Integration of Regional and Local Scale Air Quality Modeling Research With EPA ORD’s Human Exposure and Health Research Program


WHITE PAPER:

INTEGRATION OF REGIONAL- AND LOCAL-SCALE AIR

QUALITY MODELING RESEARCH WITH EPA/ORD'S
HUMAN EXPOSURE AND HEALTH RESEARCH PROGRAM

(8 October 2010)

Atmospheric Exposure Integration Branch of the Atmospheric Modeling and

Analysis Division

Atmospheric Modeling and Analysis Division
National Exposure Research Laboratory
Office of Research and Development
U.S. Environmental Protection Agency
Research Triangle Park, NC 27711

1

-------
1.	Introduction

As part of the EPA Office of Research and Development (ORD), the National Exposure
Research Laboratory (NERL) is in the process of developing Integrated Multi-disciplinary
research programs to better address Problems of Broad, National Significance (PoBNS). These
are high priority topics that would benefit from more integrated collaborative research
implementation across ORD's Laboratories and Centers. Chemical contaminants related to
human health have been identified as a PoBNS. Understanding the magnitude and nature of
human exposure is clearly the first step in assessing the likelihood of adverse health effects that
could result from contact with environmental pollutants. To help identify air pollutant sources of
greatest risk to humans, integration of modeling tools is essential to estimating air concentrations
and exposures for a population of interest. The coupling and appropriate application of these
models will improve estimates, demonstrate utility in environmental health accountability
programs, assist in the development of risk mitigation strategies, and improve community health
or epidemiology studies. This vision is consistent with the EPA/NERL exposure science
framework and its emphasis on cross-disciplinary integration. The vision is also consistent with
the EPA/ORD research emphasis on integrated multi-disciplinary research.

This white paper addresses progress that can be made towards this vision during the next 5 years
by (1) describing why is this research important, (2) describing the key air quality-exposure
science questions relevant to the analyses of human health effects, (3) discussing EPA/NERL's
research program to improve exposure assessments and associated health studies with refined air
quality modeling tools, (4) identifying necessary collaborations, and (5) outlining the research
outcomes and products to be produced by this research program. The following sections of this
white paper provide a description of the Regional and Local-scale Air Quality Model Connection
for Human Exposure and Health Research in terms of answering some of the key science policy
and programmatic concerns.

2.	Why this research is important?

The role of exposure science in risk assessment and risk management is critical to the mission of
the EPA. A core objective of the Clean Air Act is to "protect and enhance the quality of the
Nation's air resources so as to promote the public health and welfare and the productive capacity

2

-------
of its population." The EPA evaluates the risk of threat to its air resources through the risk
assessment and risk management processes. More specifically, EPA conducts exposure
assessments to determine the route, magnitude, frequency, and distribution of exposure (US
EPA, 2009). Epidemiology studies are vital in estimating the risk of exposure to ambient
pollutants and the impact resulting from this exposure on human health. Current research in
epidemiology points out the need for improved air quality estimates.

Background

There is a long history of epidemiologic research on the relationship between spatial and
temporal variability in ambient concentrations (and exposure concentrations) and adverse human
health effects. The adverse health effects include acute and chronic effects of both particulate
and various gaseous air pollutants on morbidity and mortality. Morbidity effects of air pollution
range from decreased lung function to exacerbation of symptoms of asthma and chronic
bronchitis to more serious cardiopulmonary events, such as increased hospitalizations, and
myocardial infarctions.

There are varying research approaches in epidemiology for studying the effect of ambient air
pollution. These approaches vary by health data types and study design. The health data types
include group/population level (e.g., administrative data) and individual level (e.g., cohort or
panel data). The epidemiologic study designs can be classified as either cross-sectional,
longitudinal (time-series), or as a hybrid of the two. Cross-sectional studies observe a population
at a single point in time, whereas longitudinal studies observe a population at several points over
an extended period. A cohort study is a type of longitudinal study tracking a specific group of
people through time; a panel study selects different people from a common population at each
time point. In general the evidence for an exposure/response relationship from a cohort study is
considered to be of higher quality than from other observational designs. The requirements for
air quality data are different for each of these designs, and they vary in their temporal and spatial
resolution.

There are many examples of Local Scale Cohort Studies. Among these are:

-------
-	The Harvard Six Cities Study (Dockery et al. 1993) is a long-term, longitudinal cohort
study of the health effects (pulmonary symptoms and lung function) associated with
ambient pollutant levels across six cities.

-	The American Cancer Society (ACS) Study (Pope et al. 1995) examined mortality and
particulate air pollution across 151 US metropolitan areas for which sulfate data were
available (1980 and 1981), and 50 metropolitan areas for which fine particle data were
available (1979-1983).

-	The Women's Health Initiative (WHI) included an evaluation of data from 65,893
postmenopausal women without previous cardiovascular disease in 36 US metropolitan
areas from 1994 to 1998 for associations with long-term exposure to fine particulate
matter (Miller et al. 2007).

-	The Multi-Ethnic Study of Atherosclerosis and Air Pollution (MESA Air) implements an
extensive exposure assessment program to characterize long-term average concentrations
of ambient fine particulate matter and its species and copollutants with a particular
emphasis on capturing concentration gradients within cities.

-	The NERL Longitudinal PM Panel Studies (Baltimore MD, RTP NC) investigated
relationships between personal exposures, residential indoor, and ambient concentrations
of PM2.5 and associated gaseous co-pollutants (Wallace et al. 2004).

Some examples of Cross-Sectional or Local-Scale Hybrid Studies include:

-	Cross-sectional mortality across different metropolitan statistical areas in the US (e.g.,
Ozkaynak and Thurston 1987, Pope et al. 2009, Miller et al. 2007)

-	Hospital admissions/Medicare cohort studies (Dominici et al., 2006)

-	Hybrid time-series and multicity studies (e.g. NMAPS by the Hopkins group)

-	Intra-urban studies (e,g., Jerrett/Krewski et al 2005, or Miller et al 2007)

These health studies have demonstrated the need for accurate data on temporal and spatial
variations in ambient concentrations (or indicators of exposures). There are many key findings
from these studies:

-	There is increasing evidence that the existing monitoring network is not capturing the
sharp gradients in exposure due to high concentrations near significant sources of
pollutants (e.g., major roadways); many pollutants are not simultaneously monitored.

4

-------
- A number of publications have already raised the issue of exposure prediction errors in
the interpretation of time-series epidemiology studies of PM mortality or morbidity
effects (Zeger et al. 2000; Dominici et al. 2000; Zeka and Schwartz 2004).

- More recent studies (Sarnat et al. 2006 and Jerrett et al. 2005) have shown that more
narrowly defining the geographic domain of the study populations and improvements in
the estimates of the corresponding PM ambient concentrations lead to stronger
associations between ambient pollutant concentrations and hospital admissions and
mortality records.

- The impact of exposure misclassification or exposure prediction errors on the outcome of
air pollution epidemiology studies varies, of course, depending on the particular study
design. In general, the finer the spatial scales used in the epidemiologic studies, the
greater the likelihood that exposure prediction errors have a significant impact on the
outcome of ambient air pollution epidemiologic study results. However, the uncertainty
in estimating pollutant levels at finer scales can be quite large.

