Interim Regulatory Impact Analysis for the
PM2.5 National Ambient Air Quality Standards

US Environmental Protection Agency
Office of Air Quality, Planning and Standards

January 17, 2006


-------
Executive Summary

Overview

This interim Regulatory Impact Analysis (RIA) summarizes our analysis to date of the
monetized human health benefits and control costs associated with meeting revised standards for
fine particles (PM2.5) that were proposed by EPA on December 20, 2005, and several alternatives
that are more and less stringent than the proposal. EPA also proposed revisions that would
replace the current annual and daily PMi0 standards with a new daily standard for thoracic coarse
particles (PM10.2.5). For reasons outlined more fully in Chapter 1, this RIA does not contain an
analyses of these proposed revisions. In general, EPA expects that significantly fewer areas
would violate these proposed standards as compared to the current PM10 standards. Due to
data limitations and analytical issues, this analysis focuses on five urban areas. The RIA
accompanying the final decision will include a national assessment. The analyses summarized
in this chapter nevertheless provide a national overview of the effectiveness of current programs
in attaining the alternative standards as well as implementation insights for illustrative control
strategies to attain each alternative in the five selected areas. This summary provides an
overview of the key data and preliminary conclusions of the RIA, including:

•	Future-year predictions of PM2.5 concentrations and attainment under current, proposed
and alternative standards

•	The nature of the PM2.5 air quality problem

•	Important limitations and uncertainties in our analysis that precluded an estimation of
national benefits and costs

•	Provisional conclusions from these interim analyses

Future-year predictions of PM2.5 Concentrations and Attainment Under Alternative
Standards

Alternative PM2.5 NAAQS analyzed

The December 20, 2005 preamble to the proposed rule provides the rationale for EPA's proposed
revisions to the primary PM2.5 NAAQS and as well as other alternatives on which the Agency is
requesting comment. In our analyses, we have selected a subset of options designed to
encompass the range of alternative standards upon which the Agency is requesting comment.

This analysis focused on both the incremental national air quality of the proposed and alternative
standards, and the incremental costs and benefits in five urban areas, as compared to a regulatory
base case and implementing the current standards as of 2015. The alternatives analyzed are
summarized in the following table as combinations of the annual and daily PM2.5 standards:

ES-1


-------
Table 1: Annual and Daily PM2.sNAAQS Considered in This Analysis

Combination of Annual and 98th
percentile Daily Values, in pig/m3

Notes

15/65
15/40
15/35
14/35

Proposed revision
Alternative for comment

Current standards

Alternative for comment

15/30

Alternative for comment

Overview of Air Quality Modeling Methodology

As a first step in the national assessment of alternatives, the analysis forecast emissions and air
quality in 2015 under a regulatory base case that incorporates national, regional, state and local
regulations that are already promulgated and/or adopted. This base case does not forecast
actions states may take to implement the existing PM25 standards. The regulatory base case
includes recent rules that affect the PM-related emissions from the power generation sector and a
number of mobile source categories including the Clean Air Interstate Rule (CAIR), the Clean
Air Mercury Rule (CAMR), and the Clean Air Visibility Rule (CAVR which also affects some
industrial boiler emissions), and national mobile source rules for light and heavy-duty vehicles
and non-road mobile sources. Current state programs that address these and other source
categories that were on the books as of early 2005 are also modeled for future years. Based on
the emissions forecasts, EPA developed annual and daily PM25 projections using the CMAQ
model.1

Summary of Attainment Analyses

Table 2 summarizes the results of these analyses in terms of the projected numbers of counties
with monitors that would and would not attain the standards alternatives under the same
regulatory base case for 2015. This is not a prediction of the air quality EPA would expect to
occur in this future year because the baseline analyzed contains only current programs but not
the additional reductions that will be made in response to State Implementation Plans (SIPs)
designed to meet the current PM25 or 8-hour ozone NAAQS. The PM2.5 SIPs are due in April
2008 and the ozone SIPS are due in June 2007. The Clean Air Act presumptively requires each
area to attain the current PM2.5 standards within 5 years of designation, by 2010, with authority
for EPA to grant a state an attainment date extension of up to an additional 5 years for specific
areas.

