sources other than anthropogenic emissions within the U.S. to
seasonal mean IV1DA8 levels can range from 43 to 66 percent across these 12 urban areas. These
fractions are consistent with the national ratios summarized in section 3, although the urban fractions of
background tend to be smaller than at rural sites. As shown in section 4, the fractional contributions
from background are smaller on days with high modeled ozone (i.e., days that may exceed the level of
the NAAQS). Table Ib provides the fractional contributions from these non-U.S. sources, only
considering days in which base model IV1DA8 ozone was greater than 60 ppb. As expected, the fractional
background contributions are less and range from 31 to 55 percent. Rather than taking the fractions of
the seasonal means (as in Table la), Table Ic displays the mean and median daily IV1DA8 background
fractions. These metrics may be more appropriate for application to health studies, but as can be seen
the fractional contribution to backgrounds calculated via this approach are very similar to the Table la
calculations. For completeness sake, although EPA expects the source apportionment results to provide
a more realistic estimate of fractional background values, for completeness, we also provide USB
fractions based on zero out modeling for the 12 cities (see Table Id). The results are similar to the
source apportionment findings (compare against Table la), but the zero out technique provides slightly
higher background proportions.
All days, CAMx
Model MDAS seasonal mean
Model MDAS seasonal mean
from emissions other than
U.S. anthropogenic sources
Fractional contribution from
background
ATL
59.3
25.3
0.43
BAL
54.4
25.9
0.48
BOS
43.0
26.2
0.61
CLE
48.9
25.7
0.52
DEN
47.3
31.3
0.66
DET
39.1
23.3
0.60
HOU
48.5
27.0
0.56
LA
51.1
29.1
0.57
NYC
45.4
24.5
0.54
PHI
48.7
24.2
0.50
SAC
46.4
29.7
0.64
STL
49.8
24.3
0.49
773
774
775
776
Table la. April-October average MDAS ozone, average MDAS ozone from sources other than U.S.
manmade emissions, and the fractional contribution of these background sources in the 12 REA urban
study areas, as estimated by a 2007 CAMx simulation.
2A-36
-------
Only days w/ base
MDA8 > 60 ppb
Model MDAS seasonal mean
Model MDAS seasonal mean
from emissions other than
U.S. anthropogenic sources
Fractional contribution from
background
ATL
74.0
25.4
0.34
BAL
75.3
23.7
0.31
BOS
70.7
24.4
0.35
CLE
72.0
25.4
0.35
DEN
67.5
37.3
0.55
DET
68.9
24.4
0.35
HOU
70.3
28.0
0.40
LA
74.4
31.9
0.43
NYC
74.1
23.5
0.32
PHI
74.0
22.9
0.31
SAC
68.3
32.1
0.47
STL
70.0
25.4
0.36
777
778
779
780
781
782
Table Ib. Average MDAS ozone, average MDAS ozone from sources other than U.S. manmade
emissions, and the fractional contribution of these background sources in the 12 REA areas, as
estimated by a 2007 CAMx simulation using site-days in which base MDAS ozone exceeded 60 ppb.
Mean of daily MDAS
background fractions
Median of daily MDAS
background fractions
ATL
0.46
0.43
BAL
0.53
0.51
BOS
0.68
0.73
CLE
0.58
0.54
DEN
0.69
0.69
DET
0.64
0.66
HOU
0.59
0.59
LA
0.61
0.60
NYC
0.61
0.63
PHI
0.56
0.54
SAC
0.67
0.66
STL
0.52
0.49
783
784
785
786
787
Table Ic. Fractional contribution of non-U.S. manmade emissions sources in the 12 REA urban study
areas, as estimated by a 2007 CAMx simulation using means and medians of daily MDAS fractions.
All days, CMAQ
Model MDAS seasonal mean
Model MDAS seasonal mean
from USB emissions
Fractional contribution from
background
ATL
58.6
30.0
0.51
BAL
55.6
29.9
0.54
BOS
45.2
28.5
0.63
CLE
51.8
31.6
0.61
DEN
57.1
42.2
0.74
DET
43.5
31.7
0.73
HOU
49.4
33.0
0.67
LA
54.8
33.3
0.61
NYC
47.7
29.1
0.61
PHI
50.5
29.4
0.58
SAC
51.9
34.4
0.66
STL
52.6
32.0
0.61
788
789
790
791
Table Id. April-October average MDAS ozone, average MDAS ozone from USB, and the fractional
contribution of these background sources in the 12 REA urban study areas, as estimated by two
separate 2007 CMAQ simulations.
2A-37
-------
792 7. Background ozone and W126
793 As discussed in section 5 of the second draft policy assessment, EPA is considering the adequacy
794 of the current secondary standard to protect against welfare effects. One metric that has been
795 considered previously as a potential cumulative seasonal index is the W126 metric. The W126 index is a
796 sigmoidally weighted sum of all hourly O3 concentrations observed during a specified daily and seasonal
797 time window, where each hourly O3 concentration is given a weight that increases from 0 to 1 with
798 increasing concentration (Lefohn et al, 1988). The weights are defined such that values of 0.060 ppm
799 get a weight of ~0.3; 0.070 ppm values get a weight of ~0.6; and 0.085 ppm values get a weight of ~0.9.
800 The remainder of this section uses the 2007 zero out modeling to conduct a limited assessment of the
801 role of background ozone on W126 levels over the U.S.
802 The analysis of background influence on W126 is not as detailed as the analyses related to
803 seasonal mean IV1DA8 ozone. Instead of considering impacts at every monitoring location, EPA assessed
804 NB, NAB, and USB influences at four sample locations: Atlanta GA, Denver CO, Farmington NM, and
805 Riverside CA. Each of these four locations had relatively high observed values of W126 in 2010-2012.
806 Atlanta is an urban area in the Eastern U.S. with high primary ozone design values but relatively low
807 levels of seasonal background ozone. Riverside and Denver also have high primary ozone design values
808 but are in the Western U.S. where background ozone levels are generally higher. Farmington NM was
809 chosen as a site that has relatively lower primary ozone design values along with its relatively high W126
810 levels. The varying characteristics of each of these locations perhaps allows broader national
811 extrapolation of the 4-site results.
812 In previous EPA reviews of the O3 NAAQS, the influence of background ozone was estimated
813 according to a counterfactual (i.e., how much ozone would exist in the absence of certain sets of
814 emissions). In the current review, EPA is supplementing the counterfactual assessment with analyses
815 that estimate the fraction of the existing ozone that is due to background sources. This has important
816 ramifications for assessing the influence of background on W126 concentrations, because of the non-
817 linear weighting function used in the metric which emphasizes high ozone hours (e.g., periods in which
818 ozone is greater than ~60 ppb). As an example, consider a sample site in the intermountain western
819 U.S. region with very high modeled estimates of U.S background (e.g., seasonal mean USB of 45 ppb
820 with some days as high as 65 ppb). Even at this high background location, the calculated annual W126
821 values in the USB scenario are quite low, on the order of 3 ppm-hrs. Most sites in the domain where
822 background levels are lower than the location cited above will have even smaller background W126
823 estimates, on the order of 1 ppm-hrs, which is consistent with values mentioned in past reviews (USEPA,
824 2007). Using the counterfactual scenarios, background ozone has a relatively small impact on W126
825 levels across the U.S.
826 However, because of the non-linear weighting function used in the W126 calculation, the sum of
827 the W126 from the USB scenario and the W126 resulting from US anthropogenic sources will not equal
828 the total W126. In most cases, the sum of those two components will be substantially less than total
829 W126. As a result, EPA believes it is more informative to estimate the fractional contribution of
830 background ozone to W126 levels. The 5-step methodology for assessing the fractional influence of
2A-38
-------
831 background ozone to annual W126 levels in the four locations is described below. The fractional
832 influence methodology essentially places higher weights on background fractions on days that are going
833 to contribute most substantially to the yearly W126 value.
834 • Step la: Calculate the IV1DA8 ozone values from the base and the three zero out
835 modeling scenarios at each grid cell containing a site in an area.
836 • Step Ib: Calculate the W126 daily index for the base model scenario.
837 • Step 2: For each site, find the three months with highest summed W126 daily indices.
838 • Step 3: Normalize the daily IV1DA8 values in the base, NB, NAB, and USB scenarios by the
839 corresponding W126 daily index from the base scenario.
840 • Step 4: Calculate the average W126-weighted IV1DA8 values over the three month
841 period for each of the four scenarios (base, NB, NAB, USB).
842 • Step 5: Calculate the NB/Base, NAB/Base, and USB/Base ratios based on step 4 outputs.
843 These values represent an estimate of the fractional influence of background ozone on
844 modeled W126 levels.
845 Figure 7a shows the estimated fractional influence of the three background definitions on W126
846 levels in Atlanta, Denver, Farmington, and Riverside. Based on this limited assessment, natural
847 background sources are estimated to contribute 29-50% of the total modeled W126 with the highest
848 relative influence in the intermountain western U.S. (e.g., Farmington NM) and the lowest relative
849 influence in the eastern U.S. (e.g., Atlanta). U.S. background is estimated to contribute 37-65% of the
850 total modeled W126. Figure 7b compares the relative influence of background on W126 versus seasonal
851 mean IV1DA8 ozone. The proportional impacts of background are slightly less for the W126 metric than
852 for seasonal mean IV1DA8 (discussed in section 2.4.2), because of the weighting function that places
853 more emphasis on higher ozone days when background fractions are generally lower.
854 There are several caveats associated with this analysis. First, only the zero out modeling was
855 used to assess the fractional influence of background sources on W126. The source apportionment
856 approach estimated slightly smaller relative contributions for seasonal mean IV1DA8 levels, so from that
857 perspective the zero out estimates could represent the high end of background influence on W126.
858 Additionally, the methodology used for this analysis relies on daily IV1DA8 values as a surrogate (the data
859 were readily available) for the 8a-8p time period relevant to the W126 metric. The key conclusion from
860 this cursory analysis is that background ozone may comprise a non-negligible portion of current W126
861 levels across the U.S. This fractional influence is greatest in the intermountain western U.S. and are
862 slightly smaller than the seasonal mean IV1DA8 metric. In the counterfactual cases, when non
863 background sources are completely removed, the remaining W126 levels are low (< 3 ppm-hrs).
2A-39
-------
864
865
866
867
868
869
100%
Farmington
Denver
Riverside
Atlanta
Figure 7a. Fractional contribution of background sources to W126 levels in four sample locations.
Model estimates based on 2007 CMAQ zero out modeling.
100%
MDA8
IW126
Farmington
Denver
Riverside
Atlanta
Figure 7b. Fractional contribution of U.S. background to seasonal mean MDA8 ozone and W126 levels
in four sample locations. Model estimates based on 2007 CMAQ zero out modeling.
2A-40
-------
870 8. Summary
871 The precise definition of background ozone can vary depending upon context, but it generally
872 refers to ozone that is formed by sources or processes that cannot be influenced by local control
873 measures. Background ozone can originate from natural sources of ozone and ozone precursors, as well
874 as from upwind manmade emissions of ozone precursors. In order to help further characterize
875 background ozone levels over the U.S., EPA has completed additional air quality modeling analyses
876 subsequent to the lst-draft policy assessment. As shown above, the results are largely consistent with
877 previous determinations about the magnitude of background ozone contributions across the U.S.
878 For a variety of reasons, it is challenging to present a comprehensive summary of all the
879 components and implications of background ozone. In many forums the term "background" is used
880 generically and the lack of specificity can lead to confusion as to what sources are being considered.
881 Additionally, it is well established that the impacts of background sources can vary greatly over space
882 and time which makes it difficult to present a simple summary of background ozone levels. Further,
883 background ozone can be generated by a variety of processes, each of which can lead to differential
884 patterns in space and time, and which often have different regulatory ramifications. Finally, background
885 ozone is difficult to measure and thus, typically requires air quality modeling which has inherent
886 uncertainties and potential errors and biases. Even with all of these complexities in mind, EPA believes
887 the following concise and step-wise summary of background ozone is appropriate as based on previous
888 modeling exercises and the more recent EPA analyses summarized herein.
889 • The most fundamental definition of background is "natural background" (NB). NB ozone is that
890 which is produced by processes other than manmade emissions. Examples of sources of natural
891 background include: stratospheric ozone intrusions, wildfire emissions, and biogenic emissions
892 from vegetation and soils. To date, NB ozone has been estimated to be that ozone that would
893 exist in the absence of anthropogenic ozone precursor emissions worldwide. Modeling analyses
894 have shown that NB levels can vary in time and space. As shown in Section 3, April-October
895 average NB levels range from approximately 15-35 ppb with the highest values in the spring and
896 at higher-elevation sites.
897
898 • More expansive definitions of background include North American background (NAB) and U.S.
899 background (USB). These definitions represent the ozone that originates from sources and
900 processes other than North American or U.S. anthropogenic sources. Sources of NAB and USB
901 include all the same sources of natural background, plus manmade ozone precursors emitted
902 outside the North America or the U.S. Modeling analyses have shown that NAB and USB
903 background levels can vary in time and space. As discussed in Section 3, seasonal mean NAB
904 and USB background levels range from approximately 25-45 ppb with the highest values in the
905 spring and at higher-elevation sites. USB levels are slightly higher than NAB, usually by less than
906 2 ppb.
907
2A-41
-------
908 • Estimates of seasonal mean background ozone levels are valuable in terms of a first-order
909 characterization, however because levels can vary significantly from day-to-day, it is also
910 instructive to consider the distribution of daily model estimates of background ozone over a
911 season. Typically, model background is slightly higher in the April-June period than in the later
912 portion of the ozone season (July-October) (EPA, 2012). More importantly, the modeling shows
913 that the days with highest ozone levels, on average, have similar background levels to days with
914 lower values. As a result, the proportion of total ozone that has background origins is smaller
915 on high ozone days (e.g., days > 70 ppb) than the more common lower ozone days that drive
916 seasonal means. Section 4 provides information about the distribution of background ozone
917 fractions. Based on the source apportionment modeling, it is shown that U.S. anthropogenic
918 emissions typically comprise the majority of the total ozone on site-days with base modeled
919 ozone MDAS values greater than 60 ppb.
920
921 • While it is important to recognize that most high ozone days (i.e., potential exceedance days)
922 are estimated to be driven predominantly by non-background emissions, the recent EPA
923 modeling also shows times and locations in which background contributions are estimated to
924 approach 60-80 ppb. As described in Sections 4 and 6 of this document, these occurrences are
925 relatively infrequent. While the modeling was not expressly developed to capture these types
926 of events, ambient observations have also shown relatively rare events where background
927 ozone sources (wildfires, stratospheric intrusions) have overwhelmingly contributed to an
928 ozone exceedance. From a policy perspective, these background events must be viewed in the
929 context of their relative infrequency and the existing mechanisms within the Clean Air Act (e.g.,
930 exceptional event policy, 179B international determinations) that help ensure States are not
931 required to control for events that are inherently outside their ability to influence. While
932 background ozone levels can approach and periodically exceed the NAAQS at some locations,
933 these conditions are not a constraining factor in the selection of a NAAQS. The Clean Air Act
934 requires the NAAQS to be set at a level requisite to protect public health and welfare. Case law
935 makes it clear that attainability and technical feasibility are not relevant considerations. In
936 previous reviews, EPA assessed the proximity of potential levels to peak background levels as a
937 secondary consideration between levels where health and welfare was protected.
938
939 • Section 5 shows that the contributions to background are multi-dimensional. Daily peak 8-hour
940 ozone values over the U.S. are a function of local and regional anthropogenic emissions,
941 anthropogenic emissions from outside the U.S. (including shipping emissions), natural and
942 anthropogenic methane emissions, wildfire emissions, and purely natural sources. While local
943 and regional controls are still considered to be the most effective at reducing local ozone levels,
944 any measures to reduce the international contributions or methane-induced background will
945 also be valuable.
946
947 • In previous ozone NAAQS reviews, EPA estimated risk from exposure only to ozone
948 concentrations above background. In the first drafts of the REA and PA for the current ozone
2A-42
-------
949 review, EPA estimated risk from exposure to total measured ozone concentrations, which
950 include those concentrations from background sources. EPA will continue to provide estimates
951 of risk from exposure to total ozone, consistent with CASAC advice, in the second draft policy
952 assessment. The recent EPA modeling was completed to assist in determining, in a limited
953 sense, the risk attributable to background ozone. The fractional values of background
954 contributions in the 12 REA study areas (43-66 percent) could be used as first order
955 approximations of the risk due to ozone background.
956
957 9. References
958 Arunachalam, S. (2009). Peer Review of Source Apportionment Tools in CAMx and CMAQ. Prepared
959 under EPA Contract #EP-D-07-102, Institute of the Environment, University of North Carolina,
960 Chapel Hill, NC, 42pp,
961 http://www.epa.gov/scram001/reports/SourceApportionmentPeerReview.pdf.