New development in environmental epidemiology and exposure research
Epidemiology studies have traditionally relied upon imperfect surrogates of personal exposures,
such as area-wide ambient air pollution levels based on readily available outdoor concentrations
from a central monitoring site. This approach has resulted in some major technical challenges.
For example, epidemiologic studies typically use ambient monitoring data collected for
regulatory purposes or for selected short duration research studies for PM, ozone and other
gaseous criteria pollutants (NO2, CO, SO2). These studies assume that a central site monitor, or
average of a few monitors, is representative of the complex spatial and temporal patterns of air
quality within a large urban area and reflects the levels found in areas where sensitive and
disadvantaged populations are located. For many pollutants (e.g., toxic pollutants), ambient
monitoring data are often non-existent or limited. In addition, many of these pollutants exhibit
large concentration gradients and temporal variability, especially near large emitters such as
major roadways, factories, ports, etc.

Refined estimates of exposure to ambient pollutants are necessary when exposures to stressors
with significant spatial and temporal variability occur or when cumulative risks from exposures
to multiple stressors exist (US EPA, 2009). In those cases where exposures are relatively

-------
homogeneous in the atmosphere (e.g., fine particulate sulfate), simpler characterizations of
exposures may be used. Researchers are investigating the use of improved air quality
concentration fields as surrogates for exposure when such characterization is appropriate (US
EPA, 2009). Such methods apply dispersion and photochemical grid-based air quality models,
and hybrid approaches that combine modeled and observed air quality data to improve estimates
of ambient pollutant concentrations (see Figure 1). More refined approaches are currently being
used by epidemiologists to enhance the spatial resolution of monitoring data such as GIS-based
interpolation methods that can approximate outdoor concentrations near communities. These
Land-use Regression (LUR) models incorporate landscape characteristics, such as proximity to
roadways and other outdoor sources of air pollution to capture small-scale intra-urban variability
(English, 1999; Jerrett, 2005; Smith, 2006; Arain, 2007; Marshall, 2008). LUR modellers face
challenges in applying their models in areas with sparse monitoring data and when extrapolating
the use of these m odels to study areas with different land-use and topography (Jerrett, 2005), and
estimating exposures for future emissions scenarios.

Exposure Information Used in Health Effects Studies

Tiers of Exposure Metrics

( Ambient Monitoring Data }

Land-Use Regression
Modeling

Air Quality Modeling
(CMAQ, AERMOD, hybrid)

Statistical modeling
(Data blending)

Combined
Air Quality / Exposure
Modeling

Input data

Monitoring Data

Monitoring Data

Emissions Data

Land-Use/Topography

Emissions Data

Meteorological Data

Land-Use/Topography

Monitoring Data

Emissions Data

Meteorological Data

Land-Use/Topography

Monitoring Data

Emissions Data

Meteorological Data

Land-Use/Topography

Personal Behavior/Time Activity
Microenvironmental Characteristics

\

, Health data analysis

Epidemiological statistical models:

logfECyj) = a + ^Ambient Pollution^ + ZkYkareaM+ ...other covariates

Figure 1. Various approaches to estimate exposures in support of environmental health studies.

6

-------
Some researchers are now beginning to use fine-scale air quality/exposure modeling to improve
exposure estimates in support of environmental health studies. An important advantage of this
combination of an air quality and exposure model is to account for exposure to indoor and
outdoor sources, in the same manner that the personal monitoring data can (see Figure 2).
Among these studies are: Burke, 2001; Ozkaynak, 2008; Isakov, 2009; Isakov and Ozkaynak,
2008; Isakov, 2006; Georgopoulos, 2005; Wang, 2009; Jimenez-Guerrero, 2007; Levy, 2002;
Hoffmann, 2007. Further advantages of this new approach are: (1) the improvement of the
spatial resolution of ambient outdoor concentrations by accounting for local-scale influences on
dispersion with a physics-based approach; and (2) providing concentration estimates that can
span several years (up to decades) and can match the period of epidemiological data collection.
These longer periods of records would provide more robust statistical results.

Linking Air Quality Models to Exposure Models

Input from
AQ Model

•	Modeled
ambient conc.
at census tract
centroids

•	using

-	Emissions

-	Meteorology

Ambient
Concentrations

Input
Databases

Census

Human Activity
Food Residues
¦ Recipe/Food
Diary

Exposure
Factor
Distributions

Algorithms

IkJ



LiJ

!~!

Calculate
Individual
Exposure/Dose
Profile

Inhalation

lo TIME li

Ingestion Dermal

a: :

\\ !



iA A J

Exposure
Model Output

Population
Exposure

Population
Dose

Figure 2. Combined air quality/exposure modeling in support of environmental health studies.

The use of fine-scale air quality modeling to support environmental health studies is particularly
important since individuals spend time in different micro-environments during the day (e.g.,
indoors and outdoors at residences, commuting, at school or workplace, etc.). In particular, in
order to accurately characterize exposure, it is important to determine the relative contributions
of air pollutants in each microenvironment of concern (Klepeis, 2001; Ozkaynak, 1996; Kinney,
2002). Additionally, the US EPA's Office of Air Quality Planning and Standards applies the

7

-------
Hazardous Air Pollutants Exposure Model (HAPEM) to provide screening estimates of exposure
to toxics air pollutants (Ozkaynak, et al. 2008). The US EPA's Office of Research and
Development conducts research using the Stochastic Human Exposure and Dose Simulation
model (SHEDS) to simulate an individual using sequential time-location-activity data,
probabilistically assigning that individual's contact with environmental concentrations (PM2.5,
air toxics, and multiple pollutants), estimating the individual's exposure time profile and
applying Monte Carlo sampling to simulate the population exposures. This model is being
updated to increase its capabilities for PM2.5 and air toxics (Burke, et. al. 2001, Ozkaynak et. al.
2009). As exposure models continue to evolve, it is necessary to continue providing improved
modeled air quality estimates at finer scales (in various micro-environments) and to work to
better link air quality with exposure models.

In summary, there is a long history of research that demonstrates the need to improve our
understanding of human exposure to ambient pollutants, particularly at fine temporal and spatial
scales. Epidemiological studies have relied upon imperfect surrogates of exposure such as
concentrations from a central monitoring site. There is increasing evidence that existing
monitors are not capturing sharp gradients at fine scales, and are limited in the number of species
monitored. Researchers are now beginning to use fine scale air quality/exposure modeling in
support of environmental health studies. Within this overarching context, the following section
examines research questions that are specific to the current and future research needed to link air
quality and exposure data and models across temporal and spatial scales in order to solve a range
of complex human exposure problems.

3. What are the key science questions for linking air quality modeling to
exposure and environmental health studies?

The growing need for integrated exposure and risk-based approaches to health and
environmental impact assessments poses additional requirements for air quality models beyond
traditional regulatory applications. Regulatory models are generally used to predict the peak
value of the concentration distribution, unpaired in time and space, for comparison to air quality
standards that are applicable across broad areas such as a metropolitan area. However, air
quality models used for exposure and risk assessment studies must be able to predict highly-
resolved concentration distributions in space and time in small geographic regions to account for

-------
the changes in exposure as individuals or population groups move between microenvironments.
Thus, there are many scientific issues and science questions about air quality models that are
posed by exposure modelers. Among the key questions are:

-	Are current air quality models adequate to address exposure/health research needs?