This regulatory base case scenario analysis suggests that EPA's recently promulgated national
rules, in combination with existing state and local programs will make significant contributions
to reducing projected PM2 5 nonattainment in the eastern US under any of the standards
alternatives analyzed, as compared to current air quality levels. EPA modeling indicates that by

1 The methodologies for forecasting emissions and air quality and associated uncertainties are detailed in the
Technical Support Document - "Air Quality Modeling Technique used for Multi-Pollutant Analysis"
(http://www.epa.gov/airmarkets/mp/aqsupport/airquality.pdf). The methodology used to derive the 98th percentile
24-hour values is summarized in Appendix E of this RIA.

ES-2


-------
2015, 84 counties with monitors will attain the existing PM2.5 standards out of the 116 such
counties currently out of attainment just based on regulatory programs already in place. In
addition, all areas in the eastern United States will have lower PM2.5 concentrations in 2015
relative to present-day conditions. In most cases, the predicted improvement in PM2.5 ranges
from 10% to 20%.

Table 2. Summary of Projected County Attainment and Nonattainment Counts: Projected
2010 and 2015*

Projected with Regulatory Base Case

2010	2015

Standard Alternative

(annual/daily in ftg/m3)	National	East	West	National	East	West

15/65—

Attain**

77

75

2

84

84

0

current
standard

Non-AUain

39

27

12

32

18

14



Attain**

81



6

90

84

6

15/40

Non-Attain

"

27

30

48

18

30



Attain**

102

98

4

1 15

1 1 1

4

15/35—
Proposed

Non-Attain

89

43

46

76

30

46



Attain**

125

121

4

139

135

4

14/35

Non-Attain

1 10

64

46

96

50

46



Attain**

129

129

0

148

148

0

15/30

Non-Attain

197

135

62

178

1 16

62

*See Appendix E for details on projection method used here (i.e., Speciated Modeled Attainment Test-SMAT).
There are some counties which may have complete ambient data for 24-hour standard, but incomplete data for the
annual standard. These counties were not included in this analysis.

**These are counties with monitors that reported concentrations above the respective NAAQS alternative levels
based on 2002-2004 data that are projected to attain the alternative in the forecast years noted.

Chapter 2 presents a series of maps with more specific details of the current, proposed and
alternative PM2.5 NAAQS attainment analyses results. Figure 1 below summarizes projected air
quality and attainment status under each of the individual annual (14 and 15 |ig/m3) and daily
(30, 35, 40, and 65 |ig/m3) standards considered in this analysis.

The major insights from this national scale analysis of alternatives include the following:

• As compared to the current standards, the proposed tighter daily standard of 35 ug/m3
appears to have a bigger impact in the West than in the East, particularly after the
forecast regulatory base case controls are more fully implemented. Most of the eastern
counties that would not attain the standard in 2015 are part of nonattainment areas that
are required to adopt further controls under the current standards. The increment above
the daily standard is generally less than 5 |ig/m3.

ES-3


-------
•	Southern and central California, which have a number of counties that violate the current
daily standard, have increments in the range of 20 to 48 ug/m3 above the proposed daily
standard.

•	Most of the counties that would not attain the proposed daily standard in the northwestern
quadrant of the US currently attain the annual and 24-hour NAAQS. These areas have
lower annual averages, but can have high daily peaks during the winter months related to
meteorological inversions and increased wood combustion emissions. The increment
above the daily standard varies from 3 to 7 |ig/m3 in this region.

ES-4


-------
Figure 1: Counties Exceeding Current and Alterative Annual and 24-hour PM2.s
NAAQS under Projected 2015 Regulatory Base Case

Exceeds Daily Standard of 65
Exceeds Daily Standard of 35
Exceeds Daily Standard of 40
Exceeds Daily Standard of 30

ES-5


-------
Nature of the PM2.5 Air Quality Problem

Particulate matter (PM) is a highly complex mixture of solid particles and liquid droplets that
occur in the atmosphere together with numerous pollutant gases that interact with them.