962 Emery, C; Jung, J; Downey, N; Johnson, J; Jimenez, M; Yarwood, G; Morris, R. (2012). Regional and global
963 modeling estimates of policy relevant background ozone over the United States. Atmospheric
964 Environment, 47: 206-217. http://dx.doi.Org/10.1016/i.atmosenv.2011.ll.012.
965 Environ (2011). User's Guide: Comprehensive Air Quality Model with Extensions, Version 5.40; Novato
966 CA, 306pp. http://www.camx.com/files/camxusersguide v5-40.aspx.
967 Fiore, A; Jacob, DJ; Liu, H; Yantosca, RM; Fairlie, TD; Li, Q. (2003). Variability in surface ozone background
968 over the United States: Implications for air quality policy. J Geophys Res, 108: 4787.
969 http://dx.doi.org/10.1029/2003JD003855.
970 Henderson, BH; Possiel N; Akhtar F; Simon H. (2012). Memo to the Ozone NAAQS Review Docket EPA-
971 HQ-OAR-2012-0699: Regional and Seasonal Analysis of North American Background Ozone
972 Estimates from Two Studies.
973 http://www.epa.gOV/ttn/naaqs/standards/ozone/s o3 2008 td.html.
974 Lefohn, A. S.; Laurence, J. A.; Kohut, R. J. (1988). A comparison of indices that describe the relationship
975 between exposure to ozone and reduction in the yield of agricultural crops. Atmos. Environ. 22:
976 1229-1240.
977 Simon, H., Baker, K.P., Phillips, S. (2012) Compilation and interpretation of photochemical model
978 performance statistics published between 2006 and 2012. Atmospheric Environment, 61, 124-
979 139.
980 U.S. Environmental Protection Agency (2005). Technical Support Document for the Final Clean Air
981 Interstate Rule Air Quality Modeling. Office of Air Quality Planning and Standards, Research
982 Triangle Park, NC, 285pp. http://www.epa.gov/cair/technical.html.
983 U.S. Environmental Protection Agency. (2007). Review of the National Ambient Air Quality Standards for
984 Ozone: Policy Assessment of Scientific and Technical Information - OAQPS Staff Paper, U.S.
985 Environmental Protection Agency, Research Triangle Park, NC. EPA 452/R-07-007.
2A-43
-------
986 U.S. Environmental Protection Agency (2011). Air Quality Modeling Final Rule Technical Support
987 Document. Office of Air Quality Planning and Standards, Research Triangle Park, NC, 363pp.
988 http://www.epa.gov/airtransport/CSAPR/techinfo.html.
989 U.S. Environmental Protection Agency (2012). Policy Assessment for the Review of the Ozone National
990 Ambient Air Quality Standards: First External Review Draft. EPA-452/P-12-002, 297pp.
991 U.S. Environmental Protection Agency. (2013). Integrated Science Assessment for Ozone and Related
992 Photochemical Oxidants, U.S. Environmental Protection Agency, Research Triangle Park, NC.
993 EPA/600/R-10/076.
994 U.S. Environmental Protection Agency. (2014). Health Risk and Exposure Assessment for Ozone, Second
995 External Review Draft, U.S. Environmental Protection Agency, Research Triangle Park, NC. EPA
996 xxx/P-xx-xxx.
997 Zhang, L; Jacob, DJ; Downey, NV; Wood, DA; Blewitt, D; Carouge, CC; Van donkelaar, A; Jones, DBA;
998 Murray, LT; Wang, Y. (2011). Improved estimate of the policy-relevant background ozone in the
999 United States using the GEOS-Chem global model with 1/2 2/3 horizontal resolution over North
1000 America. Atmospheric Environment 45: 6769-6776.
1001 http://dx.doi.0rg/10.1016/i.atmosenv.2011.07.054.
2A-44
-------
i APPENDIX 2B
2 MONITORING DATA ANALYSIS OF RELATIONSHIPS
3 BETWEEN CURRENT STANDARD AND W126 METRIC
4 Presented here are monitoring data analyses evaluating relationship between ozone (63)
5 concentrations in the averaging time and form of the current secondary standard (3-year average
6 of the annual 4th highest daily maximum 8-hour concentrations, in parts per billion), and a three-
7 year W126 metric (3-year average of the annual maximum 3-month sum of weighted daytime
8 concentrations, in parts per million-hours). We also consider the responsiveness of these two
9 metrics to historical changes in air quality related to ozone precursor emissions.
10 For this analysis, we chose to examine monitoring data from a base period (2001-2003)
11 as well as a recent period (2009-2011). The base period was chosen to represent air quality
12 conditions before the implementation of the 1997 national ambient air quality standard
13 (NAAQS) for O3 (0.08 ppm). In 2004, EPA designated 113 areas as nonattainment for the 1997
14 standard, which required many areas to begin precursor emissions control programs for the first
15 time. At about the same time, EPA began implementation of the NOx Budget Trading Program
16 under the NOx State Implementation Plan, also known as the "NOx SIP Call1," which required
17 summertime reductions in NOx emissions from power plants and other large sources throughout
18 the Eastern U.S. These programs were successful in reducing peak O3 concentrations, especially
19 in the Eastern U.S., and as a result only 8 of the original 113 nonattainment areas were still
20 violating the 1997 O3 NAAQS during the 2009-2011 period.
21 Hourly O3 concentration data were retrieved from EPA's Air Quality System (AQS)
22 database2 for both periods, and used to calculate design values for the current standard as well as
23 3-year average W126 values for both periods. The procedures for calculating design values for
24 the current standard from hourly O3 concentration data are described in 40 CFR Part 50,
25 Appendix P, and the procedures for calculating the 3-year average W126 values are described in
26 section 4.3.1. of the 2nd draft Welfare Risk and Exposure Assessment (WREA). There were 838
27 monitoring sites with sufficient data to calculate these values for both periods. In order to
28 identify regional patterns in the relationships, these sites were grouped into the nine NOAA
1 http://www.epa.gov/airmarkets/progsregs/nox/sip.html
2 EPA's Air Quality System (AQS) database is a national repository for many types of air quality and
related monitoring data. AQS contains monitoring data for the six criteria pollutants dating back to the 1970's, as
well as more recent additions such as PM2.5 speciation, air toxics, and meteorology data. At present, AQS receives
hourly O3 monitoring data collected from nearly 1,400 monitors operated by over 100 state, local, and tribal air
quality monitoring agencies.
2B-1
-------
29 climate regions (Karl and Koss, 1984) used in the WREA. Figure 2B-1 presents a map of these
30 regions, which are color-coded to match the scatter plots in the subsequent figures.
31 Figures 2B-2a, 2B-2b, 2B-3a and 2B-3b show scatter plots of the design values for the
32 current standard (x-axis) versus 3-year average W126 values (y-axis) for the base period and
33 recent period, respectively. Most monitors in the U.S. both exceeded the current standard of 75
34 ppb and a three-year average W126 value of 15 ppm-hrs during the base period. During the
35 recent period, both the design values and 3-year average W126 values were much lower, and
36 there also appears to be less scatter between the two metrics. In both periods, the highest design
37 values and W126 values occurred in the West region which includes California. Finally, it is
38 worth noting that monitors in the Southwest and West regions tend to have higher W126 values
39 relative to their design values than in other regions.
40 Figure 2B-4 shows a scatter plot of the design values for the current standard for the base
41 period (x-axis) versus for the recent period (y-axis), while Figure 2B-5 shows this same
42 relationship based on the 3-year average W126 values. The relationship between the two periods
43 appears to be fairly linear for both metrics, indicating that larger decreases in these metrics
44 tended to occur at monitors with higher base values. Figures 2B-6 and 2B-7 show design values
45 for the current standard and 3-year average W126 values, respectively, compared to the unit
46 changes in those values between the base period and recent period. Figures 2B-6 and 2B-7 show
47 the difference between each point and the one-to-one lines in Figures 2B-4 and 2B-5,
48 respectively. In particular, these figures highlight that there were some monitors where design
49 values for the current standard and/or W126 values increased. However, those monitors also
50 tended to have lower base values, and were mostly located outside of areas subject to emissions
51 controls under the 1997 standard.
52 Finally, Figure 2B-8 compares the unit change in design values (in ppb; x-axis) to the
53 unit change in 3-year average W126 values (in ppm-hrs; y-axis). This figure shows that in most
54 locations, the current standard metric and the W126 metric exhibit similar responses to changes
55 in precursor emissions. In particular, the NOx SIP Call, which was implemented in the states
56 east of the Mississippi River, was effective at reducing both design values and W126 values at
57 nearly all monitors in the Eastern U.S. The relationship was much more variable in the
58 remaining regions, where emissions control programs were mostly local and limited to areas
59 which were violating the NAAQS.
60 Based on this analysis of ambient monitoring data, we can make the following general
61 conclusions about the relationship between the design value metric for the current O^ standard
62 and the 3-year average W126 metric:
63 1. There is a fairly strong, positive degree of correlation between the two metrics.
2B-2
-------
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
2. Monitors in the West and Southwest regions tend to have higher W126 values relative to
their design values than in other regions.
3. Reducing precursor emissions, especially NOx, is an effective strategy for lowering both
design values and W126 values. In particular, regional control programs such as the NOx
SIP call are effective at reducing both metrics over a broad area.
In addition, Figure 2B-9 examines the number of counties with 8-hour design values
meeting the current standard and 3-year average W126 index values greater than 15 ppm-hrs.
Most of these counties were located in the Southwest region of the country. There were no
counties in any of the studied 3-year periods that had design values less than or equal to 65 ppb
and 3-year average W126 index values greater than 15 ppm-hrs.
Central
EastlMorthCentral
NorthEast
Northwest
South
SouthEast
Southwest
West
WestNorthCentral
Figure 2B-l.Map of the 9 NOAA climate regions (Karl and Koss, 1984), color coded to
match the subsequent scatter plots.
2B-2
-------
CD _
CD
CD
CD
un
CD
CD _
CD
iTN
CD _
CD
Central
* EastNorthCentral
• NorthEast
* NorthWest
South
SouthEast
SouthWest
* West
* We stNorth Central
*
•
•
. W*$
• *»*
• * •*
i
79 40 60
80 Figure 2B-2a. Design values for
*
• *
*
• *
w
•
• * *
*
**
•
* •**
* •
* * * * *
;..—.. >; • .
* * * *
-------
in
CN
CO
UM
CN~
cn
T~
f— _
-~
cn —
r
ITJ —
CO —
* Central
EastNorthCentral
NorthEast
* Northwest
1 South
* SouthEast
* Southwest
* West
* We stNorth Central
* •
*
, 1 ; - ! ,
i
60 6
|
*
I
•
• •
• •
*
I * *
* 1
t •
5 7
• *
0 »
* * * '
* t
•
* t !
. .
1 ;
! : « !
! • !
1 • •
] 7
i
i
5
84
85
86
87
88
89
90
Figure 2B-2b. Design values for the current Os standard in ppb (x-axis) versus 3-year
average W126 values in ppm-hrs (y-axis) based on ambient monitoring data
for 2001-2003 with a focus on monitors with 2001-2003 design values below 75
ppb.
2B-5
-------
CD
CD
CD
Lfi
CD
TN
Central
EastNorthCentral
NorthEast
NorthWest
South
SouthEast
SouthWest
West
WestNorthCentral
*
.*.
*
*
91
40
I
60
I
80
100
120
92
93
94
95
Figure 2B-3a. Design values for the current Os standard in ppb (x-axis) versus 3-year
average W126 values in ppm-hrs (y-axis) based on ambient monitoring data
for 2009-2011.
2B-6
-------
CO
en —
i-
in —
* Central
EastNorthCentral
NorthEast
* Northwest
1 South
* SouthEast
* Southwest
* West
* We stNorth Centra I
•
'
• .
w
*
-•-
M
i ; :
* 1 i I
'
-
•
•
.
1
60
65
70
75
96
97
98
99
100
101
102
Figure 2B-3b. Design values for the current Os standard in ppb (x-axis) versus 3-year
average W126 values in ppm-hrs (y-axis) based on ambient monitoring data
for 2009-2011 with a focus on monitors with 2009-2011 design values below 75
ppb.
2B-7
-------
CD
C--I
CD
CD
CD
CO
CD
CD
1 Central
* EastNorthCentral
NorthEast
* NorthWest
• South
SouthEast
SouthWest
* West
* WestNorthCentral
103
40
I
60
I
80
100
120
104 Figure 2B-4.Design values for the current Os standard in ppb based on ambient
105 monitoring data for 2001-2003 (x-axis) versus 2009-2011 (y-axis).
2B-8
-------
CD
CD
CD
Lfi
CD
TN
Central
* EastNorthCentral
NorthEast
* NorthWest
South
SouthEast
SouthWest
* West
* WestNorthCentral
* * * *
106
0
1
10
20
30
40
50
!
60
70
107
108
109
Figure 2B-5. Three-year average W126 values in ppm-hrs based on ambient monitoring
data for 2001-2003 (x-axis) versus 2009-2011 (y-axis).
2B-9
-------
110
111
112
113
114
CD
CN
Central
• EastNorthCentral
NorthEast
* NorthWest
South
SouthEast
SouthWest
* West
* WestNorthCentral
40
60
I
80
100
120
Figure 2B-6. Design values for the current Os standard in ppb based on ambient
monitoring data for 2001-2003 (x-axis) versus unit (ppb) change in design
values from 2001-2003 to 2009-2011 (y-axis).
2B-10
-------
» *
* *
Central
* EastNorthCentral
* NorthEast
* NorthWest
South
SouthEast
SouthWest
* West
* WestNorthCentral
115
116
117
118
119
0
I
10
20
30
40
50
60
I
70
Figure 2B-7. Three-year average W126 values in ppm-hrs based on ambient monitoring
data for 2001-2003 (x-axis) versus unit (ppm-hr) change in 3-year average
W126 values from 2001-2003 to 2009-2011 (y-axis).
2B-11
-------
120
121
122
123
124
125
Central
• EastNorthCentral
: NorthEast
* NorthWest
•• South
SouthEast
SouthWest
* West
* WestNorthCentral
-20
-10
0
10
Figure 2B-8. Unit (ppb) change in design values for the current Os standard from 2001-
2003 to 2009-2011 (x-axis) versus unit (ppm-hr) change in 3-year average
W126 values from 2001-2003 to 2009-2011 (y-axis).
2B-12
-------
126
75-
m
-60-
t=
(§45-
15-
j-
Regions
IQhioValley
Rockies
Southeast
South
Southwest
West
12-
^ 9-
13
O
O
•5 B-
E ,_
0-
-
-
-
•
*.
-•
-
-
-
•
• i
o
127
128
129
130
131
132
133
134
Years
Regions
• Rockies
Southwes"
Years
Figure 2B-9. Number of counties where the 8-hour design value is meeting the current standard and 3-year average W126
index value is greater than 15 ppm-hrs (left), and number of counties where the 8-hour design value is less than
or equal to 70 ppb and 3-year average W126 index value is greater thanlS ppm-hrs (right)3.
REFERENCES
Karl, T.R. and Koss, W.J., 1984: "Regional and National Monthly, Seasonal, and Annual Temperature Weighted by Area, 1895-1983." Historical Climatology
Series 4-3, National Climatic Data Center, Asheville, NC, 38 pp.
No counties in any of the studied 3-year periods were at or below a 3-year average of 4th highest daily maximum 8-hour averages of 65 ppb and also
2B-13 Draft - Do Not Quote or Cite
above a 3-year W126 index value of 15 ppm-hrs
January 2014
-------
1 APPENDIX 2C
2 INTER-ANNUAL VARIABILITY IN W126 INDEX VALUES:
3 COMPARING ANNUAL AND 3-YEAR AVERAGE METRICS (2008-2010)
4 2C.1 OVERVIEW
5 This appendix describes an analysis comparing values for a single-year or annual W126
6 metric to a W126 metric averaged over three consecutive years. The purpose of this analysis is
7 to compare values based on a 3-year average of annual W126 indices to values based on a single
8 annual W126 index. The deviations of the annual W126 index values in 2008, 2009, and 2010
9 from the 2008-2010 average W126 index values are presented.
10 2C.2 GENERAL DATA PROCESSING
11 The air quality data for this analysis originated from EPA's Air Quality System (AQS)
12 data base, the official repository of ambient air measurements. The data used in this analysis
13 consisted of W126 index values calculated from hourly ozone concentrations measured at 1082
14 ozone monitors nationwide. Ozone monitors must have submitted data to AQS for at least 75%
15 days in their required ozone monitoring season in 2008, 2009, and 2010 to be included in the
16 analysis.
17 2C.3 RESULTS & CONCLUSION
18 The figure below shows a scatter plot of the deviations in the annual W126 index from
19 the 3-year average by monitor. The solid curves represent the average deviation in a moving
20 window along the x-axis for each year. From this figure, it is apparent that the highest annual
21 W126 index value occurred in 2008 for most monitoring locations, the lowest annual W126
22 index value occurred in 2009 for most monitoring locations, and the 2010 W126 index value was
23 generally somewhere in between. It is also apparent that the inter-annual variability in the W126
24 index increases along with the 3-year average. For monitors with 3-year average W126 values
25 near 15 ppm-hrs, the average deviation was +3.5 ppm-hrs in 2008 and -3.8 ppm-hrs in 2009.