-	What developmental and evaluation databases (e.g., wind tunnel studies, field studies,
numerical modeling studies) are needed for continued development and improvement of air
quality models for exposure applications?

-	How can atmospheric (e.g., CMAQ, AERMOD) and exposure models (e.g., SHEDS, APEX)
be linked with other exposure-related data (e.g., air quality measurements, census data,
housing information, time-activity data, transportation surveys, etc.) to improve the
prediction of personal or population exposures in the analyses of the relationships between
ambient air pollution and adverse heath outcomes?

-	How can the spatial and temporal resolution of ambient air concentration data (either
modeled or measured) used in support of epidemiologic studies be improved (e.g., by
improved fine-scale dispersion algorithms, by hybrid air quality modeling, hierarchical
Bayesian modeling or fusion of measurement and modeling data, wind-tunnel modeling,
land-use regression modeling, etc.)?

4. Where we are going and why?

Recent efforts

In the context of these questions, AMAD has been investigating the use of improved air quality
concentration fields for exposure assessments (US EPA, 2009). NERL scientists initiated several
research projects focused on developing techniques to combine air quality monitoring and
modeling data and to link regional and fine-scale air-quality models to provide spatially and
temporally resolved air pollutant concentration estimates; and to develop and enhance methods
to characterize local-scale, high-pollutant concentrations (hotspots), such as near roadways,
which impact a large percentage of the population. Specific efforts have included:

• Hybrid modeling for spatially and temporally resolved concentrations of particulate
matter, nitrogen oxides, carbon monoxide, benzene and other air toxics at census block
groups in New Haven, CT based on combining regional scale (CMAQ) with local scale
(AERMOD) modeling results for use in accountability studies (Isakov et al. 2009)

9

-------
• Conducting wind tunnel andfield tracer studies to provide databases for dispersion
model development (Heist et al., 2009).

• Linking air quality models to human exposure models to simulate exposures to particulate
matter in Philadelphia and in North Carolina, and particulate matter, ozone, nitrogen
dioxide and benzene in New Haven (Burke et al. 2001, EPA 2004, Ozkaynak et. al.

2009).

While these past efforts have resulted in novel approaches to improve, link and apply air quality
models, recent decisions to expand the research program (described in this white paper)
necessitated a more comprehensive approach beyond the initial projects listed above.

Identifying research needs

To identify gaps and limitations in existing approaches and improve the methodologies, EPA
conducted a 2-day workshop on March 16- 17, 2009, that assembled atmospheric scientists,
chemists and exposure and health scientists from the ORD and OAQPS. The workshop assessed
the needs, state-of-science, capabilities, gaps and resources associated with a fine-scale, air-
quality modeling research program. While discussions addressed a range of topics relevant to
fine-scale modeling (e.g., regulatory applications, accountability), here we report the outcomes
relevant to improving fine-scale air-quality models and linking them to human exposure and
health. In particular, discussions focused on the adequacy of current air quality modeling
approaches to meet human exposure and health study needs. These modeling approaches
include: (1) the Eulerian photochemical Community Multiscale Air Quality (CMAQ) model; (2)
the AERMOD dispersion model; (3) hybrid approaches that combine two or more of these; and
(4) other modeling approaches. Additionally there was a discussion of the databases needed for
model development and evaluation including those generated from wind tunnel experiments and
field studies.

CMAQ has typically been run at coarse resolutions (e.g., 36 km x 36 km or 12 km x 12 km
horizontal grid) and represents regional-scale processes. Driven by meteorology and emission
inputs, CMAQ is a multi-pollutant model that computes the fate and transport of over 100
chemicals and their species on regional to urban scales. AERMOD, also driven by meteorology
and emissions, is used to simulate the dispersion of individual pollutants in a local environment
(urban and finer scales) and does not include photochemistry and particle reactions or

-------
interactions, but does include limited simplified gaseous reactions. Hybrid approaches, as used
here, refer to a combination of the regional-scale CMAQ model and the local-scale AERMOD
model.

Any model development and evaluation program depends heavily on the availability of high
quality databases. With a focus on the improvement of models at urban and fine scale, wind
tunnel experiments and field studies are critical assets for developing and evaluating model
algorithms of local scale dispersion. Due to costs and instrumentation limitations, field
campaigns are rarely capable of capturing the full spatial fields of flow and concentration needed
to understand the physical processes controlling dispersion. Wind tunnel studies can provide
complete spatial distributions in cost-effective, controlled environments and are therefore critical
to model algorithm development. Field studies, however, are most critical for model evaluation
where variations in actual sources and meteorology are included. Laboratory studies
complement field data both in the interpretation of the field measurements and in the design of
optimum studies in complex environments. Finally, a wind tunnel is an important tool for
examining the effectiveness of proposed or planned air pollution mitigation techniques.

Examples of the use of laboratory (wind tunnel) studies in the development of improved
dispersion models include Perry et al. (1994), Perry and Snyder (1989), Perry (1992), Schulman
et al. (2000) and Heist et al (2009). Examples of such studies used in model evaluation include
Petersen and Perry (1996), Schulman et al. (2000), and Perry et al. (2005).

The research planning workshop examined the progress made with these past efforts and
established future directions needed to fill research gaps and evolve the program to the next
stage. These future directions fall into five major categories:

1. Fine-scale CMAQ modeling. Recognizing the possible benefits of running the multi-pollutant
CMAQ model at spatially resolved scales, this research will explore the development and
application of an enhanced version of CMAQ at 1 km horizontal spatial resolution as part of the
2011 CMAQ release. Research to support this version includes characterizing urban
morphology, developing approaches to run the 2-way coupled meteorology model (WRF) at 1
km resolutions, and applying and evaluating fine-scale modeling results.

-------
Implications for Model Development and Evaluation: Inherent in the success of any modeling
program is a strong model development and evaluation research program. The challenges
associated with model development and evaluation are magnified with a fine-scale modeling
program. Uncertainty associated with the emission and meteorology inputs will be enhanced at
finer resolutions. At coarser resolutions, concentration estimates resulting from modeled
physical and chemical processes are aggregated, reducing (to some degree) the impact of model
bias. At fine-resolutions, however, these inaccuracies can be more critical—particularly in the
applications that these results are used (i.e., estimating human exposure near roadways and at the
census tract level). Thus, improving upon and quantifying the estimates provided at this fine
resolution is critical for such applications.