Particles span many sizes and shapes and consist of thousands of different chemicals. Particles
are emitted directly from sources and also are formed through atmospheric chemical reactions
and are designated as 'primary' and 'secondary' particles, respectively. Particle pollution varies
by season and location and is strongly affected by day to day variations in meteorology, such as
temperature, stagnation, wind, cloud cover, and humidity. Daily PM2.5 values at some locations
can be high at any time of the year. For example in Seattle, which has low annual values, the
highest levels of PM2.5 concentrations occur during the winter months and are composed of
carbon particles associated with wood and waste burning. By contrast, many areas in the eastern
US have higher annual values with the highest daily values in the summer.

Because of their size and formation mechanisms, fine particles can be transported hundreds to
thousands of miles from emissions sources. For this reason, fine particle concentrations in a
particular area may have a substantial contribution from regional transport as well as local
sources. The major PM2 5 components, or species, are various elemental and organic
carbonaceous compounds, sulfate and nitrate compounds, and crustal/metallic materials such as
soil and ash.. Primary PM2.5, that is matter which is originally emitted in particulate form,
consists of carbonaceous material (e.g. soot)—emitted from cars, trucks, heavy equipment, forest
fires, and burning waste, as well as from coke ovens, metals from combustion and industrial
processes, with some small contribution from crustal materials. Secondary PM2 5 forms in the
atmosphere from precursor gases including sulfur and nitrogen oxides from power, industrial and
other combustion and process sources, certain reactive organic gases from diesel and other
mobile sources, solvents, fires, and biogenic sources such as trees, and ammonia from
agricultural operations, natural, and other sources.

The chemical makeup of particles varies across the United States (Figure 2). For example, fine
particles in the eastern half of the United States contain more sulfates than those in the West,
while fine particles in southern California contain more nitrates than other areas of the country.
Carbon is a substantial component of fine particles everywhere. Note that particle mass and
composition can vary substantially by season, so annual averages should not be considered
representative of specific high PM2.5 days. These averages include both local and regional
transport contributions to PM2.5 based on recent data. Figure 3 focuses on estimates of the
relative contribution of various components from local sources. Sources of carbonaceous
particles appear to be the most important local contributors to fine particles in all of the urban
areas shown.

ES-6


-------
WEST

EAST

Northwest

(3

Southern
California

©

Upper
Midwest

©

Southwest

Industrial
Midwest

©

Southeast
©

Northeast

0

Sulfates
Nitrates
Carbon
Crustal

Circle size corresponds

Figure 2. Average PM2 5 composition in urban areas by region, 2003,

i.

Fresrto

nj

Nfesoula

I

SLC

SullalG:
_ a ~

0,0 0.4 0,9
Ammonium:

_ ¦ I

0.0 0.9 1.9

Nitrate;

I

0.4 3.5 6.5

Total Carbon Mass
(TCM) (k=i.S):

2.9 8,1	132

Crucial"

_ -	¦

0,0 0,4	O B

JL

Si. Louis

j j Cleveland M Bran*

¦E

Tulsa

Bailimofe

R^nownd

Birmingham

ID

Aflanla

J."

Chartone

Figure 3. Estimated 'urban excess' of 13 urban areas by PM2.S species component 2003. The urban
excess is estimated by subtracting the measured PM2 5 species at a regional monitor location
(assumed to be representative of regional background) from those measured at an urban location.2

2 The light grey in this bar graph is organic carbon and the dark grey is elemental carbon. Total Carbon Mass (TCM)
is the sum of Organic Carbon (OC) and Elemental Carbon (EC).

ES-7


-------
Limitations and Uncertainties in this RIA that Precluded a National Assessment

In developing this RIA, we planned to provide national cost and benefit estimates of illustrative
control strategies to assess the nation's ability to reach the proposed PM2.5 standards and
alternative standards options. As we developed that analysis, we reached the conclusion that, for
the proposed rulemaking, our available data and tools are insufficient to develop cost and benefit
information that would accurately reflect the range of possible options that the States may choose
to implement.