26 This represents a 1-year swing of-7.3 ppm-hrs.
27 The model-based air quality adjustments in the 2n draft of the Os Welfare REA show
28 that reducing NOx emissions is effective for reducing 3-year average W126 levels. In Appendix
29 2B, the analyses based on ambient monitoring data also show that large-scale reductions in NOx
30 emissions are associated with lower W126 levels. Finally, the data analysis presented in this
31 appendix shows that the inter-annual variability in the annual W126 index tends to decrease with
32 decreasing W126 levels. Thus, it is expected that reductions in NOx emissions will not only
2C-1
-------
33 result in lower 3-year average W126 levels, but also result in less inter-annual variability
34 associated with annual W126 levels.
35 The W126 index is based on a logistic weighting function that increases the weights
36 assigned to hourly ozone concentrations very rapidly. Hourly ozone concentrations of 50 parts
37 per billion are given a weight of about 10% while concentrations of 80 parts per billion are given
38 a weight of nearly 90%. The annual W126 index is calculated as a 3-month sum of weighted
39 ozone concentrations during daylight hours, which amounts to a sum of roughly 1100 weighted
40 hourly concentrations. Thus, even a modest change in the average daily ozone level may have a
41 significant impact upon the annual W126 index. Since ozone formation is heavily influenced by
42 meteorology, the inter-annual variability in meteorological conditions tends to cause a large
43 inter-annual variability in the W126 index.
44 In conclusion, this evaluation indicates the extent to which a form for the secondary
45 ozone standard that averages the annual W126 index values over three consecutive years might
46 be expected to account for the annual variability in this index since the 3-year period would be
47 expected to include year(s) below as well as above the 3-year average.
48
2C-2
-------
Annual W126 Index vs. 3-Year Average W126 Index
49
50
51
52
«J
I
Q
o 2008
n 2009
A 2010
5 10 20
3-Year Average W126 Index, 2003-2010
50
Figure 2C-1. Deviation of the annual W126 index values in 2008, 2009, and 2010 (y-axis)
from the 3-year average W126 index value (x-axis).
2C-3
-------
1 APPENDIX 3A
2 MODES OF ACTION SUMMARY
3 The initial key event in the toxicity pathway of O^ is the formation of secondary
4 oxidation products in the respiratory tract (ISA, section 5-3, U.S. EPA, 2013). This mainly
5 involves direct reactions with components of the extracellular lining fluid (ELF)1. The resulting
6 secondary oxidation products transmit signals to the epithelium, pain receptive nerve fibers and,
7 if present, immune cells (i.e., eosiniphils, dendritic cells and mast cells). Thus, the effects of 63
8 are mediated by components of ELF and by the multiple cell types found in the respiratory tract.
9 Further, oxidative stress2 is an implicit part of this initial key event.
10 Another key event in the toxicity pathway of Os is the activation of neural reflexes which
11 lead to lung function decrements. Evidence is accumulating that secondary oxidation products
12 are responsible for this effect. Different receptors on bronchial sensory nerves (i.e., C-fibers)
13 have been shown to mediate separate effects of 63 on pulmonary function. For example, pain
14 (i.e., nociceptive) sensory nerves are involved in the involuntary truncation of inspiration which
15 results in decreases in FVC, FEVi, tidal volume and pain upon deep inspiration. Ozone exposure
16 also results in activation of vagal sensory nerves and a mild increase in airway obstruction
17 measured as increased sRaw. Activation of neural reflexes also results in extrapulmonary effects
18 such as slow resting heart rate (i.e., bradycardia).
19 Initiation of inflammation is also a key event in the toxicity pathway of O^. Secondary
20 oxidation products, as well as cell signaling molecules (i.e., chemokines and cytokines) from
21 airway epithelial cells and white blood cells (i.e., macrophages), have been implicated in the
22 initiation of inflammation. Airways neutrophilia has been demonstrated in bronchoalveolar
23 lavage fluid (BALF), mucosal biopsy and induced sputum samples. Influx of other cells (i.e.
24 mast cells, monocytes and macrophages) also occur. Inflammation further contributes to
25 Os-mediated oxidative stress. It should be noted that inflammation, as measured by airways
1 The ELF is a complex mixture of lipids (fats), proteins, and antioxidants that serve as the first barrier and
target for inhaled O3. The antioxidant substances present in the ELF appear in most cases to limit interaction of O3
with underlying tissues and to prevent penetration of O3 deeper into the lung. However, as the ELF thickness
decreases and becomes ultra thin in the alveolar region, it may be possible for direct interaction of O3 with the
underlying epithelial cells to occur. The formation of secondary oxidation products is likely related to the
concentration of antioxidants present and the quenching ability of the ELF.
Oxidative stress reflects an imbalance between the systemic manifestation of reactive oxygen species,
such as superoxide, and a biological system's ability to readily detoxify the reactive intermediates or to repair the
resulting damage.
3A-1
-------
1 neutrophilia, is not correlated with decrements in pulmonary function as measured by
2 spirometry.
3 A fourth key event in the toxicity pathway of 63 is alteration of epithelial barrier
4 function. Increased permeability3 occurs as a result of damage to tight junctions between
5 epithelial cells subsequent to Os-induced injury and inflammation. It may play a role in allergic
6 sensitization and in airway hyperresponsiveness (AHR). Genetic susceptibility has been
7 associated with this pathway.
8 A fifth key event in the toxicity pathway of Oi is the sensitization of bronchial smooth
9 muscle. Airway hyperresponsiveness (AHR), or increased bronchial reactivity, can be both a
10 rapidly occurring and a persistent response. The mechanisms responsible for early and later AHR
11 are not well-understood. Tachykinins, peptides that can excite neurons and cause smooth muscle
12 contraction, and the secondary oxidation products of Os have been proposed as mediators of the
13 early response and inflammation-derived products have been proposed as mediators of the later
14 response. Other chemical signaling molecules (i.e., cytokines and chemokines) have been
15 implicated in the AHR response to Os in animal models. Antioxidants may confer protection.
16 A sixth key event in the toxicity pathway of Os is the modification of innate/adaptive
17 immunity. While the majority of evidence for this key event comes from animal studies, there
18 are several studies suggesting that this pathway may also be relevant in humans. Ozone exposure
19 of human subj ects resulted in recruitment of activated innate immune cells to the airways.
20 Animal studies further linked Os-mediated activation of the innate immune system to the
21 development of nonspecific AHR, demonstrated an interaction between allergen and Os in the
22 induction of nonspecific AHR, and found that Os acted as an adjuvant for allergic sensitization
23 through the activation of both innate and adaptive immunity. These studies provide evidence that
24 Os can alter host immunologic response and lead to immune system dysfunction. These
25 mechanisms may underlie the exacerbation and induction of asthma, as well as decreases in lung
26 host defense.
27 Another key event in the toxicity pathway of Os is airways remodeling. Persistent
28 inflammation and injury, which are observed in animal models of chronic and intermittent
29 exposure to Os, are associated with morphologic changes such as mucous cell metaplasia4 of
3 Cells in epithelium are very densely packed together, leaving very little intercellular space. All epithelial
cells rest on a basement membrane, a thin sheet of fibers that acts as scaffolding on which epithelium can grow and
regenerate after injuries. Epithelial tissue is innervated but avascular; it must be nourished by substances diffusing
from the blood vessels in the underlying tissue. Injury to epithelial cells, such as caused by oxidative stress, can
cause the epithelium to become more permeable to substances in the underlying vasculature.
4 Metaplasia is the reversible replacement of one differentiated cell type with another mature differentiated
cell type. The change from one type of cell to another may generally be a part of normal maturation process or
3A-2
-------
1 nasal epithelium, bronchiolar metaplasia of alveolar ducts and fibrotic changes in small airways
2 (see Section 7.2.3 of the ISA, U.S. EPA 2013). Mechanisms responsible for these responses are
3 not well-understood. However, a recent study in mice demonstrated a key role for a signaling
4 pathway in the deposition of collagen in the airway wall following chronic intermittent exposure
5 to 63. Chronic intermittent exposure to 63 has also been shown to result in effects on the
6 developing lung and immune system.
7 Systemic inflammation and vascular oxidative/nitrosative stress are also key events in the
8 toxicity pathway of 03. Extrapulmonary effects of O^ occur in numerous organ systems,
9 including the cardiovascular, central nervous, reproductive, and hepatic systems (see
10 Sections 6.3 to 6.5 and Sections 7.3 to 7.5 of the ISA, U.S. EPA, 2013). It has been proposed that
11 lipid oxidation products resulting from reaction of 63 with lipids and/or cellular membranes in
12 the ELF are responsible for systemic responses; however, it is not known whether they gain
13 access to the circulation. Alternatively, release of diffusible mediators from the lung into the
14 circulation may initiate or propagate inflammatory responses in the circulation or other organ
15 systems.
16 Responses to 63 exposure are variable within the population. Studies have shown a large
17 range of pulmonary function responses to O?, among healthy young adults. Since individual
18 responses were relatively consistent across time in some of these studies, it was thought that
19 responsiveness reflected an intrinsic characteristic of the subject (Mudway and Kelly, 2000).
20 Other responses to 03 have also been characterized by a large degree of inter-individual
21 variability. For example, inter-individual variability in the neutrophilic response has been noted
22 in human subjects. Two studies demonstrated a 3- to 20-fold difference in airways neutrophilia,
23 under different exposure conditions (Schelegle et al., 1991 and Devlin et al., 1991, respectively).
24 Reproducibility of intraindividual responses to 63 exposures in human subjects, measured as
25 sputum neutrophilia, was demonstrated by Holz et al (1999). While the basis for the observed
26 inter-individual variability in responsiveness to 63 is not clear, both dosimetric and mechanistic
27 factors are likely to contribute. These are discussed in sections 5.4.1 and 5.4.2 in the ISA (U.S.
28 EPA, 2013). Section 5.4.2 of the ISA discusses studies that provide evidence for the mechanisms
29 that may underlie the variability in responsiveness seen among individuals. Certain functional
30 genetic polymorphisms, pre-existing conditions or diseases, nutritional status, lifestages, and co-
31 exposures contribute to altered risk of (Vinduced effects.
32
caused by some sort of abnormal stimulus. In simplistic terms, it is as if the original cells are not robust enough to
withstand the new environment, and so they change into another type more suited to the new environment. If the
stimulus that caused metaplasia is removed or ceases, tissues return to their normal pattern of differentiation.
3A-3
-------
1 3.1 REFERENCES
2 Devlin, RB; Mcdonnell, WF; Mann, R; Becker, S; House, DE; Schreinemachers, D; Koren, HS. (1991). Exposure of
3 humans to ambient levels of ozone for 6.6 hours causes cellular and biochemical changes in the lung. Am J
4 Respir Cell Mol Biol 4: 72-81.
5 Holz, O; Torres, RA; Timm, P; Mucke, M; Richter, K; Koschyk, S; Magnussen, H. (1999). Ozone-induced airway
6 inflammatory changes differ between individuals and are reproducible. Am J Respir Crit Care Med 159:
7 776-784.
8 Mudway, IS; Kelly, FJ. (2000). Ozone and the lung: A sensitive issue. Mol Aspects Med 21: 1-48.
9 Schelegle, ES; Siefkin, AD; McDonald, RJ. (1991). Time course of ozone-induced neutrophilia in normal humans.
10 Am J Respir Crit Care Med 143: 1353-1358.
11 U.S. EPA (2013). Integrated Science Assessment of Ozone and Related Photochemical Oxidants (Final Report).
12 U.S. Environmental Protection Agency, Washington, DC. EPA/600/R-10/076F. Available at:
13 http://www.epa.gov/ttn/naaqs/standards/ozone/s_o3_2008_isa.html
3A-4
-------
1 APPENDIX 3B
2 RECENT STUDIES OF RESPIRATORY-RELATED
3 EMERGENCY DEPARTMENT VISITS AND HOSPITAL
4 ADMISSIONS
5 Hospital Admissions for All Respiratory Causes
6 The APHENA study (APHENA is for Air Pollution and Health: A European and North
7 American Approach) analyzed air pollution and health outcome data from existing Canadian,
8 European, and U.S. multi-city studies and examined the influence of varying model specification
9 to control for season and weather (Katsouyanni et al., 2009). The U.S.-based portion of the
10 APHENA study utilized the National Morbidity, Mortality, and Air Pollution Study (NMMAPS)
11 cohort which, for the Katsouyanni et al. (2009) analysis, comprised respiratory hospital
12 admissions among individuals 65 years of age and older from 14 US cities with Oj, data from
13 1985-1994 (7 cities had summer only 63 data). For the year round analysis, Katsouyanni et al.
14 (2009) reported consistently positive, and statistically significant in models with 8 degrees of
15 freedom per year (U.S. EPA, 2013, section 6.2.7.2), associations between 1-hour 63
16 concentrations and respiratory hospital admissions across the datasets from the U.S., Canada, and
17 Europe (U.S. EPA 2013, Figure 6-15).5 In co-pollutant models adjusting for PMio, O3 effect
18 estimates remained positive, though effect estimates were somewhat attenuated in the U.S. and
19 European datasets, possibly due to the PM sampling schedule (U.S. EPA 2013, Figure 6-15).
20 Effect estimates for the warm season were larger than for the year-round analysis in the
21 Canadian dataset, but generally similar in magnitude to the year-round analysis in the U.S. and
22 European datasets.
23 Several additional multicity studies examined respiratory disease hospital admissions in
24 Canada and Europe. Cakmak et al. (2006) reported a statistically significant increase in
25 respiratory hospital admissions in 10 Canadian cities (4.4% increase per 20 ppb increase in 24-
26 hour average Os, 95% CI: 2.2, 6.5%). In analyses of potential effect modifiers of the Os-
27 respiratory hospital admission relationship, individuals with an education level less than the 9th
28 grade were found to be at greater risk. Dales et al. (2006) reported a 5.4% (95% CI: 2.9, 8.0%)
29 increase in neonatal respiratory hospital admissions for a 20 ppb increase in 24-hour average O^
The study by Katsouyanni et al. (2009) evaluated different statistical models. Although the investigators
did not identify the model they deemed to be the most appropriate for comparing the results across study locations,
they did specify that "overall effect estimates (i.e., estimates pooled over several cities) tended to stabilize at high
degrees of freedom" (Katsouyanni et al., 2009). In discussing of the results of this study, the ISA focused on models
with 8 degrees of freedom per year (US EPA, 2012a, section 6.2.7.2).
3B-1
-------
1 concentrations in 11 Canadian cities from 1986 to 2000. In contrast, Biggeri et al. (2005) did not
2 detect an association between short-term Oj, exposure and respiratory hospital admissions in four
3 Italian cities from 1990 to 1999.
4 In addition to the large multi-city studies discussed above, several smaller-scale studies
5 have also reported associations with total respiratory hospital admissions. Specifically, Lin et al.
6 (2008) reported a positive association between 63 and pediatric (i.e., <18 years) respiratory
7 admissions in an analysis of 11 geographic regions in New York state from 1991 to 2001, though
8 results were not presented quantitatively. In co-pollutant models with PMio, the authors reported
9 that region-specific 63 associations with respiratory hospital admissions remained relatively
10 robust.
11 Cause-Specific Hospital Admissions
12 With regard to cause-specific respiratory outcomes, the limited evidence available in the
13 last review indicated that the strongest findings were for ambient 63 associated asthma and
14 chronic obstructive pulmonary disease (COPD) respiratory hospital admissions (U.S. EPA 2013,
15 6.2.7.2). Since the last review, a few additional studies have investigated cause-specific
16 respiratory admissions (i.e., COPD, asthma, pneumonia) in relation to 63 exposure (Medina-
17 Ramon et al, 2006; Yang et al., 2005; Zanobetti and Schwartz, 2006; Silverman and Ito, 2010).
18 Medina-Ramon et al. (2006) examined the association between short-term ambient 03
19 concentrations and Medicare hospital admissions for COPD among individuals > 65 years of age
20 for COPD in 35 cities in the U.S. for the years 1986-1999. The authors reported an increase in
21 COPD admissions for lag 0-1 day in the warm season for a 30 ppb increase in 8-h max Os
22 concentrations. The authors found no evidence for such associations in cool season or in year
23 round analyses. In a co-pollutant model with PMio, the association between 03 and COPD
24 hospital admissions remained robust. In Vancouver from 1994-1998, a location with low ambient
25 O3 concentrations (U.S. EPA, 2013, Table 6-26), Yang et al. (2005) reported a statistically non-
26 significant increase in COPD admissions per 20 ppb increase in 24-hour average Oj
27 concentrations. In two-pollutant models with every-day data for NO2, SO2, CO, and PMio, Os
28 risk estimates remained robust, though not statistically significant (U.S. EPA, 2013, Figure 6-20;
29 Table 6-29). In addition, Wong et al. (2009) reported increased Os-associated COPD admissions
30 during periods of increased influenza activity in Hong Kong.