The model development and evaluation activities required for this research corresponds
with the two major modeling approaches discussed throughout the paper; CMAQ and
AERMOD. CMAQ model development activities will be focused on running CMAQ at more
spatially- and temporally-resolved scales (e.g., 1-4 km) and include:

- Evaluating urban-scale applications of WRF and CMAQ

- Developing and testing local scale modeling techniques for line sources (roadways) in
dispersion models and in CMAQ (with near-source chemical and physical processes)

- Developing and testing integrated point source plume technique in CMAQ

- Refinement and testing of hybrid modeling with CMAQ/AERMOD and/or CMAQ and
other local source models

2. Local-scale dispersion modeling. This research will result in application driven improvements
to enhance and optimize AERMOD. For example, AERMOD has historically been used for
applications such as air-quality permitting that require single, prescribed meteorology inputs for
estimating dispersion and the resulting maximum exposure. AERMOD is now being used to
simulate pollutant concentrations from multiple sources across urban environments. Therefore,
this research will focus on developing methods to optimize emission inputs, such as modified
approaches for estimating local-scale mobile source emissions. This emissions optimization will
reduce the costs of running AERMOD and improve model results. Also critical to modeling
urban environments, this research will include development of improved AERMOD algorithms
to incorporate the influence of roadway configurations. In addition, the use of gridded
meteorological data as input to AERMOD is being investigated.

-------
Implications for Model Development and Evaluation: AERMOD model development and
evaluation will be conducted through the Atmospheric Exposure and Regulatory Model
Improvement Committee (AERMIC). The main goal of AERMIC is to provide scientific advice
and support to EPA in the area of near-source/short-range dispersion modeling and implies the
following objectives:

1.	Support EPA offices/programs concerned with or having a need for simulating short-
range dispersion in relation to permit applications, exposure research, and human health
risk assessments.

2.	Ensure that short-range dispersion programs include/incorporate state-of-the-art science.

3.	Assist with the transfer of state-of-art science and modeling techniques into
practical/workable tools applicable to key programs such as regulatory modeling
(permitting applications, SIP revisions, and regulation development in support of the
CAA, human and ecological exposure studies, source culpability analyses, air-
accountability studies (AAS), National Ambient Air Quality Standard (NAAQS)
assessments, etc.

4.	Interface with regional-scale modeling programs to address multi-scale issues/impacts.

5.	Ensure that EPA's near-field model(s) have the capabilities to address current and future
needs for air quality assessment and management.

The difference in program office/ORD roles in CMAQ and AERMOD model
development/evaluation are shown in a schematic figure below.

NERL NERL NERL/OAQPS OAQPS

Atmospheric 	* CMAQ Model 	. Base Year 	. Future Baseline &

Chemistry Data &1 |) Development & EZj) Model IZZj) Control Case
Models	Testing	Evaluation	Applications

R&D 		 Testing & Evaluation 	* Applications/

Reg Assessment

AERMIC AERMIC/NERL/OAQPS OAQPS

n u t-x ^ o k AERMOD Model K Performance K Model

W1 Development & l~> Evaluation a CZ> Applications a
9ori ms	Testing Consequence Analysis Guidance

13

-------
3. Hybrid modeling approaches. The success of pilot projects conducted in this area has
resulted in increased demand for fine-scale hybrid modeling by the epidemiology community.
This research will be application driven and includes developing improved techniques to
combine or statistically blend modeled and observed air quality data. In addition, comparison of
the various fine-scale modeling approaches, such as fine-scale CMAQ and other hybrid
approaches, will be a critical component of this research.

Implications for Model Development and Evaluation: As discussed earlier, model development
and evaluation activities associated with Hybrid modeling approaches are recognized as a desired
component of our research program.

4. CMAQ integrated with a particle puff model. Participants in the workshop recognized that
integrating a particle puff model into CMAQ may be a critical component for improving fine-
scale CMAQ model results. Suggestions included the incorporation of a simplified plume-in-
grid (PinG) approach, bringing Advanced Plume Treatment (APT) in-house for evaluation, and
incorporating a particle dispersion model. We propose to begin exploratory research in this area
within the Division, building off our collaboration on application of the HYSPLIT model and our
experience with the PinG methodology.

Implications for Model Development and Evaluation: As discussed earlier, model development
and evaluation activities associated with incorporating a particle puff model into CMAQ are
recognized as a desired component of our research program, and the Division has committed
recently to devoting a research scientist towards exploring this area.

5. Air quality model application and evaluation. The fine-scale nature of this research presents
unique challenges in evaluating approaches and results, and will require wind tunnel studies,
field studies and Computational Fluid Dynamic (CFD) modeling to improve and evaluate the
models. Recognizing the importance of model application and evaluation, this research includes
the evaluation of newly developed algorithms, overall model performance and model sensitivity.
This research includes 1) conducting additional wind tunnel and field experiments to create
databases for model evaluation, 2) characterizing uncertainty/variability and communicating
results as distributions vs. absolute values, and 3) evaluating the benefit of providing more finely
resolved spatial and temporal air quality data on health studies outcomes.

-------
Future model and methodolosv development

Addressing the research priorities listed above, EPA/NERL is currently involved in several
activities to develop and evaluate techniques in support of exposure and health studies. This
research is focused on linking air quality models to exposure and health studies. Critical to that
is the improvement of fine-scale air quality models. For example, EPA/NERL is collaborating
with the National Children's Study (NCS) to advance the state-of-the-science for human
exposure assessment. In this effort, we are developing a methodology to provide spatially
resolved concentrations for the multiple NCS study areas. We will evaluate the methodology
using the results from previous studies in Baltimore, New Haven and Detroit. We will conduct
research to develop additional approaches such as spatio-temporal covariate analyses to identify
areas of high concentrations for community-based cumulative exposure assessment to the NCS.

Participants in the workshop recognized a variety of fine-scale modeling needs including: 1)
improvements to local-scale dispersion models (e.g., AERMOD) to adequately simulate
pollutant concentrations from multiple sources in complex urban environments 2) the
development and application of a more spatially resolved regional-scale model (e.g., CMAQ),
and 3) the incorporation of a simplified plume in grid particle model in CMAQ.

Focusing on urban areas dictates an examination of exposures of large populations to a variety of
source types arrayed in a relatively small geographical area. Major roadways are an important
source type to be accounted for in urban areas. Additionally, a growing number of health studies
has linked adverse health effects to populations living and working near major roadways.
Therefore, EPA/ORD has developed a near-road research program to better understand the
relationship between roadway sources and health outcomes.

In part of the near-road research program, AMAD is developing a refined line-source algorithm
for inclusion in applied urban dispersion models (e.g., AERMOD) based on laboratory,
numerical and field studies. This refined algorithm will be used to provide estimates of near-road
concentrations of air pollutants for exposure models. The increased temporal and spatial
resolution of the input to the model will greatly enhance exposure assessments.

-------
In support of this model development, AMAD is conducting a series of wind tunnel experiments
on near-road dispersion examining the influences of meteorology (wind direction, approach
flow), roadway source characteristics (vehicle-induced turbulence), and near-road structures
(depressed roadways, barriers, vegetation, and buildings). In addition to characterizing
concentration gradients near roadways, these databases provide detailed distributional
representations of concentrations in these complex urban settings useful for model evaluation.