The most significant data limitation we encountered is related to our controls and emissions
inventory databases. In developing air quality modeling scenarios, we discovered significant
limitations in both the number and effectiveness of control technologies for which adequate cost
information is available. Further, existing emissions inventories appear to understate the extent
of mobile source direct PM emissions and provide limited information on the extent of available
controls that have already been applied to some source categories. These limitations prevented us
from performing a national analysis. The scope of our air quality modeling was also constrained
by both the time in which we had to complete the analysis and the breadth of controls available
to evaluate in the model runs on a national basis. For this reason, we chose to perform air quality
modeling on an urban scale in 5 areas by using a screening-level air quality model that we
describe below.

Thus, we concluded that the national-scale analysis based on our current data and tools would
not properly reflect the incremental costs and benefits of moving from the current standards to
progressively more health-protective standards. We are taking steps to ensure that we will
complete this national-scale analysis in time for publication with the final rule (September 2006).

Overview of the Five City Analysis and Provisional Conclusions

Analytical Approach

In this RIA we selected a handful of urban areas in which to estimate control costs and
monetized human health benefits. Our selection of these urban areas was greatly influenced by
the air quality model that we utilized to perform the analysis. The Response Surface Model
(RSM) discussed in Chapter 2 can analyze air quality changes resulting from the application of
both local and regional controls within nine pre-selected urban areas. These nine urban areas
represent areas within the air quality modeling domain for which we could analyze control
strategies without such controls affecting other RSM urban areas. The air quality model
estimated that six of these urban areas would be out of attainment for some level of the annual
and daily standard under analysis; these areas include Atlanta, New York/Philadelphia, Chicago,
Seattle, San Joaquin and Salt Lake City, Utah.3 The latter area was excluded from the detailed
analysis because of an inadequate specification of the initial modeling domain. Although limited,
these areas do reflect a span of different nonattainment/source/composition characteristics across
the US (e.g. Figure 2), including areas with high and low regional backgrounds, and different

3 These nine urban areas include Seattle WA, San Joaquin CA, Phoenix AZ, Denver CO, Dallas TX, Chicago IL,
New York/Philadelphia NY/PA, Atlanta GA and Salt Lake City UT. We excluded Phoenix, Denver and Dallas
because they did not violate any of the standard alternatives we analyzed.

ES-8


-------
local source characteristics ranging from midwestern industry to woodsmoke and agricultural
sources. Table 3 summarizes some of the key characteristics of these 5 cities.

Table 3: Summary of Current and 2015 Future Projected Emissions and Air Quality Information Across
Modeled Urban Areas

New York/

	Atlanta Chicago Philadelphia} Seattle San Joaquin

Annual Mean

17.5

16

16.8/15.4

10.5

21.7

98th Percentile Daily Average

39

43

42/38

41

62

2015 Basecase Desien Value
Annual Mean

16.4

16.9

14.3 / 14.6

10.5

26.1

98th Percentile Daily Average

35

39

34/36

39

83

2015 Basecase Precursor Emissions (thousands of tons)









NOx

128

282

617

74.5

161

S02

89.7

286

341

7.4

15.8

VOC

156

309

674

90.8

111

NH3

26

24.2

67.6

10.5

156

Primary Organic and Primary Elemental
Carbon

15

22.7

43.3

4.8

9.1

2015 Percentage of Emissions Reduced After Application of All Effective Controls in AirControlNet*



NOx**







10.3%

13.6%

S02

66.1%

48.9%

16.5%

12%

20%

voc**

NH3**

Primary Organic and Primary Elemental
Carbon

23.6%

34%

31.8%

34.4%

9.1%

¦"Calculations do not reflect EGU NOx and S02 controls in Eastern U.S. urban areas (Atlanta, Chicago, NY/Philadelphia)

**We chose not to include NOx controls in the eastern U.S. and VOC and NH3 controls in either the eastern or western U.S.

***2015 future projected design values were based on the 1999-2003 five year weighted average concentrations.

/New York/Philadelphia 2002-2004 design values and 2015 base design values are based New York Co. and Philadelphia Co.,

respectively.