31 The ISA assessed a study that evaluated asthma-related hospital admissions in New York
32 City (U.S. EPA, 2013, section 6.2.7.2) (Silverman and Ito, 2010). This study examined the
33 association of 8-hour max Os concentrations with severe acute asthma admissions (i.e., those
34 admitted to the Intensive Care Unit [ICU]) during the warm season in the years 1999 through
3B-2
-------
1 2006 (Silverman and Ito, 2010)). The investigators reported positive associations between O^ and
2 ICU asthma admissions for the 6- to 18-year age group for a 30 ppb increase in max 8-hour
3 average Os concentrations, but little evidence of associations for the other age groups examined
4 (<6 years, 19-49, 50+, and all ages). However, positive associations were observed for each
5 age-stratified group and all ages for non-ICU asthma admissions, but again the strongest
6 association was reported for the 6- to 18-years age group. In two-pollutant models, Os effect
7 estimates for both non-ICU and ICU hospital admissions remained robust to adjustment for
8 PM2.s. In an additional analysis, using a smooth function, the authors examined whether the
9 shape of the concentration-response curve for O?, and asthma hospital admissions (i.e., both
10 general and ICU for all ages) is linear. When comparing the curve to a linear fit line, the authors
11 found that the linear fit was a reasonable approximation of the concentration-response
12 relationship between O?, and asthma hospital admissions, but the limited data density at relatively
13 low 63 concentrations contributes to uncertainty in the shape of the concentration-response
14 relationship at the low end of the distribution of Os concentrations (U.S. EPA, 2013, Figure 6-
15 16).
16 In contrast to COPD and asthma, the evidence for pneumonia-related admissions was less
17 consistent. Medina-Ramon et al. (2006) examined the association between short-term ambient Oi
18 concentrations and Medicare hospital admissions among individuals > 65 years of age for
19 pneumonia. The authors reported an increase in pneumonia hospital admissions in the warm
20 season for a 30 ppb increase in 8-hour max 63 concentrations, with no evidence of an association
21 in the cool season or year round. In two-pollutant models restricted to days for which PMio data
22 was available, the association between Oj, exposure and pneumonia hospital admissions
23 remained robust. In contrast, Zanobetti and Schwartz (2006) reported a decrease in pneumonia
24 admissions for a 20 ppb increase in 24-hour average O^ concentrations in Boston for the average
25 of lags 0 and 1 day.
26 The magnitude of associations with respiratory-related hospital admissions may be
27 underestimated due to behavioral modification in response to forecasted air quality (U.S. EPA,
28 2013, section 4.6.6). Recent studies (Neidell and Kinney, 2010; Neidell, 2009) conducted in
29 Southern California demonstrates that controlling for avoidance behavior increases 63 effect
30 estimates for respiratory hospital admissions, specifically for children and older adults. This
31 study shows that on days where no public alert warning of high 63 concentrations was issued,
32 there was an increase in asthma hospital admissions. Although only one study has examined
33 averting behavior and this study is limited to the outcome of asthma hospital admissions in one
34 location and time period (i.e., Los Angeles, CA for the years 1989-1997), it does provide
35 preliminary evidence indicating that some epidemiologic studies may underestimate associations
3B-3
-------
1 between 03 and health effects by not accounting for behavioral modification when public health
2 alerts are issued.
3 Emergency Department Visits for All Respiratory Causes
4 A large single-city study conducted in Atlanta by Tolbert et al. (2007), and subsequently
5 reanalyzed by Darrow et al. (2011) using different air quality data and evaluating associations
6 with different metrics, provides evidence for associations between short-term exposures to
7 ambient 63 concentrations and respiratory emergency department visits. Tolbert et al. (2007)
8 reported an increase in respiratory emergency department visits for a 30 ppb increase in 8-hour
9 max Oj, concentrations during the warm season. In copollutant models with CO, NC>2, and PMio,
10 limited to days in which data for all pollutants were available, associations between 63 and
11 respiratory emergency department visits remained positive, but were attenuated. Darrow et al.
12 (2011) reported the strongest associations with respiratory emergency department visits for 8-
13 hour daily max, 1-hour daily max, and day-time Os exposure metrics (all associations positive
14 and statistically significant), while positive, but statistically non-significant, associations were
15 reported with 24-hour average and commuting period exposure metrics. In addition, a negative
16 association was observed when using the night-time exposure metric (U.S. EPA, 2013, Figure 6-
17 17). The results of Darrow et al. (2011) suggest that averaging over nighttime hours may lead to
18 smaller 63 effect estimates for respiratory emergency department visits due to dilution of
19 relevant Os concentrations (i.e., the higher concentrations that occur during the daytime); and
20 potential negative confounding by other pollutants (e.g., CO, NO2) during the nighttime hours
21 (U.S. EPA, 2013, section 6.2.7.3)
22 Cause-Specific Emergency Department Visits
23 In evaluating asthma emergency department visits in an all-year analysis, a Canadian
24 multi-city study (Stieb et al., 2009) reported that 24-hour Os concentrations were positively
25 associated with emergency department visits for asthma at lag 1 and lag 2. Though the authors
26 did not present seasonal analyses, they stated that no associations were observed with emergency
27 department visits in the winter season, suggesting that the positive associations reported in the
28 all-year analysis were due to the warm season (Stieb et al., 2009). In addition to asthma, the
29 authors reported that Oj was positively associated with COPD emergency department visits in
30 all-year analyses, but that associations with COPD visits were statistically significant only for the
31 warm season (i.e., April-September).
3B-4
-------
1 Several single-city studies have also provided evidence for positive associations between
2 asthma emergency department visits and ambient O?, concentrations. Ito et al. (2007) reported
3 positive and statistically significant associations with asthma emergency department visits in
4 New York City during the warm season, and an inverse association in the cool season, for a 30
5 ppb increase in 8-hour max 63 concentrations. In two-pollutant models with PM2.5, NC>2, 862,
6 and CO, the authors found that 63 risk estimates were not substantially changed during the warm
7 season (U.S. EPA, 2013, Figure 6-20; Table 6-29).
8 Strickland et al. (2010) examined the association between 63 exposure and pediatric
9 asthma emergency department visits (ages 5-17 years) in Atlanta using air quality data over the
10 same years as Darrow et al. (2011) and Tolbert et al. (2007), but using population-weighting to
11 combine daily pollutant concentrations across monitors. Strickland et al. (2010) reported an
12 increase in emergency department visits for a 30 ppb increase in 8-hour max OT, concentrations
13 in an all-year analysis. In seasonal analyses, stronger associations were observed during the
14 warm season (i.e., May-October) than the cold season. In co-pollutant analyses that included CO,
15 NO2, PM2.5 elemental carbon, or PM2.5 sulfate, Strickland et al. (2010) reported that Os risk
16 estimates were not substantially changed. The authors also examined the concentration-response
17 relationship between 03 exposure and pediatric asthma emergency department visits and
18 reported that positive associations with Os persist at 8-hour ambient Os concentrations (3-day
19 average of 8-hour daily max concentrations) at least as low as 30 ppb.
20 In a single-city study conducted in Seattle, WA, Mar and Koenig (2009) examined the
21 association between 03 exposure and asthma emergency department visits for children (< 18)
22 and adults (> 18). For children, positive and statistically significant associations were reported
23 across multiple lags, with the strongest associations observed at lag 0 and lag 3. Ozone was also
24 found to be positively associated with asthma emergency department visits for adults at all lags,
25 except at lag 0. The slightly different lag times for children and adults suggest that children may
26 be more immediately responsive to O^ exposures than adults (Mar and Koenig, 2009).
27 In addition to the U.S. single-city studies discussed above, a single-city study conducted
28 in Alberta, Canada (Villeneuve et al., 2007) provides support for the findings from Stieb et al.
29 (2009), but also attempts to identify those lifestages at greatest risk for Os-associated asthma
30 emergency department visits. Villeneuve et al. reported an increase in asthma emergency
31 department visits in an all-year analysis across all ages with associations being stronger during
3B-5
-------
1 the wanner months. When stratified by age, the strongest associations were observed in the
2 warm season for individuals 5-14 and 15-44. These associations were not found to be
3 confounded by the inclusion of aeroallergens in age-specific models.
4 3.2 REFERENCES
5 Biggeri, A; Baccini, M; Bellini, P; Terracini, B. (2005). Meta-analysis of the Italian studies of short-term effects of
6 air pollution (MISA), 1990-1999. Int J Occup Environ Health 11: 107-122.
7 Cakmak, S; Dales, RE; Judek, S. (2006b). Respiratory health effects of air pollution gases: Modification by
8 education and income. Arch Environ Occup Health61: 5-10. http://dx.doi.org/10.3200/AEOH.61.L5-10
9 Dales, RE; Cakmak, S; Doiron, MS. (2006). Gaseous air pollutants and hospitalization for respiratory disease in the
10 neonatal period. Environ Health Perspect 114: 1751-1754. http://dx.doi.org/10.1289/ehp.9044
11 Darrow, LA; Klein, M; Sarnat, JA; Mulholland, JA; Strickland, MJ; Sarnat, SE; Russell, AG; Tolbert, PE. (2011).
12 The use of alternative pollutant metrics in time-series studies of ambient air pollution and respiratory
13 emergency department visits. J Expo Sci Environ Epidemiol 21: 10-19.
14 Jto, K; Thurston, GD; Silverman, RA. (2007b). Characterization of PM2.5, gaseous pollutants, and meteorological
15 interactions in the context of time-series health effects models. J Expo Sci Environ Epidemiol 17: S45-S60.
16 Katsouyanni, K; Samet, JM; Anderson, HR; Atkinson, R; Le Tertre, A; Medina, S; Samoli, E; Touloumi, G;
17 Burnett, RT; Krewski, D; Ramsay, T; Dominici, F; Peng, RD; Schwartz, J; Zanobetti, A. (2009). Air
18 pollution and health: A European and North American approach (APHENA). (Research Report 142).
19 Boston, MA: Health Effects Institute. http://pubs.healtheffects.org/view.php?id=327
20 Lin, S; Bell, EM; Liu, W; Walker, RJ; Kim, NK; Hwang, SA. (2008a). Ambient ozone concentration and hospital
21 admissions due to childhood respiratory diseases in New York State, 1991-2001. EnvironRes 108: 42-47.
22 http://dx.doi.0rg/10.1016/j.envres.2008.06.007
23 Mar, TF; Koenig, JQ. (2009). Relationship between visits to emergency departments for asthma and ozone exposure
24 in greater Seattle, Washington. Ann Allergy Asthma Immunol 103: 474-479.
25 Medina-Ramon, M; Zanobetti, A; Schwartz, J. (2006). The effect of ozone and PM10 on hospital admissions for
26 pneumonia and chronic obstructive pulmonary disease: A national multicity study. Am J Epidemiol 163:
27 579-588. http://dx.doi.org/10.1093/aje/kwj078
28 Neidell, M. (2009). Information, avoidance behavior, and health: The effect of ozone on asthma hospitalizations.
29 Journal of Human Resources 44: 450-478.
30 Neidell, M; Kinney, PL. (2010). Estimates of the association between ozone and asthma hospitalizations that
31 account for behavioral responses to air quality information. Environ Sci Pol 13: 97-103.
32 http://dx.doi.0rg/10.1016/j.envsci.2009.12.006
33 Silverman, RA; Ito, K. (2010). Age-related association of fine particles and ozone with severe acute asthma in New
34 York City. J Allergy Clin Immunol 125: 367-373. http://dx.doi.org/10.1016/jjaci.2009.10.061
3 5 Stieb, DM; Szyszkowicz, M; Rowe, BH; Leech, JA. (2009). Air pollution and emergency department visits for
36 cardiac and respiratory conditions: A multi-city time-series analysis. Environ Health Global Access Sci
37 Source 8: 25. http://dx.doi.org/10.1186/1476-069X-8-25
3B-6
-------
1 Strickland, MJ; Darrow, LA; Klein, M; Flanders, WD; Sarnat, JA; Waller, LA; Sarnat, SE; Mulholland, JA; Tolbert,
2 PE. (2010). Short-term associations between ambient air pollutants and pediatric asthma emergency
3 department visits. Am J Respir Crit Care Med 182: 307-316. http://dx.doi.org/10.1164/rccm.200908-
4 1201OC
5 Tolbert, PE; Klein, M; Peel, JL; Sarnat, SE; Sarnat, JA. (2007). Multipollutant modeling issues in a study of ambient
6 air quality and emergency department visits in Atlanta. J Expo Sci Environ Epidemiol 17: S29-S35.
7 http://dx.doi.org/10.1038/sj.jes.7500625
8 U.S. EPA (2013). Integrated Science Assessment of Ozone and Related Photochemical Oxidants (Final Report).
9 U.S. Environmental Protection Agency, Washington, DC. EPA/600/R-10/076F. Available at:
10 http://www.epa.gov/ttn/naaqs/standards/ozone/s_o3_2008_isa.html
11 Villeneuve, PJ; Chen, L; Rowe, BH; Coates, F. (2007). Outdoor air pollution and emergency department visits for
12 asthma among children and adults: A case-crossover study in northern Alberta, Canada. Environ Health
13 Global Access Sci Source 6: 40. http://dx.doi.org/10.1186/1476-069X-6-40
14 Wong, CM; Yang, L; Thach, TQ; Chau, PY; Chan, KP; Thomas, GN; Lam, TH; Wong, TW; Hedley, AJ; Peiris, JS.
15 (2009). Modification by influenza on health effects of air pollution in Hong Kong. Environ Health Perspect
16 117: 248-253. http://dx.doi.org/10.1289/ehp.11605
17 Yang, Q; Chen, Y; Krewski, D; Burnett, RT; Shi, Y; Mcgrail, KM. (2005). Effect of short-term exposure to low
18 levels of gaseous pollutants on chronic obstructive pulmonary disease hospitalizations. Environ Res 99: 99-
19 105. http://dx.doi.0rg/10.1016/j.envres.2004.09.014
20 Zanobetti, A; Schwartz, J. (2006). Air pollution and emergency admissions in Boston, MA. J Epidemiol Community
21 Health 60: 890-895. http://dx.doi.org/10.1136/iech.2005.039834
3B-7
-------
1 APPENDIX 3C
2 AT-RISK POPULATIONS
3 People with Specific Genetic Variants
4 Overall, for variants in multiple genes there is adequate evidence for involvement in
5 populations being more at-risk than others to the effects of Os exposure on health (U.S. EPA,
6 2013, section 8.1). Controlled human exposure and epidemiologic studies have reported evidence
7 of Os-related increases in respiratory symptoms or decreases in lung function with variants
8 including GSTM1, GSTP1, HMOX1, and NQO1. NQO1 deficient mice were found to be
9 resistant to (^-induced AHR and inflammation, providing biological plausibility for results of
10 studies in humans. Additionally, studies of rodents have identified a number of other genes that
11 may affect Os-related health outcomes, including genes related to innate immune signaling and
12 pro- and anti-inflammatory genes, which have not been investigated in human studies.
13 People with Asthma
14 Previous 63 AQCDs identified individuals with asthma as a population at increased risk
15 of Os-related health effects. Multiple new epidemiologic studies included in the ISA have
16 evaluated the potential for increased risk of Os-related health effects in people with asthma,
17 including: lung function; symptoms; medication use; airway hyperresponsiveness (AHR); and
18 airway inflammation (also measured as exhaled nitric oxide fraction, or FeNO). A study of
19 lifeguards in Texas reported decreased lung function with short-term 63 exposure among both
20 individuals with and without asthma, however, the decrease was greater among those with
21 asthma (Thaller et al., 2008). A Mexican study of children ages 6-14 detected an association
22 between short-term 63 exposure and wheeze, cough, and bronchodilator use among asthmatics
23 but not non-asthmatics, although this may have been the result of a small non-asthmatic
24 population (Escamilla-Nufiez et al., 2008). A study of modification by AHR (an obligate
25 condition among asthmatics) reported greater short-term (^-associated decreases in lung
26 function in elderly individuals with AHR, especially among those who were obese (Alexeeff et
27 al., 2007). With respect to airway inflammation, in one study, a positive association was reported
28 for airway inflammation among asthmatic children following short-term 63 exposure, but the
29 observed association was similar in magnitude to that of non-asthmatics (Barraza-Villarreal et
30 al., 2008). Similarly, another study of children in California reported an association between 63
31 concentration and FeNO that persisted both among children with and without asthma as well as
32 those with and without respiratory allergy (Berhane et al., 2011). Finally, Khatri et al. (2009)
33 found no association between short-term 63 exposure and altered lung function for either
3C-1
-------
1 asthmatic or non-asthmatic adults, but did note a decrease in lung function among individuals
2 with allergies.
3 New evidence for difference in effects among asthmatics has been observed in studies
4 that examined the association between O^ exposure and altered lung function by asthma
5 medication use. A study of children with asthma living in Detroit reported a greater association
6 between short-term Os and lung function for corticosteroid users compared with
7 noncorticosteroid users (Lewis et al., 2005). Conversely, another study found decreased lung
8 function among noncorticosteroid users compared to users, although in this study, a large
9 proportion of non-users were considered to be persistent asthmatics (Hernandez-Cadena et al.,
10 2009). Lung function was not related to short-term Oj, exposure among corticosteroid users and
11 non-users in a study taking place during the winter months in Canada (Liu et al., 2009).
12 Additionally, a study of airway inflammation reported a counterintuitive inverse association with
13 63 of similar magnitude for all groups of corticosteroid users and non-users (Qian et al., 2009).