AMAD is developing a statistical "data blending" modeling approach that combines regional-
scale modeling, local-scale modeling, and observations to provide spatially- and temporally-
resolved concentrations in support of exposure and health studies. The "data blending" approach
has several advantages: 1) it produces accurate, spatially resolved estimates of the daily air
pollution field at receptors of interest within a metropolitan area; 2) it harnesses the strengths and
compensates for the weaknesses of each data source; accounts for bias in the numerical models;
3) accounts for spatial and temporal dependence; and 5) avoids the limitations of other statistical
methods. The data blending model has been initially tested in the Baltimore area for January and
July of 2002 for PM2.5 and NOx- The model utilized the CMAQ and AERMOD output with
monitor observations. After additional evaluation and refinements based on the results of the
Baltimore study, the data blending model will be applied in the Atlanta metropolitan area for
several pollutants over all of 2002. Collaborators at Emory University will compare this ambient
concentration estimate to a number of other exposure metrics.

An important issue is the practicality of the modeling tools for exposure assessments. The
uncertainty of model predictions at fine scale paired in space and time is expected to be high,
mainly because of the uncertainty in emission and meteorological inputs. Therefore, providing
distributional representations of predicted concentrations seems to be a more practical solution.
This could be accomplished, for example, using a computational scheme that allows the
estimation of concentrations over micro-environments inside grid-cells, where specific emission
activity dominates (Valari and Menut, 2010). The contribution of different emission activities
can be also estimated by using emission data at sub-grid scale. The advantage of this approach is
that modeled concentrations are directly associated with human-activities during the day and are
therefore easily adapted to human exposure models. This approach would require additional
information to estimate concentrations within grid cells. One way of providing this information

-------
is to conduct wind tunnel experiments. The results from these wind tunnel studies can provide
detailed sub-grid scale concentrations for evaluation of these approaches. The workshop
participants recommended conducting research to investigate the applicability of this statistical
approach for exposure assessments. However, since epidemiologists use concentrations paired
in space and time in their analyses (e.g. based on central monitor site and/or land-use regression
models), changing their methodology to use distributional representation is not going to be
achieved easily, and will require close collaboration. Furthermore, other modeling needs will
emerge. Therefore, it is essential that AQ models begin to be tested and used in ongoing
epidemiological studies. The EPA cooperative research projects, described below, are key
examples of this type of collaboration needed for distributional representations to be accepted by
the epidemiological community.

Future application and evaluation of the methodology in exposure and health studies

To test the newly developed techniques in support of exposure and health studies, AMAD's
scientists participate in several cooperative research projects as a part of the EPA/NERL
Cooperative Agreement Program. The overall goal of this "Air Pollution Exposure and Health
Program" is to enhance the results from epidemiologic studies of ambient PM and gaseous air
pollution through the use of more reliable approaches for characterizing personal and population
exposures. The main objectives of research carried out under this Coop program are to: 1)
develop and apply innovative approaches for exposure prediction by utilizing available or
modeled information based on monitoring data, meteorological data, exposure factors data, air
quality and exposure modeling results, and 2) conduct analyses of available health data using the
various tiers and alternative exposure metrics developed. This research will allow us to evaluate
the benefits of using various levels of detail in air quality modeling to support exposure and
health studies. AMAD scientists are involved in three exposure health cooperative programs
with Emory University, University of Washington and Rutgers University to examine the utility
of refining exposure metrics used in air pollution studies (Neas and Ozkaynak 2006). Various
cooperative research projects are summarized in a table below.

-------
Overview of NERL Coops: Evaluating Alternative Exposure Estimates with Health Data

Overall Goal: Enhance results of epi studies through the use of improved exposure metrics

Collaborators

Pollutants

Exposure Approach

Health Study

EPA Role

Rutgers/LBNL

•	PM2.5

•	PM2.5
species:

EC,0C,S04,
N03,NH3)

•	Ambient Monitoring

•	SHEDS

•	Infiltration Model
(LBNL)

•	SHEDS-Infiltration
Hybrid

•	NJ MI

•	NJ Birth
Outcomes

•	SHEDS

•	CMAQ*

Emory/Ga
Tech

•	PM2.5
EC

•	CO

•	NOx

•	Ozone

•	Ambient Monitoring

•	Interpolation

•	CMAQ

•	Mobile Emissions Model

•	CMAQ+AERMOD

•	Air Exchange/Infiltration

•	Atlanta ED

•	Atlanta
ICD

•	CMAQ

•	AERMOD

•	Data Fusion

•	EMI*

•	SHEDS*
NOTES: * Being
considered

U. of

Washington

•	PM2.5

•	NOx*

Assimilated Data from
Ambient
• CMAQ+AERMOD**
Spatio-temporal models
(GIS & physically-based)

•	MESA-Air

•	WHI-OS

•	CMAQ**

•	AERMOD**
NOTES: ** For selected
MESA Cities (e.g.,
Baltimore, New York
City)

University of
Michigan

PM2.5,
EC/OC, BC,
CO, NOx,
VOCs, and
Air Toxics

Health Effects of Near-Roadway
Exposures to Air Pollution

Detroit
(Childhood
Health Effects
from Roadway
and Urban
Pollutant
Burden Study)

EPA is providing
monitoring data, and
air quality modeling
to support the health
data analysis

The Rutgers/EOHSI-LBL group is examining associations between PM2.5 mass and species and
adverse health for two established epidemiology studies using different tiers for exposure
specification. In this cooperative research project, the team will investigate the use of air quality
modeled concentrations as inputs to epidemiologic analysis. AMAD scientists provided 12 by 12
km CMAQ results for the entire state of New Jersey to be used as refined exposure surrogates in

18

-------
two epidemiology studies: The New Jersey (NJ) Triggering of Myocardial Infarction (MI) Study
and the NJ Adverse Birth Outcomes Study.

In another cooperative research project between EPA/NERL and Emory University, researchers
from the Emory University, Georgia Tech, and EPA/NERL will evaluate various exposure
metrics for ambient traffic-related (CO, NOx, PM2.5 and PM2.5 EC) and regional (O3 and SO42")
pollutants in Atlanta. These metrics include various tiers of exposure refinement, including the
use of 1) ambient receptor data from interpolated measurements, 2) ambient concentrations from
the hybrid modeling approach (Isakov et. al., 2009), and 3) exposure factors data. These
exposure metrics will be applied to two ongoing Emory epidemiologic studies examining
ambient air pollution and acute morbidity in Atlanta, GA.

The University of Washington coop is motivated by the Multi-Ethnic Study of Atherosclerosis
and Air Pollution (MESA Air), a cohort study funded by the U.S.EPA. MESA Air exposure
assessment emphasizes accurate prediction of individual ambient-source exposures in order to
accomplish its primary aim of assessing the relationship between chronic exposure to air
pollution and progression of sub-clinical cardiovascular disease. Air pollution cohort studies
have increasingly used predicted ambient air quality based on GIS-based covariates in "land use
regression" models, and more recently some studies have used predicted exposure from spatial
regression models to account for spatial correlation structure. In this study, EPA is providing
monitoring data, regional and local-scale air quality modeling, data fusion methodologies, and
comparing different approaches for air quality estimates as alternative personal exposure
indicators of multiple pollutants in conjunction with the Baltimore MESA Air study to develop a
better linkage to exposure models and human health studies.