The analysis of the 5 cities used a reduced form air quality model called the Response Surface
Model (RSM). The RSM is a screening-level air quality modeling tool built from a complex
design of photochemical model simulations using the CMAQ Modeling System. Using this set
of air quality model simulations permits a quick assessment of the estimated air quality changes
at monitored locations (and elsewhere) throughout the United States for any combination of
emissions reductions within a range of 10 to 120 percent of baseline emissions for a set of 12
source/emission factors. The RSM was used in the interim RIA to identify effective emissions
reductions strategies (e.g. comparison across sectors; comparison across pollutants) and to
develop (in combination with cost and effectiveness information) appropriate area specific
combinations of controls to attain standards.

ES-9


-------
The details of these analyses are presented in Appendix A. Chapter 3 summarizes illustrative
estimates of control costs and the monetized human health benefits to simulating attainment, or
near-attainment, with each of the standard alternatives in the five urban areas in 2015. In
developing insights and conclusions from these results, there are several important strengths,
limitations and uncertainties that apply to our air quality modeling, controls analysis and benefits
assessment that are important to note. Collectively, these limitations argue against placing
significant weight on the specific quantitative estimates presented. The following sections
summarize the key issues and uncertainties in each of these general categories of the analysis.

Emissions Forecasting and Air Quality Modeling Uncertainties

Chapters 2 and 3 summarize some of the key uncertainties associated with forecasting emissions
and modeling air quality for the multiple pollutants that contribute to ambient fine particle
concentrations. While EPA's regional scale air quality modeling system has been extensively
peer reviewed and represents the state of the science in terms of the formation and fate of PM2.5
in the atmosphere, a number of factors affect the conclusions that can be reached about the
effectiveness, costs, and benefits of alternative control strategies in the five city analyses:

•	Overall, the air quality model performs well in predicting monthly to seasonal
concentrations, similar to other recent model applications for PM2.5 The model is less
well suited to predicting 24-hour values.

•	In general, model performance is better for the eastern U.S. than for the West. The air
quality model performs well in predicting the formation of sulfates, which are the
dominant species in the East. It does not perform as well for nitrates and secondary
organic particles from anthropogenic and natural sources.

•	A number of uncertainties arise from use of baseline data from EPA's National
Emissions Inventory, especially in terms of the overall magnitude of emissions of
primary particles from stationary and mobile sources, spatial allocation of area and other
source categories, and the relative split of emissions into PM2.5 species. Of particular
concern is the apparent disparity between estimated contributions of mobile source
emissions with receptor modeling results based on ambient air quality data. These
comparisons suggest that our base emissions inventory significantly underestimates the
emissions of mobile sources. In addition, the RSM system does not include primary
emissions of metals or related inorganic emissions from industrial processes or
combustion. This limits control options for primary particles to carbonaceous emissions.

•	Additional uncertainty is introduced through our future year projections of emissions due
to unrefined growth rates and limited information on the effectiveness of control
programs.

•	The RSM based air quality modeling likely understates the effectiveness of urban-area
controls. The CMAQ air quality model that provides the basis for the RSM uses a 36
kilometer receptor grid, which effectively spreads point and mobile source emissions that
may be concentrated in particular locations across a wide area. The serves to obscure

ES-10


-------
local-scale air quality improvements that result from urban-area controls.4 To the extent
that this occurs, our estimates may underestimate the effectiveness of local or urban-area
controls relative as compared to broad scale regional controls.

Cost and Emissions Uncertainties

The limitations in our control strategy technology and cost data noted above also affect the five
city analyses. As discussed more fully in the RIA and appendix, a number of approximations and
assumptions were required to complete the analysis for all of the standards alternatives analyzed.
The more important of these include:

•	Progress attainable through controls known to be available is underestimated. The
analysis does not consider all known control measures, and as a result understates the
emissions reductions and progress toward attainment that can be achieved through known
measures.

•	Attainment cost estimates are highly dependent on costs of measures not currently in
EPA's database. In part due to the database limitations discussed above, the analysis of
the costs of meeting the current standards and more stringent alternatives rely on
innovative and emerging controls with derived costs. Many emission controls employed
to meet the more stringent standards are based on unspecified measures with assumed
costs. Therefore the incremental attainment cost estimates for more stringent standards,
and any cost-benefit comparisons, are subject to an unusually high degree of uncertainty.