14 Controlled human exposure studies that have examined the effects of Os on adults with
15 asthma and healthy controls are limited. Based on studies reviewed in the 1996 and 2006 63
16 AQCDs, subjects with asthma appeared to be more sensitive to acute effects of 63 in terms of
17 FEVi and inflammatory responses than healthy non-asthmatic subjects. For instance, Horstman
18 et al. (1995) observed that mild-to-moderate asthmatics, on average, experienced double the
19 Os-induced FEVi decrement of healthy subjects (19% versus 10%, respectively, p = 0.04).
20 Moreover, a statistically significant positive correlation between FEVi responses to 03 exposure
21 and baseline lung function was observed in individuals with asthma, i.e., responses increased
22 with severity of disease. Minimal evidence exists suggesting that individuals with asthma have
23 smaller Os-induced FEVi decrements than healthy subjects (3% versus 8%, respectively)
24 (Mudway et al., 2001). However, the asthmatics in that study also tended to be older than the
25 healthy subjects, which could partially explain their lesser response since FEVi responses to O^
26 exposure diminish with age. Individuals with asthma also had significantly more neutrophils in
27 the BALF (18 hours postexposure) than similarly exposed healthy individuals (Peden et al.,
28 1997; Scannell et al., 1996; Basha et al., 1994). Furthermore, a study examining the effects of O^
29 on individuals with atopic asthma and healthy controls reported that greater numbers of
30 neutrophils, higher levels of cytokines and hyaluronan, and greater expression of macrophage
31 cell-surface markers were observed in induced sputum of atopic asthmatics compared with
32 healthy controls (Hernandez et al., 2010). Differences in Os-induced epithelial cytokine
33 expression were noted in bronchial biopsy samples from asthmatics and healthy controls (Bosson
34 et al., 2003). Cell-surface marker and cytokine expression results, and the presence of
35 hyaluronan, are consistent with Os having greater effects on innate and adaptive immunity in
36 these asthmatic individuals. In addition, studies have demonstrated that O^ exposure leads to
3C-2
-------
1 increased bronchial reactivity to inhaled allergens in mild allergic asthmatics (Kehrl et al., 1999;
2 Torres et al., 1996) and to the influx of eosinophils in individuals with pre-existing allergic
3 disease (Vagaggini et al., 2002; Peden et al., 1995). Taken together, these results point to several
4 mechanistic pathways which could account for the enhanced sensitivity to Os in subjects with
5 asthma (see Section 5.4.2.2 in the ISA).
6 Toxicological studies provide additional evidence of the biological basis for the greater
7 effects of Os among those with asthma or AHR (U.S. EPA, 2013, section 8.2.2). In animal
8 toxicological studies, an asthmatic phenotype is modeled by allergic sensitization of the
9 respiratory tract. Many of the studies that provide evidence that O^ exposure is an inducer of
10 AHR and remodeling utilize these types of animal models. For example, a series of experiments
11 in infant rhesus monkeys have shown these effects, but only in monkeys sensitized to house dust
12 mite allergen. Similarly, adverse changes in pulmonary function were demonstrated in mice
13 exposed to 63; enhanced inflammatory responses were in rats exposed to 63, but only in animals
14 sensitized to allergen. In general, it is the combined effects of Os and allergic sensitization which
15 result in measurable effects on pulmonary function. In a pulmonary fibrosis model, exposure 63
16 for 5 days increased pulmonary inflammation and fibrosis, along with the frequency of
17 bronchopneumonia in rats. Thus, short-term exposure to O^ may enhance damage in a previously
18 injured lung (U.S. EPA, 2013, section 8.2.2).
19 In the 2006 Os AQCD, the potential for individuals with asthma to have greater risk of
20 Os-related health effects was supported by a number of controlled human exposure studies,
21 evidence from toxicological studies, and a limited number of epidemiologic studies. In section
22 8.2.2, the ISA reports that in the recent epidemiologic literature some, but not all, studies report
23 greater risk of health effects among individuals with asthma. Studies examining effect measure
24 modification of the relationship between short-term 63 exposure and altered lung function by
25 corticosteroid use provided limited evidence of Os-related health effects. However, recent studies
26 of behavioral responses have found that studies do not take into account individual behavioral
27 adaptations to forecasted air pollution levels (such as avoidance and reduced time outdoors),
28 which may underestimate the observed associations in studies that examined the effect of 03
29 exposure on respiratory health (Neidell and Kinney, 2010). This could explain some
30 inconsistency observed among recent epidemiologic studies. The evidence from controlled
31 human exposure studies provides support for increased detriments in FEVi and greater
32 inflammatory responses to 63 in individuals with asthma than in healthy individuals without a
33 history of asthma. The collective evidence for increased risk of Os-related health effects among
34 individuals with asthma from controlled human exposure studies is supported by recent
35 toxicological studies which provide biological plausibility for heightened risk of asthmatics to
3C-3
-------
1 respiratory effects due to Os exposure. Overall, the ISA finds there is adequate evidence for
2 asthmatics to be an at-risk population.
3 Children
4 Children are considered to be at greater risk from Os exposure because their respiratory
5 systems undergo lung growth until about 18-20 years of age and are therefore thought to be
6 intrinsically more at risk for Os-induced damage (U.S. EPA, 2006b). It is generally recognized
7 that children spend more time outdoors than adults, and therefore would be expected to have
8 higher exposure to 63 than adults. The ventilation rates also vary between children and adults,
9 particularly during moderate/heavy activity. Children aged 11 years and older and adults have
10 higher absolute ventilation rates than children aged 1-11 years. However, children have higher
11 ventilation rates relative to their lung volumes, which tends to increase dose normalized to lung
12 surface area. Exercise intensity has a substantial effect on ventilation rate, with high intensity
13 activities resulting in nearly double the ventilation rate during moderate activity among children
14 and those adults less than 31 years of age. For more information on time spent outdoors and
15 ventilation rate differences by age group, see Section 4.4.1 in the ISA (U.S. EPA, 2013).
16 The 1996 Oj, AQCD reported clinical evidence that children, adolescents, and young
17 adults (<18 years of age) appear, on average, to have nearly equivalent spirometric responses to
18 Os exposure, but have greater responses than middle-aged and older adults (U.S. EPA, 1996a).
19 Symptomatic responses (e.g., cough, shortness of breath, pain on deep inspiration) to 63
20 exposure, however, appear to increase with age until early adulthood and then gradually decrease
21 with increasing age (U.S. EPA, 1996). Complete lung growth and development is not achieved
22 until 18-20 years of age in women and the early 20s for men; pulmonary function is at its
23 maximum during this time as well.
24 Recent epidemiologic studies have examined different age groups and their risk to
25 Os-related respiratory hospital admissions and emergency department (ED) visits. Evidence for
26 greater risk in children was reported in several studies. A study in Cyprus of short-term O^
27 concentrations and respiratory hospital admissions (HA) detected possible effect measure
28 modification by age with a larger association among individuals < 15 years of age compared
29 with those > 15 years of age; the effect was apparent only with a 2-day lag (Middleton et al.,
30 2008). Similarly, a Canadian study of asthma-ED visits reported the strongest (Vrelated
31 associations among 5- to 14-year olds compared to the other age groups (ages examined 0-75+)
32 (Villeneuve et al., 2007). Greater (^-associated risk in asthma-related ED visits were also
33 reported among children (<15 years) as compared to adults (15 to 64 years) in a study from
34 Finland (Halonen et al., 2009). A study of New York City hospital admissions demonstrated an
35 increase in the association between 63 exposure and asthma-related hospital admissions for 6- to
3C-4
-------
1 18-year olds compared to those < 6 years old and those > 18 years old (Silverman and Ito, 2010).
2 When examining long-term Os exposure and asthma HA among children, associations were
3 determined to be larger among children 1 to 2 years old compared to children 2 to 6 years old
4 (Lin et al., 2008). A few studies reported positive associations among both children and adults
5 and no modification of the effect by age.
6 The evidence reported in epidemiologic studies is supported by recent toxicological
7 studies which observed (Vinduced health effects in immature animals. Early life exposures of
8 multiple species of laboratory animals, including infant monkeys, resulted in changes in
9 conducting airways at the cellular, functional, ultra-structural, and morphological levels. The
10 studies conducted on infant monkeys are most relevant for assessing effects in children. Carey et
11 al. (2007) conducted a study of Os exposure in infant rhesus macaques, whose respiratory tract
12 closely resemble that of humans. Monkeys were exposed either acutely or in episodes designed
13 to mimic human exposure. All monkeys acutely exposed to 63 had moderate to marked
14 necrotizing rhinitis, with focal regions of epithelial exfoliation, numerous infiltrating neutrophils,
15 and some eosinophils. The distribution, character, and severity of lesions in episodically exposed
16 infant monkeys were similar to that of acutely exposed animals. Neither exposure protocol for
17 the infant monkeys produced mucous cell metaplasia proximal to the lesions, an adaptation
18 observed in adult monkeys exposed in another study (Harkema et al., 1987). Functional and
19 cellular changes in conducting airways were common manifestations of exposure to Os among
20 both the adult and infant monkeys (Plopper et al., 2007). In addition, the lung structure of the
21 conducting airways in the infant monkeys was significantly stunted by 63 and this aberrant
22 development was persistent 6 months postexposure (Fanucchi et al., 2006).
23 Age may also affect the inflammatory response to 63 exposure. Toxicological studies
24 reported that the difference in effects among younger lifestage test animals may be due to
25 age-related changes in antioxidants levels and sensitivity to oxidative stress. Further discussion
26 of these studies may be found in section 8.3.1.1 of the ISA (U.S. EPA, 2013, p. 8-18).
27 The previous and recent human clinical and toxicological studies reported evidence of
28 increased risk from O^ exposure for younger ages, which provides coherence and biological
29 plausibility for the findings from epidemiologic studies. Although there was some inconsistency,
30 generally, the epidemiologic studies reported positive associations among both children and
31 adults or just among children. The interpretation of these studies is limited by the lack of
32 consistency in comparison age groups and outcomes examined. However, overall, the
33 epidemiologic, controlled human exposure, and toxicological studies provide adequate evidence
34 that children are potentially at increased risk of (Vrelated health effects.
3C-5
-------
1 Older Adults
2 The ISA notes that older adults are at greater risk of health effects associated with O3
3 exposure through a variety of intrinsic pathways (U.S. EPA, 2013, section 8.3.1.2). In addition,
4 older adults may differ in their exposure and internal dose. Older adults were outdoors for a
5 slightly longer proportion of the day than adults aged 18-64 years. Older adults also have
6 somewhat lower ventilation rates than adults aged 31 - less than 61 years. For more information
7 on time spent outdoors and ventilation rate differences by age group, see Section 4.4 in the ISA
8 (U.S. EPA, 2013). The gradual decline in physiological processes that occur with aging may lead
9 to increased risk of O3-related health effects (U.S. EPA, 2006a). Respiratory symptom responses
10 to Os exposure appears to increase with age until early adulthood and then gradually decrease
11 with increasing age (U.S. EPA, 1996a); lung function responses to O3 exposure also decline
12 from early adulthood (U.S. EPA, 1996a). The reductions of these responses with age may put
13 older adults at increased risk for continued O3 exposure. In addition, older adults, in general,
14 have a higher prevalence of preexisting diseases compared to younger age groups and this may
15 also lead to increased risk of O3-related health effects (U.S. EPA, 2013, section 8.3.1.2). With
16 the number of older Americans increasing in upcoming years (estimated to increase from 12.4%
17 of the U.S. population to 19.7% between 2000 to 2030, which is approximately 35 million and
18 71.5 million individuals, respectively) this group represents a large population potentially at risk
19 of O3-related health effects (SSDAN CensusScope, 2010a; U.S. Census Bureau, 2010).
20 The majority of recent studies reported greater effects of short-term O3 exposure and
21 mortality among older adults, which is consistent with the findings of the 2006 O3 AQCD. A
22 study (Medina-Ramon and Schwartz, 2008) conducted in 48 cities across the U.S. reported larger
23 effects among adults >65 years old compared to those < 65 years; further investigation of this
24 study population revealed a trend of O3-related mortality risk that gets larger with increasing age
25 starting at age (Zanobetti and Schwartz, 2008). Another study conducted in 7 urban centers in
26 Chile reported similar results, with greater effects in adults >65 years old (Cakmak et al., 2007).
27 More recently, a study conducted in the same area reported similar associations between O3
28 exposure and mortality in adults aged < 64 years old and 65 to 74 years old, but the risk was
29 increased among older age groups (Cakmak et al., 2011). A study performed in China reported
30 greater effects in populations >45 years old (compared to 5 to 44 year olds), with statistically
31 significant effects present only among those >65 years old (Kan et al., 2008). An Italian study
32 reported higher risk of all-cause mortality associated with increased O3 concentrations among
33 individuals >85 year old as compared to those 35 to 84 years old (Stafoggia et al., 2010). The Air
34 Pollution and Health: A European and North American Approach (APHENA) project examined
35 the association between O3 exposure and mortality for those <75 and > 75 years of age. In
3C-6
-------
1 Canada, the associations for all-cause and cardiovascular mortality were greater among those
2 >75 years old. In the U.S., the association for all-cause mortality was slightly greater for those
3 <75 years of age compared to those >75 years old in summer-only analyses. No consistent
4 pattern was observed for CVD mortality. In Europe, slightly larger associations for all-cause
5 mortality were observed in those <75 years old in all-year and summer-only analyses. Larger
6 associations were reported among those <75years for CVD mortality in all-year analyses, but the
7 reverse was true for summer-only analyses (Katsouyanni et al., 2009).
8 With respect to epidemiologic studies of Os exposure and hospital admissions, a positive
9 association was reported between short-term 63 exposure and respiratory hospital admissions for
10 adults >65 years old but not for those adults aged 15 to 64 years (Halonen et al., 2009). In the
11 same study, no association was observed between Oj, concentration and respiratory mortality
12 among those >65 years old or those 15 to 64 years old. No modification by age (40 to 64 year
13 olds versus >64 year olds) was observed in a study from Brazil examining O?, levels and COPD
14 ED visits.
15 Although some outcomes reported mixed findings regarding an increase in risk for older
16 adults, recent epidemiologic studies report consistent positive associations between short-term
17 Os exposure and mortality in older adults. The evidence from mortality studies is consistent with
18 the results reported in the 2006 63 AQCD and is supported by toxicological studies providing
19 biological plausibility for increased risk of effects in older adults. Also, older adults may be
20 experiencing increased exposure compared to younger adults. Overall, the ISA (U.S. EPA 2013)
21 concludes adequate evidence is available indicating that older adults are at increased risk of
22 O3-related health effects.
23 People with Diets Lower in Vitamins C and E
24 Diet was not examined as a factor potentially affecting risk in previous Os AQCDs, but
25 recent studies have examined modification of the association between Oj, and health effects by
26 dietary factors. Because Oj mediates some of its toxic effects through oxidative stress, the
27 antioxidant status of an individual is an important factor that may contribute to increased risk of
28 Os-related health effects. Supplementation with vitamins C and E has been investigated in a
29 number of studies as a means of inhibiting Os-mediated damage.
30 Two epidemiologic studies have examined effect measure modification by diet and found
31 evidence that certain dietary components are related to the effect O^ has on respiratory outcomes.
32 In one recent study the effects of fruit/vegetable intake and Mediterranean diet were examined.
33 Increases in these food patterns, which have been noted for their high vitamins C and E and
34 omega-3 fatty acid content, were positively related to lung function in asthmatic children living
3C-7
-------
1 in Mexico City, and modified by Os exposure (Romieu et al., 2009). Another study examined
2 supplementation of the diets of asthmatic children in Mexico with vitamins C and E (Sienra-
3 Monge et al., 2004). Associations were detected between short-term Os exposure and nasal
4 airway inflammation among children in the placebo group but not in those receiving the
5 supplementation.
6 The epidemiologic evidence is supported by controlled human exposure studies,
7 discussed in section 8.4.1 of the ISA (U.S. EPA 2013), that have shown that the first line of
8 defense against oxidative stress is antioxidants-rich extracellular lining fluid (ELF) which
9 scavenge free radicals and limit lipid peroxidation. Exposure to 63 depletes antioxidant levels in
10 nasal ELF probably due to scrubbing of Cb; however, the concentration and the activity of
11 antioxidant enzymes either in ELF or plasma do not appear to be related to 63 responsiveness.
12 Controlled studies of dietary antioxidant supplementation have demonstrated some protective
13 effects of a-tocopherol (a form of vitamin E) and ascorbate (vitamin C) on spirometric measures
14 of lung function after 63 exposure but not on the intensity of subjective symptoms and
15 inflammatory responses. Dietary antioxidants have also afforded partial protection to asthmatics
16 by attenuating postexposure bronchial hyperresponsiveness. Toxicological studies discussed in
17 section 8.4.1 of the ISA (U.S. EPA 2013) provide evidence of biological plausibility to the
18 epidemiologic and controlled human exposure studies.
19 There is adequate evidence that individuals with diets lower in vitamins C and E are at
20 risk for Os-related health effects. The evidence from epidemiologic studies is supported by
21 controlled human exposure and toxicological studies.