In another cooperative project between EPA/NERL, state agency, and academia, AMAD
scientists are investigating the use of tiered data (observations only, fused air quality data, and
exposure factors) in a time series epidemiology study in New York State relating respiratory-
related hospital admissions to ozone concentrations.

Recently, Near Road Research has been identified as a Problem of Broad National Significance.
NERL scientists are involved in the cooperative research project between EPA/NERL and

19

-------
University of Michigan titled the Childhood Health Effects from Roadway and Urban Pollutant
Burden Study (CHERUBS). This project is focused on health effects of near-roadway exposures
to air pollution. In this study, researchers from the University of Michigan will study respiratory
health effects in about 105 children with asthma who live in low-traffic and high-traffic areas. In
addition to providing monitoring data, the EPA is supporting the health data analysis by
providing fine-scale (AERMOD) modeling as the primary air-quality input to the exposure
assessments, conducting wind tunnel studies to examine meteorological influences and roadway
characteristics at the study sites to help interpret the exposure and health data, and providing
limited monitoring at selected sites for selected periods.

In summary, AMAD is taking a leadership role in developing better modeling information
needed by exposure and health modelers and improving and refining these measures through
direct application with researchers in academia and regulatory agencies.

5. What collaboration is necessary?

The complexity of this effort requires collaborations with many groups:

A. Collaboration with EPA ORD scientists (NERL, NRMRL and NHEERL) and with
epidemiologists and health scientists outside of EPA:

- In multidisciplinary research, we need to collaborate with experts in several
organizations. Health data analysis and exposure modeling is not AMAD's area of
expertise; therefore collaboration is necessary.

- For the EPA/NERL collaboration with the National Children's Study, we will assist the
NCS Centers to apply the EPA/NERL modeling methodology to identify areas of high
concentrations for community-based cumulative exposure assessment to the NCS.

- Participate in on-going NERL Air Pollution Exposure and Health Coop program and
other ORD exposure and health studies to develop, test, evaluate and apply these
methodologies in different parts of the US. Results of these coops would identify future
collaboration needs.

-------
B. Collaboration/coordination with OAQPS:

OAQPS is also involved in exposure and risk modeling assessments. Even though OAQPS and
ORD have different responsibilities for exposure and risk modeling applications, we should be
using similar tools and work collaboratively. A good example of such collaboration is the "near
Road Research Program". Current OAQPS areas of interest are:

- Increasing demands for more spatially and temporally resolved pollutant concentration
fields at multiple scales and local-scale (or "Hotspot") analysis for risk analysis (e.g., SIP
& conformity analysis for mobile sources)

- Driven by:

o New demands for improved characterization of ambient pollutants for exposure

modeling, health studies and risk assessments
o Inform monitoring network design

o Residual PM2.5 non-attainment analyses to address excesses in urban increment
o Multi-pollutant approaches to air quality management that require multiple scales

6. What research products and outcomes are likely to emerge within 5 years?

The overall goal of this research is to enhance epidemiological and health effect risk assessment
studies of ambient air pollution by incorporating into exposure models more refined estimates of
air pollution (due to better air quality inputs) than those currently available from ambient air
quality measurements. Using advanced modeling approaches will strengthen the air quality and
human exposure aspects of on-going or future epidemiologic studies. We anticipate the
following products within next 5 years:

1. Advancement in modeling approaches:

- Advanced modeling approaches (based on improved model algorithms, improved input
data, etc.). For example, a new line source algorithm in AERMOD for the purposes of
getting improved estimates near roadways. Also, improved source characterization (i.e.,
initial dispersion) in AERMOD for near road structures (noise barriers, depressed
sections, vegetation, etc.).

o Schedule: improved line-source algorithm - within 1-2 years; more complex road
configuration and vehicle-induced turbulence - within next 3-5 years.

-------
- Improving methodologies for emission inputs for mobile sources. For example,
developing link-based emission inventory using EPA/OTAQ latest research products.

o Schedule: methodologies for link-based emission inventory - within 1-2 years;
refined methodology - within next 3-5 years.

- Databases for model development and evaluation of these advanced modeling
approaches. Make available to the scientific community the results of the EPA sponsored
field studies such as the Idaho Falls tracer study, wind tunnel experiments, near-road field
studies in Raleigh, Las Vegas, and Detroit.

o Schedule: Analysis of field study results from Idaho Falls, Las Vegas experiments
- within 1-2 years; Detroit - within next 3-5 years. Wind tunnels results for
effects of road structures (road barriers and depressed sections, for perpendicular
winds) - 1-2 years; for expanded set of meteorological conditions and urban
morphology - 3-5 years

- Successful demonstrations of linked air-quality and exposure modeling to support health
studies. For example, the demonstration of the hybrid modeling approach and/or
statistical "data blending" techniques can be used as a starting point for developing a
methodology for linkage between air quality modeling and environmental health.

o Schedule: within 1-2 years

- Developing distributional representations of exposures to support environmental
exposure and health studies. For example, using a computational scheme that allows the
estimation of concentrations over micro-environments inside grid-cells, where specific
emission activity dominates. The advantage of this approach is that modeled
concentrations are directly associated with human-activities during the day and are
therefore easily adapted to human exposure models.

o Schedule: within 3-5 years

2. Better linkage between sources of air pollution and health effects

- Understanding the limitation of various modeling approaches is a key research product
from the cooperative projects. For example, the main objective of the cooperative
research project with Emory University is to evaluate the benefits of using different tiers
of exposure metrics in support of environmental epidemiologic studies. These tiers
include: 1) central site monitor data; 2) interpolated observations; 3) regional modeling

-------
(CMAQ) results; 3) hybrid modeling (CMAQ and AERMOD) results; and 5) statistical
"data blending" modeling results. The research findings from this research project would
help identify gaps and develop future research priorities in developing modeling tools to
support environmental health studies.

o Schedule: results from the pilot studies within 2 years, refinements - within 3-5
years

- Increase communication and collaboration with epidemiology researchers. For example,
continue providing air quality estimates at the urban scale such as the Baltimore area with
the University of Washington and the Atlanta area with Emory University and Georgia
Tech.

o Schedule: within 1-2 years

- Demonstrate the utility of using fine scale model estimates. Test the impacts of
increasing the level of spatial estimates, e.g., CMAQ estimates alone, AERMOD plus
CMAQ estimates, hybrid model estimates and statistical blending techniques on
epidemiological model predictions in Baltimore and Atlanta areas.

o Schedule: within 1-2 years

- Encourage epidemiologist to use physics based model estimates instead of empirical land
use regression models. For example, test the statistical utility of using CMAQ estimates
in New Jersey by Rutgers University.

o Schedule: within 1-2 years

- Work with collaborators to publish research results in peer reviewed journal articles.

o Schedule: identified 4 to 5 papers within 1-2 years; additional papers within 3-5
years.