•	Analysis assumes attainment of new standards within 5 years. Although subpart 1 of Part
D of the Clean Air Act allows nonattainment areas to qualify for an extension of up to 10
years from designation for an area to attain, the analysis for simplicity assumes that all
areas must attain within 5 years (i.e., in 2015). This assumption tends to overestimate
costs associated with attainment for areas qualifying for an extension (to 2020) because
federal programs (e.g., on-road and non-road vehicle and engine standards and the Clean
Air Interstate Rule) achieve greater emissions reductions over time, so that most areas
become cleaner in the base case beyond 2015. Based on current information, it does not
appear possible to attain the proposed NAAQS in the San Joaquin area by 2015.

Benefits Uncertainty

The benefits estimates generated for this proposal RIA are subject to a number of assumptions
and uncertainties, which are discussed throughout the document. For example, key assumptions
underlying the primary estimate for the mortality category include the following:

4 This is illustrated in figure 2-26, in chapter 2, which displays the geographical distribution of the results of the
local scale modeling within a 36-kilometer CMAQ grid cell.

ES-11


-------
1. Inhalation of fine particles is causally associated with premature death at
concentrations experienced by many Americans on a regular basis. Although
biological mechanisms for this effect have not yet been completely established,
the weight of the available epidemiological and experimental evidence supports
an assumption of causality

2. The analysis also assumes that all components of fine particles have equal

toxicity. While it is reasonable to expect that the potency of components may vary
across the numerous effect categories associated with particulate matter, EPA's
interpretation of scientific information considered to date is that such information
does not yet provide a basis for quantification beyond using fine particle mass.
While EPA has not performed formal sensitivity analysis of this assumption in its
analysis for the proposed PM NAAQS RIA, the Agency is exploring ways to
present the importance of this assumption in estimating benefits and its
implications for control strategy development and assessment as a part of the
analysis for the final RIA.

3. One source of uncertainty that has received recent attention from several scientific
review panels is the shape of the concentration-response function for PM-related
mortality, and specifically whether there exists a threshold below which there
would be no benefit to further reductions in PM25. That is, the hypothesized
relationship includes the possibility that there exists a PM concentration level
below which further reductions no longer yield premature mortality reduction
benefits. To consider the impact of a threshold in the response function for the
chronic mortality endpoint on the primary benefits estimates, we constructed a
sensitivity analysis by assigning different cutpoints below which changes in PM2.5
are assumed to have no impact on premature mortality.

Conclusions

While the results of the analyses presented in this RIA must be interpreted within the context of
uncertainties and limitations summarized above, we believe that the following conclusions are
appropriate:

•	Recently promulgated regional and national programs will make significant progress in
reducing daily as well as annual PM2.5 by 2015 under the current and proposed NAAQS
as well as the alternatives.

•	Current standards can be met in all areas analyzed with no additional controls beyond the
current regulatory base case programs (2 areas) or with the addition of controls on local
sources.

•	The proposed new daily NAAQS would be met in 2 of the 3 eastern areas through
programs designed to meet the current annual NAAQS. The proposed daily NAAQS
can be met with local controls in Seattle and New York/Philadelphia. Based on current
information, it does not appear possible to attain the proposed NAAQS in the San
Joaquin area by 2015. They would likely need to consider a combination of intrastate

ES-12


-------
regional and technology forcing local controls appear to be necessary to attain by 2020 or
beyond.

•	Based on the current analyses, it appears that the more stringent annual and daily
alternatives (14 |ig/m3 or 30 |ig/m3) would drive States to consider additional regional
reductions in the Eastern US, as well as new intrastate regional reductions in the West.
Because the limitations of the analyses likely understate the cost-effectiveness of existing
and new controls on local sources, the point at which incremental regional controls would
become necessary or significantly more cost effective is not clear because the limitations
of the analyses may understate the cost-effectiveness of existing and new controls on
local sources.

•	Within the context of the limitations of the analysis, costs and benefits of the proposed
NAAQS and alternatives are generally within the same order of magnitude. Given the
uncertainties and limitations, no general conclusions are possible with respect to the most
optimal approach.

ES-13


-------