22 Outdoor Workers
23 Studies included in the 2006 63 AQCD reported that individuals who participate in
24 outdoor activities or work outside to be a population at increased risk based on consistently
25 reported associations between 63 exposure and respiratory health outcomes in these groups (U.S.
26 EPA, 2006b). Outdoor workers are exposed to ambient Oj concentrations for a greater period of
27 time than individuals who spend their days indoors. As discussed in Section 4.7 of the ISA (U.S.
28 EPA, 2013) outdoor workers sampled during the work shift had a higher ratio of personal
29 exposure to fixed-site monitor concentrations than health clinic workers who spent most of their
30 time indoors. Additionally, an increase in dose to the lower airways is possible during outdoor
31 exercise due to both increases in the amount of air breathed (i.e., minute ventilation) and a shift
32 from nasal to oronasal breathing. The association between FEVi responses to Os exposure and
33 minute ventilation is discussed more fully in Section 6.2.3.1 of the 2006 Os AQCD.
3C-8
-------
1 Previous studies have shown that increased exposure to O^ due to outdoor work leads to
2 increased risk of Cb-related health effects, specifically decrements in lung function (U.S. EPA,
3 2006b). The strong evidence from the 2006 Os AQCD which demonstrated increased exposure,
4 dose, and ultimately risk of Os-related health effects in this population supports the conclusion
5 that there is adequate evidence to indicate that increased exposure to 63 through outdoor work
6 increases the risk of Os-related health effects.
7 3.3 REFERENCES
8 Alexeeff, SE; Litonjua, AA; Suh, H; Sparrow, D; Vokonas, PS; Schwartz, J. (2007). Ozone exposure and lung
9 function: Effect modified by obesity and airways hyperresponsiveness in the VA Normative Aging Study.
10 Chest 132: 1890-1897. http://dx.doi.org/10.1378/chest.07-1126
11 Barraza-Villarreal, A; Sunyer, J; Hernandez-Cadena, L; Escamilla-Nunez, MC; Sienra-Monge, JJ; Ramirez-Aguilar,
12 M; Cortez-Lugo, M; Holguin, F; Diaz-Sanchez, D; Olin, AC; Romieu, I. (2008). Air pollution, airway
13 inflammation, and lung function in a cohort study of Mexico City schoolchildren. Environ Health Perspect
14 116: 832-838. http://dx.doi.org/10.1289/ehp.10926
15 Basha, MA; Gross, KB; Gwizdala, CJ; Haidar, AH; Popovich, J, Jr. (1994). Bronchoalveolar lavage neutrophilia in
16 asthmatic and healthy volunteers after controlled exposure to ozone and filtered purified air. Chest 106:
17 1757-1765.
18 Berhane, K; Zhang, Y; Linn, WS; Rappaport, EB; Bastain, TM; Salam, MT; Islam, T; Lurmann, F; Gilliland, FD.
19 (2011). The effect of ambient air pollution on exhaled nitric oxide in the Children's Health Study. Eur
20 Respir J 37: 1029-1036. http://dx.doi.org/10.1183/09031936.00081410
21 Bosson, J; Stenfors, N; Bucht, A; Helleday, R; Pourazar, J; Holgate, ST; Kelly, FJ; Sandstrom, T; Wilson, S; Frew,
22 AJ; Blomberg, A. (2003). Ozone-induced bronchial epithelial cytokine expression differs between healthy
23 and asthmatic subjects. Clin Exp Allergy 33: 777-782.
24 Cakmak, S; Dales, RE; Angelica Rubio, M; Blanco Vidal, C. (2011). The risk of dying on days of higher air
25 pollution among the socially disadvantaged elderly. Environ Res 111: 388-393.
26 http://dx.doi.0rg/10.1016/j.envres.2011.01.003
27 Cakmak, S; Dales, RE; Vidal, CB. (2007). Air pollution and mortality in Chile: Susceptibility among the elderly.
28 Environ Health Perspect 115: 524-527.
29 Carey, SA; Minard, KR; Trease, LL; Wagner, JG; Garcia, GJ; Ballinger, CA; Kimbell, JS; Plopper, CG; Corley,
30 RA; Postlethwait, EM; Harkema, JR; Einstein, DR. (2007). Three-dimensional mapping of ozone-induced
31 injury in the nasal airways of monkeys using magnetic resonance imaging and morphometric techniques.
32 Toxicol Pathol 35: 27-40. http://dx.doi.org/10.1080/01926230601072343
33 Escamilla-Nunez, MC; Barraza-Villarreal, A; Hernandez-Cadena, L; Moreno-Macias, H; Ramirez-Aguilar, M;
34 Sienra-Monge, JJ; Cortez-Lugo, M; Texcalac, JL; del Rio-Navarro, B; Romieu, I. (2008). Traffic-related
3 5 air pollution and respiratory symptoms among asthmatic children, resident in Mexico City: The EVA
36 cohort study. Respir Res 9: 74. http://dx.doi.org/10.1186/1465-9921-9-74
37 Fanucchi, MV; Plopper, CG; Evans, MJ; Hyde, DM; Van Winkle, LS; Gershwin, LJ; Schelegle, ES. (2006). Cyclic
3 8 exposure to ozone alters distal airway development in infant rhesus monkeys. Am J Physiol Lung Cell Mol
39 Physiol 291: L644-L650. http://dx.doi.org/10.1152/ajplung.00027.2006
3C-9
-------
1 Harkema, JR; Plopper, CG; Hyde, DM; St George, JA; Dungworth, DL. (1987a). Effects of an ambient level of
2 ozone on primate nasal epithelial mucosubstances: quantitative histochemistry. Am J Pathol 127: 90-96.
3 Hernandez, ML; Lay, JC; Harris, B; Esther, CR; Brickey, WJ; Bromberg, PA; Diaz-Sanchez, D; Devlin, RB;
4 Kleeberger, SR; Alexis, NE; Peden, DB. (2010). Atopic asthmatic subjects but not atopic subjects without
5 asthma have enhanced inflammatory response to ozone. J Allergy Clin Immunol 126: 537-544.
6 http://dx.doi.org/10.1016/jjaci.2010.06.043
7 Hernandez-Cadena, L; Holguin, F; Barraza-Villarreal, A; Del Rio-Navarro, BE; Sienra-Monge, JJ; Romieu, I.
8 (2009). Increased levels of outdoor air pollutants are associated with reduced bronchodilation in children
9 with asthma. Chest 136: 1529-1536. http://dx.doi.org/10.1378/chest.08-1463
10 Halonen, JI; Lanki, T; Tiittanen, P; Niemi, JV; Loh, M; Pekkanen, J. (2009). Ozone and cause-specific
11 cardiorespiratory morbidity and mortality. J Epidemiol Community Health 64: 814-820.
12 http://dx.doi.org/10.1136/jech.2009.087106
13 Horstman, DH; Ball, BA; Brown, J; Gerrity, T; Folinsbee, LJ. (1995). Comparison of pulmonary responses of
14 asthmatic and nonasthmatic subjects performing light exercise while exposed to a low level of ozone.
15 Toxicol Ind Health 11: 369-385.
16 Torres, R; Nowak, D; Magnussen, H; Speckin, P; Koschyk, S. (1996). The effect of ozone exposure on allergen
17 responsiveness in subjects with asthma or rhinitis. Am J Respir Crit Care Med 153: 56-64.
18 Kan, H; London, SJ; Chen, G; Zhang, Y; Song, G; Zhao, N; Jiang, L; Chen, B. (2008). Season, sex, age, and
19 education as modifiers of the effects of outdoor air pollution on daily mortality in Shanghai, China: The
20 Public Health and Air Pollution in Asia (PAPA) Study. Environ Health Perspect 116: 1183-1188.
21 http://dx.doi.org/10.1289/ehp. 10851
22 Katsouyanni, K; Samet, JM; Anderson, HR; Atkinson, R; Le Tertre, A; Medina, S; Samoli, E; Touloumi, G;
23 Burnett, RT; Krewski, D; Ramsay, T; Dominici, F; Peng, RD; Schwartz, J; Zanobetti, A. (2009). Air
24 pollution and health: A European and North American approach (APHENA). (Research Report 142).
25 Boston, MA: Health Effects Institute. http://pubs.healtheffects.org/view.php?id=327
26 Kehrl, HR; Peden, DB; Ball, BA; Folinsbee, LJ; Horstman, DH. (1999). Increased specific airway reactivity of
27 persons with mild allergic asthma after 7.6 hours of exposure to 0.16 ppm ozone. J Allergy Clin Immunol
28 104: 1198-1204.
29 Khatri, SB; Holguin, FC; Ryan, PB; Mannino, D; Erzurum, SC; Teague, WG. (2009). Association of ambient ozone
30 exposure with airway inflammation and allergy in adults with asthma. J Asthma 46: 777-785.
31 http://dx.doi.org/10.1080/02770900902779284
32 Lewis, TC; Robins, TG; Dvonch, JT; Keeler, GJ; Yip, FY; Mentz, GB; Lin, X; Parker, EA; Israel, BA; Gonzalez, L;
3 3 Hill, Y. (2005). Air pollution-associated changes in lung function among asthmatic children in Detroit.
34 Environ Health Perspect 113: 1068-1075.
35 Lin, S; Liu, X; Le, LH; Hwang, SA. (2008b). Chronic exposure to ambient ozone and asthma hospital admissions
36 among children. Environ Health Perspect 116: 1725-1730. http://dx.doi.org/10.1289/ehp.11184
37 Liu, L; Poon, R; Chen, L; Frescura, AM; Montuschi, P; Ciabattoni, G; Wheeler, A; Dales, R. (2009). Acute effects
38 of air pollution on pulmonary function, airway inflammation, and oxidative stress in asthmatic children.
39 Environ Health Perspect 117: 668-674. http://dx.doi.org/10.1289/ehpll
40 Medina-Ramon, M; Schwartz, J. (2008). Who is more vulnerable to die from ozone air pollution? Epidemiology 19:
41 672-679.
3C-10
-------
1 Middleton, N; Yiallouros, P; Kleanthous, S; Kolokotroni, O; Schwartz, J; Dockery, DW; Demokritou, P; Koutrakis,
2 P. (2008). A 10-year time-series analysis of respiratory and cardiovascular morbidity in Nicosia, Cyprus:
3 The effect of short-term changes in air pollution and dust storms. Environ Health 7:39.
4 Mudway, IS; Stenfors, N; Blomberg, A; Helleday, R; Dunster, C; Marklund, SL; Frew, AJ; Sandstrom, T; Kelly, FJ.
5 (2001). Differences in basal airway antioxidant concentrations are not predictive of individual
6 responsiveness to ozone: A comparison of healthy and mild asthmatic subjects. Free Radic Biol Med 31:
7 962-974.
8 Neidell, M; Kinney, PL. (2010). Estimates of the association between ozone and asthma hospitalizations that
9 account for behavioral responses to air quality information. Environ Sci Pol 13: 97-103.
10 http://dx.doi.0rg/10.1016/j.envsci.2009.12.006
11 Peden, DB; Boehlecke, B; Horstman, D; Devlin, R. (1997). Prolonged acute exposure to 0.16 ppm ozone induces
12 eosinophilic airway inflammation in asthmatic subjects with allergies. J Allergy Clin Immunol 100: 802-
13 808.
14 Peden, DB; Setzer, RW, Jr; Devlin, RB. (1995). Ozone exposure has both a priming effect on allergen-induced
15 responses and an intrinsic inflammatory action in the nasal airways of perennially allergic asthmatics. Am J
16 RespirCrit Care Med 151: 1336-1345.
17 Plopper, CG; Smiley-Jewell, SM; Miller, LA; Fanucchi, MV; Evans, MJ; Buckpitt, AR; Avdalovic, M; Gershwin,
18 LJ; Joad, JP; Kajekar, R; Larson, S; Pinkerton, KE; Van Winkle, LS; Schelegle, ES; Pieczarka, EM; Wu,
19 R; Hyde, DM. (2007). Asthma/allergic airways disease: Does postnatal exposure to environmental
20 toxicants promote airway pathobiology? Toxicol Pathol 35: 97-110.
21 http://dx.doi.org/10.1080/01926230601132030
22 Romieu, I; Barraza-Villarreal, A; Escamilla-Nunez, C; Texcalac-Sangrador, JL; Hernandez-Cadena, L; Diaz-
23 Sanchez, D; De Batlle, J; Del Rio-Navarro, BE. (2009). Dietary intake, lung function and airway
24 inflammation in Mexico City school children exposed to air pollutants. Respir Res 10: 122.
25 Scannell, C; Chen, L; Aris, RM; Tager, I; Christian, D; Ferrando, R; Welch, B; Kelly, T; Balmes, JR. (1996).
26 Greater ozone-induced inflammatory responses in subjects with asthma. Am J Respir Crit Care Med 154:
27 24-29.
28 Sienra-Monge, JJ; Ramirez-Aguilar, M; Moreno-Macias, H; Reyes-Ruiz, NI; Del Rio-Navarro, BE; Ruiz-Navarro,
29 MX; Hatch, G; Crissman, K; Slade, R; Devlin, RB; Romieu, I. (2004). Antioxidant supplementation and
3 0 nasal inflammatory responses among young asthmatics exposed to high levels of ozone. Clin Exp Immunol
31 138: 317-322. http://dx.doi.0rg/10.llll/j.1365-2249.2004.02606.x
32 Silverman, RA; Ito, K. (2010). Age-related association of fine particles and ozone with severe acute asthma in New
33 York City. J Allergy Clin Immunol 125: 367-373. http://dx.doi.org/10.1016/jjaci.2009.10.061
34 SSDAN CensusScope (Social Science Data Analysis Network, CensusScope). (2010a). United States: Age
35 distribution [Database]. Ann Arbor, Michigan: Social Science Data Analysis Network. Retrieved from
3 6 http://www.censusscope.org/us/chart_age.html
37 Stafoggia, M; Forastiere, F; Faustini, A; Biggeri, A; Bisanti, L; Cadum, E; Cernigliaro, A; Mallone, S; Pandolfi, P;
38 Serinelli, M; Tessari, R; Vigotti, MA; Perucci, CA. (2010). Susceptibility factors to ozone-related
39 mortality: A population-based case-crossover analysis. Am J Respir Crit Care Med 182: 376-384.
40 http://dx.doi.org/10.1164/rccm.200908-1269OC
41 Thaller, El; Petronella, SA; Hochman, D; Howard, S; Chhikara, RS; Brooks, EG. (2008). Moderate increases in
42 ambient PM2.5 and ozone are associated with lung function decreases in beach lifeguards. J Occup Environ
43 Med 50: 202-211. http://dx.doi.org/10.1097/JOM.Ob013e31816386b4
3C-11
-------
1 U.S. EPA (2013). Integrated Science Assessment of Ozone and Related Photochemical Oxidants (Final Report).
2 U.S. Environmental Protection Agency, Washington, DC. EPA/600/R-10/076F. Available at:
3 http://www.epa.gov/ttn/naaqs/standards/ozone/s_o3_2008_isa.html
4 U.S. EPA (U.S. Environmental Protection Agency). (2006b). Air quality criteria for ozone and related
5 photochemical oxidants [EPA Report]. (EPA/600/R-05/004AF). Research
6 U.S. EPA (U.S. Environmental Protection Agency). (1996a). Air quality criteria for ozone and related
7 photochemical oxidants [EPA Report]. (EPA/600/P-93/004AF). Research Triangle Park, NC.
8 U.S. EPA (U.S. Environmental Protection Agency). (2006a). Aging and toxic response: Issues relevant to risk
9 assessment [EPA Report]. (EPA/600/P-03/004A). Washington, DC.
10 http://cfpub.epa.gov/ncea/cfm/recordisplay.cfm?deid=156648
11 U.S. EPA(2006b). Air Quality Criteria for Ozone and Related Photochemical Oxidants (2006 Final). U.S.
12 Environmental Protection Agency, Washington, DC. EPA/600/R-05/004aF-cF. March 2006. Available at:
13 http://www.epa.gov/ttn/naaqs/standards/ozone/s_o3_cr_cd.html
14 U.S. EPA (1996). Air quality criteria for ozone and related photochemical oxidants [EPA Report]. (EPA/600/P-
15 93/004aF, cF). U.S. Environmental Protection Agency, Research Triangle Park, NC.
16 Vagaggini, B; Taccola, M; Clanchetti, S; Carnevali, S; Bartoli, ML; Bacci, E; Dente, FL; Di Franco, A; Giannini, D;
17 Paggiaro, PL. (2002). Ozone exposure increases eosinophilic airway response induced by previous allergen
18 challenge. Am J Respir Crit Care Med 166: 1073-1077.