- Reducing uncertainty. Reducing uncertainties in linking health and environmental
outcomes to sources of air pollution is a Long Term Goal of the ORD Clean Air Research
program. The improved methods and tools described in previous sections would help
reduce uncertainty in linking health and environmental outcomes to sources of air
pollution. The improved methods would also help better design environmental
epidemiologic studies. The results obtained from appropriately designed and conducted
epidemiologic studies help enhance public health by reducing uncertainties in exposure
and risk assessments and by providing information that can be used to develop optimum
exposure and risk mitigation strategies. Reducing uncertainties in the risk assessment

-------
process is the main objective of the ORD Human Health Research Program, whereas
establishing risk assessment and risk management is an integral part of the EPA mission
to protect public health and safeguard the environment.

This research will also be used to link sources/origins of air pollution and health effects, which
directly contributes to a LTG in the Air MYP. The AQ models being source-oriented can provide
considerable insights into emissions and dispersion/transport of non-reactive primary pollutants
and also the generation of secondary pollutants from emissions/transport/transformation of
primary pre-cursors.

Thus, there will be many exciting research products emerging within the next five years, and to
be fully successful in this effort, EPA/NERL must collaborate with internal and external groups.

24

-------
References

Arain, M. A. ; R. Blair, R.; Finkelstein, N.; Brook, J. R.; Sahsuvaroglu, T.; Beckerman, B.; Zhang, L.;
Jerrett, M. "The use of wind fields in a land use regression model to predict air pollution
concentrations for health exposure studies." Atmos. Environ. 2007 41, 3453-3464.

Burke, J.; Zufall M.; Ozkaynak, H. A population exposure model for particulate matter: Case study
results for PM2 5 in Philadelphia, PA. J. Expos. Anal. & Environ. Epidem. 2001, 11, 470-489.

Burnett, R.T., Brook, J., Dann, T., Delocla, C., Philips, O., Cakmak, S., Vincent, R., Goldberg, M.S., and
Krewski, D. Association between particulate- and gas-phase components of urban air pollution
and daily mortality in eight Canadian cities. Inhal Toxicol 2000; 12 Suppl 4:15-39.

Cohen, M.A.;Adar, S.D.;Allen, R.W.;Avol, E.;Curl, C.L.;Gould, T.;Hardie, D.;Ho, A.;Kinney, P.;Larson,
T.V.;Sampson, P.;Sheppard, L.;Stukovsky, K.D.;Swan, S.S.;Liu, L.J.S.;Kaufman, J.D. Approach
to Estimating Participant Pollutant Exposures in the Multi-Ethnic Study of Atherosclerosis and
Air Pollution (Mesa Air);Environ. Sci. Technol. 2009, 43(13), 4687-4693.

Dockery, D.W.;Pope, C.A.;Xu, X.P.;Spengler, J.D.;Ware, J.H.;Fay, M.E.;Ferris, B.G.;Speizer, F.E. An
Association between Air-Pollution and Mortality in 6 United-States Cities;New England Journal
of Medicine 1993, 329(24), 1753-1759.

Dominici, F., Zeger, S.L, and Samet, J.M. A measurement error model for time-series studies of air
pollution and mortality. Biostatistics. 2000;1: 157-175.

Dominici, F., Peng, R.D., Bell, M.L., Pham, L., McDermott, A., Zeger S.L, Samet JM. Fine particulate air
pollution and hospital admission for cardiovascular and respiratory disease. JAMA 2006;
295:1127-1134.

English, P., Neutra, R., Scalf, R., Sullivan, M., Waller, L., Zhu, L. (1999). Examining associations
between childhood asthma and traffic flow using a geographic information system." Environ.
Health Perspect. 1999, 107, 761-767.

Gauderman, W.J, Vora, H., McConnell, R., Berhane, K., Gilliland, F., Thomas, D., Lurmann, F., Avol,
E., Kunzli, N., Jerrett, M., and Peters, J. Effect of exposure to traffic on lung development from
10 to 18 years of age: a cohort study. Lancet 2007; 369 (9561):571-7.

Georgopoulos, P., Wang, S., Vyas, V., Sun, Q., Burke, J., Vedantham, R., McCurdy, T., Ozkaynak, H.
(2005) A source-to-dose assessment of population exposures to fine PM and ozone in
Philadelphia, PA, during a summer 1999 episode. J. Expos. Analy. & Environ. Epidemi., 2005,
15, 439-457.

Heist, D. K., Perry, S.G., Brixey, L.A. A wind tunnel study of the effect of roadway configurations on
the dispersion of traffic-related pollution. Submitted to Atmos. Environ. 2009.

25

-------
Hoffmann, B., Moebus, S., Mohlenkamp, S., Stang, A., Lehmann, N., Dragano, N., Schmermund, A.,

Memmesheimer, M., Mann, K., Erbel, R., Jockel, K.H., (2007). Residential exposure to traffic is
associated with coronary atherosclerosis. Circulation, 2007, 116, 489-96.

Isakov, V.; Graham, S.; Burke, J.; Ozkaynak, H. Linking Air quality and Exposure Modeling; EM
Magazine 2006, 9, 26-29.

Isakov, V.; Ozkaynak, H. In Air Pollution Modeling and its Applications XIX; Borrego, C., Ed.; Springer
Science & Business Media B.V., The Netherlands, 2008; pp 616-624.

Isakov, V.;J. Touma, J. Burke, D. Lobdell, T. Palma, A., Rosenbaum, H., Ozkaynak. Combining Regional
and Local Scale Air Quality Models with Exposure Models for Use in Environmental Health
Studies. J. A&WMA, 2009, 59, 461-472.

Ito, K., De Leon, S.F., and Lippmann, M. Associations between ozone and daily mortality: analysis and
meta-analysis. Epidemiology 2005; 16: 446-457.

Jerrett, M., Burnett, R.T., Ma, R., Pope, C.A., 3rd, Krewski, D., Newbold, K.B., Thurston, G., Shi, Y.,
Finkelstein, N., Calle, E.E., Thun, M.J. Spatial analysis of air pollution and mortality in Los
Angeles. Epidemiology 2005, 16, 727-736.

Jimenez-Guerrero, P., Jorba, O., Baldasano, J.M., Casso, S. The use of a modeling system as a tool for air
quality management: Annual high-resolution simulations and evaluation. Science of the Total
Environment, 2007, 390, 323-340.

Kinney, P.L., and Ozkaynak, H. Associations of daily mortality and air pollution in Los Angeles County.
Environ Res. 1991;54:99-120.

Kinney, P.L., Chillrud, S.N., Ramstrom, S., Ross, J., Spengler, J.D. Exposures to multiple air toxics in
New York City. Environ. Health Perspect., 2002, 110 Suppl 4: 539-46.

Klepeis, N.E., Nelson, W.C., Ott, W.R., Robinson, J.P., Tsang, A.M., Switzer, P., Behar, J.V., Hern, S.C.,
Engelmann, W.H. The National Human Activity Pattern Survey (NHAPS): a resource for
assessing exposure to environmental pollutants. J. Expo. Anal. Environ. Epidemiol., 2001, 11,
231-52.

Levy, J.I., Greco, S.L., Spengler, J.D. The importance of population susceptibility for air pollution risk

assessment: a case study of power plants near Washington, DC. Environ. Health Perspect., 2002,
110, 1253-1260.

Levy, J.I., Chemerynski, S.M. and Sarnat, J.A. Ozone exposure and mortality: an empirical Bayes
metaregression analysis. Epidemiology 2005: 16: 458-468.

Marshall, J.D., Nethery, E., Brauer, M. Within-urban variability in ambient air pollution: Comparison of
estimation methods. Atmos. Environ., 2008, 42, 1359-1369.

26

-------
Miller, K.A.;Siscovick, D.S.;Sheppard, L.;Shepherd, K.;Sullivan, J.H.;Anderson, G.L.;Kaufman, J.D.

Long-Term Exposure to Air Pollution and Incidence of Cardiovascular Events in Women;N Engl
J Med 2007, 356(5), 447-458.

Mukeijee, S., Smith, L.A., Norris, G.A., Morandi, M.T ., Gonzales, M., Noble, C.A., Neas, L.M .,

Ozkaynak, A.H.. Field method comparison between passive air samplers and continuous monitors
for volatile organiccompounds and N02 in El Paso, Texas, USA. J. A&WMA, 2004, 54,17 307-
319

Ozkaynak, H., Xue, J., Spengler, J., Wallace, L., Pellizzari, E., Jenkins, P. Personal exposure to airborne
particles and metals: results from the Particle TEAM study in Riverside, California. J. Expo.

Anal. Environ. Epidemiol., 1996, 6, 57-78.

Ozkaynak, H.; Palma, T.; Touma, J.; Thurman, J. Modeling population exposures to outdoor sources of
hazardous air pollutants. J. Expo. Sci. Environ. Epidemiol., 2008, 18, 45-58.

Ozkaynak, H., Frey, C., Burke, J., Pinder, R. Analysis of coupled model uncertainties in source-to-dose
modeling of human exposures to ambient air pollution: A PM2.5 case study. Atmospheric
Environment, 2009, 1641-1649.

Perry, S. G. and W. H. Snyder, 1989: Laboratory simulations of daytime mixed-layer heights in complex
terrain. Preprints of the Sixth Joint AMS/AWMA Conference on Application of Air Pollution
Meteorology, Jan.30 - Feb. 3, 1989, pp 256-259, American Meteorological Society, Boston, MA.

Perry, S. G., 1992: CTDMPLUS: A dispersion model for sources in complex topography. Part I:

Technical formulations. J. Appl. Meteor., 31, 633-645.

Perry, S. G., R. S. Thompson, and W. B. Petersen, 1994: Consideration for modeling small-particulate
impacts from surface coal-mining operations based on wind-tunnel simulations. Preprints of the
Eight Joint AMS/AWMA Conference on Air Pollution Meteorology, Jan 23-28, 1994, Nashville,
TN, American Meteorological Society, Boston, MA.

Perry, S.G., A. J. Cimorelli, R. J. Paine, R. W. Brode, J. C. Weil, A. Venkatram, R. B. Wilson, R. F. Lee,
and W. D. Peters, 2005: AERMOD: A dispersion model for industrial source applications. Part II:
Model performance against seventeen field-study databases. J. Appl. Meteor., 44, 694-708.

Petersen, W.B., and S.G. Perry, 1996: Improved algorithms for estimating the effects of pollution
impacts from area and open pit sources. In Air Pollution Modeling and Its Application XI.
Volume 21. S.-E. Gryning and F.A. Schiermeier (Eds.). Plenum Press, New York, NY, 379-386.

Pope, C.A., Burnett, R.T., Thun, M.J., Calle, E.E, Krewski, D., Ito, K., and Thurston, G. Lung cancer,
cardiopulmonary mortality, and long-term exposure to fine particulate air pollution. JAMA 2002;
9:1132-1141.

27

-------
Pope, C.A.;Thun, M.J.;Namboodiri, M.M.;Dockery, D.W.;Evans, J.S.;Speizer, F.E.;Heath, C.W.

Particulate Air-Pollution as a Predictor of Mortality in a Prospective-Study of Us Adults;Am. J.
Respir. Crit. Care Med. 1995, 151(3), 669-674.

Pope, C.A III, and Dockery, D.W. Critical review: health effects of fine particulate air pollution: lines that
connect. J Air Waste Manage Assoc 2006; 56: 709-742.

Sarnat, S.E., Klein, M, Peel, J.L., Mulholland, J., Sarnat, J.A., Flanders, W.D., Waller, L.A., and

Tolbert, P.E. Spatial considerations in a study of ambient air pollution and cardiorespiratory
emergency department visits. Epidemiology 2006; 17(6) Suppl:S242-S243.

Schulman, L.L., Strimaitis, D.G. and Scire, J.S., 2000. Development and evaluation of the PRIME plume
rise and building downwash model. J. Air Waste Manage. Assoc. 50, pp. 378-390.

Schwartz, J. Air pollution and hospital admissions for heart disease in eight U.S. counties. Epidemiology
1999; 10: 17-22.

Smith, L., Mukeijee, S., Gonzales, M., Stallings, C., Neas, L., Norris, G., Ozkaynak, H. Use of GIS and
ancillary variables to predict volatile organic compound and nitrogen dioxide levels at
unmonitored locations." Atmos. Environ., 2006, 40, 3773-3787.

U.S. EPA Report, A Conceptual Framework for U.S. EPA's National Exposure Research Laboratory,
2009.

Valari, M., Menut, L. Transferring the heterogeneity of surface emissions to variability in pollutant

concentrations over urban areas through a chemistry-transport model Original Research Article
Atmos. Environ., 2010, 44, 3229-3238.

Wallace, L., Williams, R., Suggs, J. et al., Exposure of High-Risk Subpopulations to Particles: Final

Report—APM-21. US EPA, National Exposure Research Laboratory, Research Triangle Park,
NC. EPA/600/R-03/145. February, 2004.

Wang, S.W., Tang, X., Fan, Z.H., Wu, X., Lioy, P. J. and Georgopoulos, P. G. Modeling of
Personal Exposures to Ambient Air Toxics in Camden, New Jersey: An Evaluation
Study. J Air Waste Manage Assoc 2009; 59, 733-746.

Ware, J.H., Spengler, J.D., Neas, L.M., Samet, J.M., Wagner, G.R., Coultas, D., Ozkaynak, H., and

Schwab, M. Respiratory and irritant health effects of ambient volatile organic compounds: The
Kanawha County Health Study. Am J Epidemiol 1993; 137(12): 1287-1301.

Woodruff, T.J., Axelrad, D.A., Caldwell, J., Morello-Frosch, R., and Rosenbaum, A. Public health

implications of 1990 air toxics concentrations across the United States. Environ Health Perspect
1998; 106:245-251.

28

-------
Zeger, S.L., Thomas, D., Dominici, F., Samet, J.M., Schwartz, J., Dockery, D., and Cohen, A Exposure
measurement error in time-series studies of air pollution: concepts and consequences, Environ
Health Perspect 2000; 108: 419-426.

Zeka, A., and Schwartz, J. Estimating the independent effects of multiple pollutants in the presence of
measurement error: an application of a measurement-error-resistant technique. Environ Health
Perspect 2004; 112: 1686-1690.

Acknowledgements

We are grateful to Dr. Haluk Ozkaynak for his input and review comments on the draft paper

29

-------