19 Villeneuve, PJ; Chen, L; Rowe, BH; Coates, F. (2007). Outdoor air pollution and emergency department visits for
20 asthma among children and adults: A case-crossover study in northern Alberta, Canada. Environ Health
21 Global Access Sci Source 6: 40. http://dx.doi.org/10.1186/1476-069X-6-40
22 Zanobetti, A; Schwartz, J. (2008a). Is there adaptation in the ozone mortality relationship: A multi-city case-
23 crossover analysis. Environ Health 7: 22. http://dx.doi.org/10.1186/1476-069X-7-22
3C-12
-------
1 APPENDIX 3-D: AMBIENT O3 CONCENTRATIONS IN LOCATIONS OF
2 HEALTH STUDIES
3 Annual 4th highest daily maximum Oj, concentrations for all U.S. monitors operating
4 during the 1975 - 2010 period were retrieved from EPA's AQS database. These data were used
5 to calculate 63 design values for the 2008 8-hour 63 NAAQS of 0.075 parts per million (ppm)
6 according to 40 CFR part 50, Appendix P. Design values were calculated for each 63 monitor
7 and each 3-year period between 1975-1977 and 2008-2010 whenever sufficient data were
8 available.
9 Ozone Design Values in Study Locations
10 Ozone monitors were matched to 200 health study locations on a case-by-case basis, using
11 the following guidelines:
12 1) Areas defined by a Metropolitan Statistical Area (MSA) were matched with 63 monitors
13 by incorporating all of the monitors located in within the MSA boundaries.
14 2) Areas not represented by a MSA were matched to monitors by incorporating all of the
15 monitors in the county central to location of the health study area.
16 3) In some cases, EPA staff made judgment calls. For example, EPA staff matched the Los
17 Angeles, CA study area to the Los Angeles-Long Beach-Santa Ana, CA MSA defined by
18 Los Angeles County, CA and Orange County, CA, while the Long Beach, CA study area
19 was matched to Los Angeles County, CA and the Santa Ana, CA study area was matched
20 to Orange County, CA.
21 In some cases, EPA staff matched two or more study areas to the same county or MSA.
22 In other cases, a study area was matched to a MSA and another study area was matched to a
23 county within the same MSA. For each 3-year period, the area design value was determined by
24 the monitor reporting the highest design value in the county or MSA. This has two implications
25 for the design values:
26 1) Design values are sensitive to changes in the monitoring network. The addition or
27 discontinuation of Oj monitors in an area may cause increases or decreases in the design
28 value trend.
29 2) Only valid design values are reported. According to 40 CFR Part 50, Appendix P, design
30 values greater than the level of the NAAQS (0.075 ppm) are always valid, while design
31 values less than or equal to 0.075 ppm must have 75% annual data completeness in order
32 to be valid. This may cause anomalies in the design value trend. For example, a monitor
33 may report a valid design value based on as few as 12 days of data, or a monitor with less
34 than 75% annual data completeness may have valid design values in some 3-year periods
35 and invalid design values in others.
3D-1
-------
36 We have identified design values for the U.S. O^ epidemiologic studies identified in Sections
37 3.1.4.2 and 3.1.4.3 of the second draft Policy Assessment (see Tables 3D-1 and 3D-2,
38 respectively). For each study, design values were identified for the cities evaluated and for the
39 years over which the study was conducted. These design values are reported in tables A-l to A-
40 22 of the Wells et al, 2012 memo "Analysis of Recent U.S. Ozone Air Quality Data to Support
41 the 63 NAAQS Review and Quadratic Rollback Simulations to Support the First Draft of the
42 Risk and Exposure Assessment".
43 Table 3D-1 Number of Study Cities from Epidemiologic Studies Using Short-Term Os
44 Metrics with 3-Year Averages of Annual 4th Highest Daily Maximum 8-hour
45 O3 Concentrations > 75, 70, 65, or 60 ppb1
Study
Belletal.20042
Bell et al. 2006, 2007, 2008
Cakmakal. 2006
Dales etal. 2006
Franklin & Schwartz, 2008
Katsouyanni et al. 20093
Katsouyannietal. 20095
Katsouyanni etal. 20096
Medina-Ramon et al. 2006
Schwartz, 2005
Stieb et al. 2009
Zanobetti & Schwartz,
2008a,b
Number
of Cities
95
98
10
11
18
894
14
12
26'
14
7
48
Study
Period
1987-2000
1987-2000
1993-2000
1986-2000
2000-2005
1987-1996
1987-1996
1987-1996
1986-1999
1986-1993
1992-2003
1989-2000
Number
(Percent) of
Cities >75
89 (94%)
92 (94%)
3 (30%)
4 (36%)
17 (94%)
83 (93%)
12 (86%)
4 (33%)
24 (92%)
13 (93%)
2 (29%)
44 (92%)
Number
(Percent) of
Cities >70
91 (96%)
94 (96%)
3 (30%)
6 (55%)
18 (100%)
84 (94%)
12 (86%)
6 (50%)
24 (92%)
13 (93%)
2 (29%)
45 (94%)
Number
(Percent) of
Cities >65
93 (98%)
96 (98%)
4 (40%)
7 (64%)
18 (100%)
86 (97%)
14 (100%)
7 (58%)
26 (100%)
14 (100%)
3 (43%)
47 (98%)
Number
(Percent) of
Cities >60
94 (99%)
97 (99%)
8 (80%)
10 (91%)
18(100%)
88 (99%)
14(100%)
11 (92%)
26(100%)
14(100%)
4 (57%)
47 (98%)
1 For U.S. study areas, we used EPA's Air Quality System (AQS) (http://www.epa.gov/ttn/airs/airsaqsA) to identify
8-hour O3 concentrations. For Canadian study areas, we used publically available air quality data from the
Environment Canada National Air Pollution Surveillance Network (http://www.etc-
cte.ec.gc.ca/napsdata/main/aspxX We followed the data handling protocols for calculating design values as detailed
in 40 CFR Part 50, Appendix P.
2We also evaluated the bayes-adjusted effect estimates for individual cities presented by Bell et al. (2004). None of
the cities for which individual city effect estimates were statistically significant would have met the current standard
over the study period.
3 U.S. cities; examining mortality
4 Study authors included 90 cities in their analyses; air quality data that met completeness criteria described above
were available for 89 cities
5 U.S. cities; examining morbidity
6 Canadian cities; examining mortality and morbidity
7 Study authors included 27 cities in their analyses; air quality data that met completeness criteria described above
were available for 26 cities
3D-2
-------
46 Table 3D-2 Number of Study Cities from Epidemiologic Studies Using Long-Term Os
47 Metrics with 3-Year Averages of Annual 4th Highest Daily Maximum 8-hour
48 Os Concentrations > 75, 70, 65, or 60 ppb
Study
Islam etal. 2008, 2009
Jerrett et al. 2009
Lin etal. 2008
Meng etal. 2010
Moore etal. 2008
Salam etal. 2009
Zanobetti & Schwartz 2011
Number
of Cities
11s
949
2610
7
8
II11
105
Study
Period
1994-2003
1977-2000
1991-2001
1997-2002
1980-2000
1992-2005
1985-2006
Number
(Percent) of
Cities with
Maximum
cone >75
11 (100%)
91 (97%)
24 (92%)
7 (100%)
8 (100%)
12 (100%)
100 (95%)
Number
(Percent) of
Cities with
Maximum
cone >70
11 (100%)
92 (98%)
24 (92%)
7(100%)
8(100%)
12 (100%)
104 (99%)
Number
(Percent) of
Cities with
Maximum cone
>65
11 (100%)
93 (99%)
26(100%)
7 (100%)
8 (100%)
12 (100%)
104 (99%)
Number
(Percent) of
Cities with
Maximum
cone >60
11 (100%)
94(100%)
26(100%)
7 (100%)
8 (100%)
12 (100%)
104 (99%)
49
50
Study authors included 12 cities in their analyses, air quality data that met completeness criteria described above
were available for 11 cities
9 Study authors included 96 cities in their analyses, air quality data that met completeness criteria described above
were available for 94 cities
10 Study authors included 27 cities in their analyses, air quality data that met completeness criteria described
above were available for 26 cities
11 Study authors included 12 cities in their analyses, air quality data that met completeness criteria described
above were available for 11 cities
3D-3
-------
51 Relationship between average 24-hour and highest 8-hour Os concentrations for cities
52 analyzed by Bell et al. (2006)
53 Bell et al. (2006) reported associations between mortality and 24-hour average 63
54 concentrations (i.e., averaged across monitors in cities with multiple monitors) in a multi-city
55 study of 98 U.S. cities. Positive associations persisted in a series of analyses that restricted 63
56 concentrations to those below various cut points (cut points ranged from 5 to 60 ppb in 5 ppb
57 increments). To facilitate consideration of these cut point analyses for the second draft of the 63
58 Policy Assessment, so as to match the form and averaging time of the existing primary standard,
59 we evaluated the relationship between 24-hour average Os concentrations, averaged across
60 monitors in cities with multiple monitors, and the highest 8-hour daily maximum Os
61 concentrations among the individual monitors in each city.
62 EPA staff retrieved daily 24-hour average and 8-hour maximum Os concentrations
63 reported to EPA by monitors in the 98 study areas defined in Bell et al. (2006) during the 1987-
64 2012 period from EPA's Air Quality System (AQS) database. Next, EPA staff obtained the
65 study area boundaries from the published study (Bell et al., 2006) and used them to determine
66 which Os monitoring sites were associated with each study area. The 24-hour average 03
67 concentrations were averaged spatially across all available monitors within each study area on
68 each day where monitoring data were collected. Next, days where the area-wide 24-hour average
69 concentration (i.e., averaged spatially across monitors in areas with multiple monitors) was
70 greater than 60 ppb were removed from the data. Based on the data remaining (i.e., with 24-hour
71 average concentrations of 60 ppb or below), the annual 4th highest 8-hour daily maximum
72 concentrations were identified for each study area and for each year from 1987-2012 (Table 3D-
73 3). This process was repeated by further removing days with area-wide 24-hour average
74 concentrations greater than 55 ppb, 50 ppb, etc., down to 5 ppb, and re-calculating the same
75 statistics after each removal. The resulting dataset consisted of the annual 4th highest 8-hour
76 daily maximum concentrations for all study areas.
77
3D-4
-------
78 Table 3D-3 Number of Study Cities with 4th Highest 8-hour Daily Maximum
79 Concentrations Greater Than the Level of the Current Standard and
80 Potential Alternative Standards For Various Cut-Point Analyses Presented
81 in Bell et al. (2006)12
Cut-point for 2-day moving average across monitors and cities (24-h avg)
20 25 30 35 40 45 50 55 60 All
Number (%) of Cities
with 4th highest >75 0(0%) 0(0%) 12(12%) 52(53%) 77(79%) 88(90%) 93(95%) 94(96%) 94(96%) 94(96%)
(any year; 1987-2000)
Number (%) of Cities
with 4th highest >70 0(0%) 3(3%) 31(32%) 77(79%) 86(88%) 93(95%) 94(96%) 94(96%) 95(97%) 95(97%)
(any year; 1987-2000)
Number (%) of Cities
with 4th highest >65 0(0%) 10(10%) 58(59%) 84(86%) 93(95%) 94(96%) 94(96%) 94(96%) 94(96%) 94(96%)
(any year; 1987-2000)
Number (%) of Cities
with 4th highest >60 1(1%) 36(37%) 74(76%) 93(95%) 96(8%) 97(99%) 97(99%) 97(99%) 97(99%) 97(99%)
(any year; 1987-2000)
82
83
12 Study authors included 98 cities in their analyses, air quality data only available for 95
3D-5
-------
84 Relationship between average and highest 8-hour daily maximum Os concentrations for
85 New York City, as analyzed by Silverman and Ito (2010)
86 EPA staff retrieved daily maximum 8-hour Os concentrations for the 13 monitors in the
87 New York City area used in the Silverman and Ito (2010) study for April-August of 1999-2006
88 from the AQS database. Next, EPA staff spatially averaged these concentrations across monitors
89 for each day during this period, and then paired them with the highest 8-hour daily maximum
90 value reported across the 13 monitors on each day.
91 Next, the range of observed average daily maximum 8-hour concentrations was broken
92 into 5 ppb increments. The number of days where the area-wide average daily maximum 8-hour
93 concentration fell within the increment and the number of days where one or more monitored 8-
94 hour daily maximum values were greater than 75, 70, 65 and 60 ppb were recorded for each 5
95 ppb increment. These numbers are summarized in Table 3D-4.
96 Table 3D-4 Summary statistics for Observed Oi Concentrations in the New York City
97 Area, April - August 1999 - 2006
2-day moving average across monitors (ppb)
11 to 20 21 to 25 26 to 30 31 to 35 36 to 40 41 to 45 46 to 50 51 to 55 56 to 60
(62 days) (92 days) (178 days) (206 days) (236 days) (196 days) (153 days) (111 days) (71 days)
98
99
Days > 75 ppb
Days > 70 ppb
Days > 65 ppb
Days > 60 ppb
0
0
0
0
0
0
0
0
1
1
1
2
0
4
6
7
1
1
5
12
2
12
18
39
9
17
37
67
15
23
42
61
20
30
45
53
3D-6
-------
100 Relationship between average and highest 8-hour daily maximum Os concentrations for
101 Atlanta, as analyzed by Strickland et al. (2010)
102 For our assessment of the Strickland et al. (2010) study, based in the Atlanta metropolitan
103 area, we retrieved 8-hour daily maximum concentration data for 4 of the 5 monitors used in the
104 study during the study period (May-October, 1993-2004) from the AQS database. The 5th
105 monitor was a part of the Southeastern Aerosol Research and Characterization (SEARCH)
106 network, which does not report data to EPA. EPA staff calculated the area-wide average of the
107 8-hour daily maximum concentrations for each day, and compared to population-weighted
108 average concentrations obtained from the author. The correlation between the arithmetic average
109 values and the population-weighted average values was very high (R = 0.985), thus EPA staff
110 deemed the arithmetic average to be a suitable surrogate for the population-weighted average
111 used in the study. Finally, 3-day moving averages were calculated from the daily area-wide
112 average values (matching the air quality metric used in the study), and paired with the highest
113 monitored 8-hour daily maximum value occurring during each 3-day period.
114 Next, the range of observed average daily maximum 8-hour concentrations was broken
115 into 5 ppb increments. The number of days where the area-wide average daily maximum 8-hour
116 concentration fell within the increment and the number of days where one or more monitored 8-
117 hour daily maximum values were greater than 75, 70, 65 and 60 ppb were recorded for each 5
118 ppb increment. These numbers are summarized in Table 3D-5.
119 Table 3D-5 Summary statistics for Observed O^ Concentrations in the Atlanta Area,
120 April - August 1999 - 2006
3-day moving average across monitors (ppb)
26-30 31-35 36-40 41-45 46-50 51-55 56-60 61-65 66-70 71 to 75 76 to 80
(75 days) (144 days) (165 days) (210 days) (235 days) (244 days) (272 days) (234 days) (169 days) (124 days) (106 days)
122
Days > 75
Days > 70
Days > 65
Days > 60
0
0
1
1
0
0
0
2
2
6
8
15
2
6
19
33
10
20
38
68
24
49
75
115
53
81
118
152
80
111
147
173
89
107
133
147
87
96
106
116
87
95
100
102
3D-7
-------
123 Relationship between annual and highest 1-hour daily maximum Os concentrations for 12
124 study areas, as analyzed by Jerrett et al. (2009)
125 The Jerrett et al. (2009) study used a long-term metric based on seasonal averages of 1-
126 hour daily maximum 63 concentrations to evaluate associations between respiratory mortality
127 and long-term or repeated exposures to 63. Authors divided study cities into quartiles based on
128 these seasonal averages of 1-hour daily 63 concentrations. Using AQS, we identified the 3-year
129 averages of annual 4th highest daily maximum 8-hour 63 concentrations in study cities during the
130 study period. Table 3D-6 presents the means and maximums of these concentrations over the
131 study period.
132 In addition, for the 12 urban case study areas included in the epidemiology-based risk
133 assessment of the 2nd draft of the Health REA we identified the seasonal averages of 1-hour daily
134 maximum concentrations (i.e., the 63 metric evaluated by Jerrett et al., 2009) for air quality
135 adjusted to the current and alternative standards. Specifically, for adjusted air quality "quarterly"
136 averages of 1-hour concentrations for April-June and July-August were calculated for each area
137 and year. The quarterly values were considered to be valid if valid daily maximum 1-hour
138 values were available for at least 75% of the days in the quarter. The two quarterly values were
139 then averaged, as was done by Jerrett et al. (2009) to generate the long-term metric used in the
140 study. This process was repeated for the various model-based adjustment scenarios in each of
141 the 12 study areas. Summary statistics based on this seasonal average of daily 03 concentrations
142 are presented in Table 3D-7 for recent air quality and for air quality adjusted to just meet the
143 current and alternative standards.
144
3D-8
-------
145
146
th
Table 3D-6 Three-Year Averages of Annual 4 Highest Daily Maximum 8-hour O
,13
Concentrations in 94 Study Areas Examined in Jerrett et al. (2009)
*t
0)
D
l/>
O
Q.
X
0)
0)
00
tc
OJ
tc
^
_QJ
t
tc
D
!T
4-»
LO
0)
_O
0)
4->
c
LO
0)
4->
U
Cities in the highest three
quartiles of average
exposure15
City
Charleston, WV
Chicago, IL
Colorado Springs, CO
Corpus Christi, TX
Detroit, Ml
Flint, Ml
Ft. Lauderdale, FL
Kansas City, MO
Lansing, Ml
Madison, Wl
Minneapolis, MN
New Orleans, LA
Orlando, FL
Portland, OR
Providence, Rl
Salinas, CA
San Antonio, TX
San Francisco, CA
San Jose, CA
Seattle, WA
Tacoma, WA
Vallejo, CA
Wichita, KS
Charleston, SC
Charlotte, NC
Chattanooga, TN
Cincinnati, OH
Cleveland, OH
Columbia, SC
Columbus, OH
Dallas/Ft Worth, TX
Dayton, OH
Denver, CO
Mean over
study
period
81
103
62
82
95
83
74
87
81
82
74
86
79
81
110
68
85
88
91
78
78
74
75
79
97
90
101
98
85
93
106
95
83
Max over
study
period
99
114
66
89
103
91
79
97
90
102
80
99
82
91
124
74
92
96
103
88
88
82
81
90
112
97
119
108
109
103
118
122
91
13 Jerrett et al. (2009) examined 96 MSAs; this analysis included the 94 cities that met data completeness criteria
described above, after linking monitors to MSAs (see lines 10-28, above).
14 Based on visual inspection of Figure 1 in Jerrett et al. (2009)
15 Based on visual inspection of Figure 1 in Jerrett et al. (2009)
3D-9
-------
El Paso, TX
Evansville, IN
Fresno, CA
Gary, IN
Greely, CO
Greensboro, NC
Greenville, SC
Harrisburg, PA
Houston, TX
Huntington, WV
Indianapolis, IN
Jackson, MS
Jacksonville, FL
Jersey City, NJ
Johnstown, PA
Kenosha, Wl
Knoxville, TN
Lancaster, PA
Las Vegas, NV
Lexington, KY
Little Rock, AR
Los Angeles, CA
Memphis, TN
Milwaukee, Wl
Nashville, TN
Nassau, NY
New Haven, CT
New York City, NY
Newark, NJ
Norfolk, VA
Oklahoma City, OK
Philadelphia, PA
Phoenix, AZ
Pittsburgh, PA
Portland, ME
Portsmouth, NH
Racine, Wl
Raleigh, NC
Reading, PA
Richmond, VA
Riverside, CA
Roanoke, VA
85
93
112
91
69
89
86
94
121
94
93
79
81
106
90
101
91
94
80
88
86
193
94
103
94
NA16
116
118
90
91
86
117
86
101
106
92
102
90
99
94
196
83
96
100
123
105
75
100
94
103
140
103
103
98
87
118
107
114
97
101
85
99
107
248
103
117
106
NA
136
129
105
101
93
136
96
123
117
104
124
104
114
104
245
95
' Air quality data did not meet completeness criteria described above
3D-10
-------
Rochester, NY
Sacramento, CA
San Diego, CA
Shreveport, LA
South Bend, IN
Springfield, MA
St Louis, MO
Steubenville, OH
Syracuse, NY
Tampa, FL
Toledo, OH
Trenton, NJ
Tucson, AZ
Ventura, CA
Washington, DC
Wilmington, DE
Worcester, MA
York, PA
Youngstown, OH
89
110
121
83
90
102
105
82
85
85
93
112
76
118
105
103
92
95
93
99
118
141
88
102
115
122
99
96
91
108
124
82
132
116
116
102
107
103
147
3D-11
-------
148
149
150
Table 3D-7 Long-Term Os Concentrations in 12 Urban Case Study Areas (Using the
Metric Evaluated by Jerrett et al., 2009) for Recent Air Quality and Air
Quality Adjusted to Meet Standard Levels of 75, 70, 65, and 60 ppb
Atlanta
Baltimore
Boston
Cleveland
Denver
Detroit
Houston
Los Angeles
New York City
Philadelphia
Sacramento
Saint Louis
Air Quality
Adjusted to:
Recent
75
70
65
60
Recent
75
70
65
60
Recent
75
70
65
60
Recent
75
70
65
60
Recent
75
70
65
60
Recent
75
70
65
60
Recent
75
70
65
60
Recent
75
70
65
60
Recent
75
70
65
60
Recent
75
70
65
60
Recent
75
70
65
60
Recent
75
70
65
60
2006
(Adj Yrs 2006-2008)
65
53
50
47
45
60
54
52
49
46
49
48
46
44
43
51
49
47
45
41
63
62
60
58
53
50
50
48
47
45
53
48
47
46
45
65
58
55
52
N/A
53
47
N/A
N/A
N/A
56
51
49
47
45
66
55
52
50
47
58
53
50
47
44
2007
(Adj Yrs 2006-2008)
63
52
49
46
44
59
54
51
49
46
50
49
47
45
43
52
50
48
45
41
63
61
59
58
53
54
52
50
49
46
48
46
45
44
43
61
59
56
53
N/A
54
47
N/A
N/A
N/A
59
52
50
48
46
59
50
48
46
44
58
53
51
48
45
2008
(Adj Yrs 2008-2010)
57
53
49
46
44
57
53
51
48
46
46
49
48
46
44
53
51
48
45
41
63
63
62
59
53
51
N/A
51
49
46
47
47
46
45
43
64
60
57
54
N/A
55
51
N/A
N/A
N/A
57
54
51
49
47
65
54
51
49
46
52
51
50
48
45
2009
(Adj Yrs 2008-2010)
50
47
44
42
40
52
49
48
46
44
45
45
44
43
41
49
47
45
43
40
58
58
58
56
51
48
N/A
49
47
45
47
48
47
46
44
62
60
58
54
N/A
48
47
N/A
N/A
N/A
51
49
47
45
43
61
51
49
47
44
51
50
48
46
43
2010
(Adj Yrs 2008- 2010)
56
52
49
46
44
60
55
53
50
48
49
48
48
46
44
54
51
48
45
42
60
60
58
55
50
52
N/A
52
50
47
46
46
46
45
44
57
58
56
53
N/A
55
51
N/A
N/A
N/A
58
54
52
49
47
55
48
46
44
42
55
54
52
49
46
151
3D-12
-------
152 REFERENCES
153 Bell, ML; Dominici, F. (2008). Effect modification by community characteristics on the short-term effects of ozone
154 exposure and mortality in 98 US communities. Am J Epidemiol 167: 986-997.
155 http://dx.doi.org/10.1093/aje/kwm396
156 Bell, ML; Kim, JY; Dominici, F. (2007). Potential confounding of paniculate matter on the short-term association
157 between ozone and mortality in multisite time-series studies. Environ Health Perspect 115: 1591-1595.
158 http://dx.doi.org/10.1289/ehp.10108
159 Bell, ML; Peng, RD; Dominici, F. (2006). The exposure-response curve for ozone and risk of mortality and the
160 adequacy of current ozone regulations. Environ Health Perspect 114: 532-536.
161 Cakmak, S; Dales, RE; Judek, S. (2006a). Do gender, education, and income modify the effect of air pollution gases
162 on cardiac disease? J Occup Environ Med 48: 89-94.
163 http://dx.doi.org/10.1097/01.jom.0000184878.11956.4b
164 Cakmak, S; Dales, RE; Judek, S. (2006b). Respiratory health effects of air pollution gases: Modification by
165 education and income. Arch Environ Occup Health 61: 5-10. http://dx.doi.org/10.3200/AEOH.61.L5-10
166 Dales, RE; Cakmak, S; Doiron, MS. (2006). Gaseous air pollutants and hospitalization for respiratory disease in the
167 neonatal period. Environ Health Perspect 114: 1751-1754. http://dx.doi.org/10.1289/ehp.9044
168 Franklin, M; Schwartz, J. (2008). The impact of secondary particles on the association between ambient ozone and
169 mortality. Environ Health Perspect 116: 453-458.
170 Islam, T; Berhane, K; McConnell, R; Gauderman, WJ; Avol, E; Peters, JM; Gilliland, FD. (2009). Glutathione-S-
171 transferase (GST) PI, GSTM1, exercise, ozone and asthma incidence in school children. Thorax 64: 197-
172 202. http://dx.doi.org/10.1136/thx.2008.099366
173 Islam, T; McConnell, R; Gauderman, WJ; Avol, E; Peters, JM; Gilliland, FD. (2008). Ozone, oxidant defense genes
174 and risk of asthma during adolescence. Am J Respir Crit Care Med 177: 388-395.
175 http://dx.doi.org/10.1164/rccm.200706-863OC
176 Jerrett, M; Burnett, RT; Pope, CA, III; Ito, K; Thurston, G; Krewski, D; Shi, Y; Calle, E; Thun, M. (2009). Long-
177 term ozone exposure and mortality. N Engl J Med 360: 1085-1095.
178 http://dx.doi.org/10.1056/NEJMoa0803894
179 Katsouyanni, K; Samet, JM; Anderson, HR; Atkinson, R; Le Tertre, A; Medina, S; Samoli, E; Touloumi, G;
180 Burnett, RT; Krewski, D; Ramsay, T; Dominici, F; Peng, RD; Schwartz, J; Zanobetti, A. (2009). Air
181 pollution and health: A European and North American approach (APHENA). (Research Report 142).
182 Boston, MA: Health Effects Institute.
183 Lin, S; Liu, X; Le, LH; Hwang, SA. (2008b). Chronic exposure to ambient ozone and asthma hospital admissions
184 among children. Environ Health Perspect 116: 1725-1730. http://dx.doi.org/10.1289/ehp. 11184
185 Medina-Ramon, M; Zanobetti, A; Schwartz, J. (2006). The effect of ozone and PM10 on hospital admissions for
186 pneumonia and chronic obstructive pulmonary disease: A national multicity study. Am J Epidemiol 163:
187 579-588. http://dx.doi.org/10.1093/aje/kwj078
188 Meng, YY; Rull, RP; Wilhelm, M; Lombardi, C; Balmes, J; Ritz, B. (2010). Outdoor air pollution and uncontrolled
189 asthma in the San Joaquin Valley, California. J Epidemiol Community Health 64: 142-147.
190 http://dx.doi.org/10.1136/jech.2008.083576
3D-13
-------
191 Moore, K; Neugebauer, R; Lurmann, F; Hall, J; Brajer, V; Alcorn, S; Tager, I. (2008). Ambient ozone
192 concentrations cause increased hospitalizations for asthma in children: An 18-year study in Southern
193 California. Environ Health Perspect 116: 1063-1070.
194 Salam, MT; Islam, T; Gauderman, WJ; Gilliland, FD. (2009). Roles of arginase variants, atopy, and ozone in
195 childhood asthma. J Allergy Clinlmmunol 123: 596-602. http://dx.doi.org/10.1016/jjaci.2008.12.020
196 Schwartz, J. (2005a). How sensitive is the association between ozone and daily deaths to control for temperature?
197 Am J Respir Crit Care Med 171: 627-631.
198 Schwartz, J. (2005b). Who is sensitive to extremes of temperature? A case-only analysis. Epidemiology 16: 67-72.
199
200 Stieb, DM; Szyszkowicz, M; Rowe, BH; Leech, JA. (2009). Air pollution and emergency department visits for
201 cardiac and respiratory conditions: A multi-city time-series analysis. Environ Health Global Access Sci
202 Source 8: 25. http://dx.doi.org/10.1186/1476-069X-8-25
203 Zanobetti, A; Schwartz, J. (2011). Ozone and survival in four cohorts with potentially predisposing diseases. Am J
204 Respir Crit Care Med 184: 836-841. http://dx.doi.org/10.1164/rccm.201102-0227OC
205 Zanobetti, A; Schwartz, J. (2008a). Is there adaptation in the ozone mortality relationship: A multi-city case-
206 crossover analysis. Environ Health 7: 22. http://dx.doi.org/10.1186/1476-069X-7-22
207 Zanobetti, A; Schwartz, J. (2008b). Mortality displacement in the association of ozone with mortality: An analysis
208 of 48 cities in the United States. Am J Respir Crit Care Med 177: 184-189.
209 http://dx.doi.org/10.1164/rccm.200706-823OC
210
3D-14
-------
Appendix 5A
Ozone-Sensitive Plant SpeciesA Used by Some Tribes*
*(Based on Feedback from 3 Tribes)
Common Name
(other common names)
Red alder (Oregon alder, Western alder)
Speckled alder (Tag alder, Gray alder, Hoary
alder)
Groundnut (Wild bean, American potato bean)
Spreading Dogbane (Common dogbane)
Common milkweed
New England Aster
Green ash
Twinberry
Bee-balm
Virginia creeper
Jack pine
Lodgepole pine
White pine
Black poplar (Balsam poplar)
Quaking aspen (Trembling aspen)
Black cherry
Choke cherry
Douglas fir
Allegheny blackberry (Common blackberry)
Thimbleberry
Cutleaf coneflower (Coneflower, Golden glow)
Pussy willow
Shinning willow
American elder (White elder)
Red elderberry
Sassafras
Goldenrod
Huckleberry
Wild grape
European wine grape
Scientific Name
Alnus rubra
Alnus rugosa (Alnus incana)
Apios americana
Apocynum androsamifolium
Asclepias syriaca
Aster novae-angliae
Symphyotrichum novae-angliae
Fraxinus pennsylvanica
Lonicera involucrate
Monarda didyma
Parthenocissus quinquefolia
Pinus banksiana
Pinus contorta
Pinus strobus
Populus balsamifera
trichocarpa
Populus tremuloides
Prunus serotina
Prunus virginiana
Pseudotsuga menziesii
Rubus allegheniensis
Rubus parviflorus
Rudbeckia laciniata
Salix discolor
Salix lucida
Sambucus canadensis
Sambucus racemosa
Sassafras albidum
Solidago altissima
Vaccinium membranaceum
Vitis spp.
Vitis vinifera
Confirmed bioindicator
species
Y
Y
Y
Y
Y
Y
Y
Y
Y
Y
Y
Y
Y
Y
Y
Y
Y
ASpecies included in this list are identified in one or more of the following sources:
!)SP2007(www.2.nature.nDS.20v/air/Pubs/Ddf/fla2/NPSozonesensDDFLAG06.Ddf)
2) NFS O^ Bioindicators 2006 (www.nature.nDS.sov/air/Pubs/bioindicators/index.cfm)
3) Kline et al., 2008; 4) Davis, 2007/ 2009; 5) Flagler, et al., eds., 1998
6) USDA FS FHM/FIA: Ozone Bioindicator Sampling and Estimation
(www.nrs.fs.fed.us/fia/tODics/ozone/Dubs/Ddfs/ozone%20estimation%20document.Ddf) and
Ozone Injury in West Coast Forests: 6 Years of Monitoring (2007).
5A-1
-------
This page left intentionally blank.
5A-2
-------
1 APPENDIX 6A
2
3 Calculation of Approximate Equivalent 12-hr SUM06 and 12-hr W126
4
5
6 SOURCE: 2007 Staff Paper, Appendix 7B (U.S. EPA, 2007).
1 Despite various metrics reported in the vegetation effects literature, there is no standard
8 method for calculating equivalent levels between metrics. The maximum 3-month 12-hr
9 SUM06 of 25 ppm-hr secondary standard that was proposed in the last review (62 FR
10 38877) was based on a yield loss prevention of approximately 10% in 50% of crop cases
11 studied in the National Crop Loss Analysis Network (NCLAN) experiments. For
12 consistency, staff judged it appropriate to use the NCLAN experiments to derive
13 equivalents between the 12-hr SUM06 and W126. For example, below are the 12-hr
14 SUM06 and W126 NCLAN equations to protect 50% of crop cases from a specified
15 percent yield loss (Lee and Hogsett 1996):
16
Metric
12-hr SUM06
1 2-hr W 126
Weibull Equation
Predicted Relative Yield Loss = 1- exp(-[SUM06/87.42]A1.82)
Predicted Relative Yield Loss = 1- exp(-[W126/96.05]A1.48)
17
18 In the first equation, solving for a SUM06 of 25 ppm-hr equals a predicted relative yield
19 loss of 10%. Solving the second equation for a 10% yield loss equals a W126 of 21 ppmhr.
20 Thus, staff considers a 12-hr SUM06 of 25 ppm-hr and a 12-hr W126 of 21 ppm-hr
21 approximately equivalent.
22
23 REFERENCES
24
25 Lee, E. H.; Hogsett, W. E. (1996) Methodology for calculating inputs for ozone secondary standard benefits
26 analysis: part II. Report prepared for Office of Air Quality Planning and Standards, Air Quality Strategies
27 and Standards Division, U.S. Environmental Protection Agency, Research Triangle Park, N.C., March.
28 U.S. EPA (U.S. Environmental Protection Agency). (2007). Review of the national ambient air quality standards for
29 ozone: Policy assessment of scientific and technical information: OAQPS staff paper [EPA Report].
30 (EPA/452/R-07/003). Research Triangle Park, NC.
31 http://www.epa.gov/ttn/naaqs/standards/ozone/data/2007_0 l_ozone_staff_paper
32
33
6A-1
-------
United States Office of Air Quality Planning and Standards Publication No. EPA-452/P-14-002
Environmental Protection Health and Environmental Impacts Division January 2014
Agency Research Triangle Park, NC
------- |