A MODELING METHODOLOGY TO SUPPORT EVALUATION OF PUBLIC
HEALTH IMPACTS OF AIR POLLUTION REDUCTION PROGRAMS

V. Isakova, H. Ozkaynakb

aNOAA Atmospheric Sciences Modeling Division (In partnership with the U.S. EPA), Research Triangle
Park, North Carolina, 27711

bOffice of Research and Development, U.S. EPA, Research Triangle Park, North Carolina, 27711
Abstract

Environmental public health protection requires a good understanding of the types and
locations of pollutant emissions of health concern and their relationship to
environmental public health indicators. Therefore, it is necessary to develop the
methodologies, data sources, and tools for assessing the public health impact of air
pollution reduction programs, also referred to as accountability analysis. Since air
quality models are among the main tools that can be used to evaluate the impacts from
emissions changes, either due to growth or implementation of source control strategies,
these approaches play a vital role in most air accountability studies.

In this study, we present a modeling methodology to estimate concentrations for
multiple pollutants that include both local features (hot spots) and regional transport.
The local impacts from mobile sources and significant stationary sources are estimated
using a dispersion model (AERMOD). These local details are combined with regional
background estimates computed by a photochemical grid model (CMAQ) in a "hybrid"
approach to derive total concentrations required for the subsequent human exposure
analysis.

We demonstrate an application of this methodology in New Haven, Connecticut. The
city of New Haven has implemented a comprehensive Clean Air Initiative, which
includes a number of federally mandated and voluntary air pollution programs. This
project is a collaborative effort with state and local agencies including government,
academia, and the New Haven community, to apply and evaluate air quality and human
exposure models that can be used with health data and to assess the feasibility of using
this information to conduct an air accountability study. Although this study is based in
one city, the methodologies developed through this project can have broad application
to other areas within the United States and internationally.

1. INTRODUCTION

Air quality has improved substantially in the United States in recent decades, in large
part due to increasingly stringent federal and state air quality regulations. While many
studies have documented links between better air quality and improvements in a variety
of human health metrics, direct evidence is lacking about the extent to which specific
control measures have improved health. Over the past decade, various epidemiological
studies have examined the relationships between acute and chronic health outcomes and
measured ambient particulate matter, ozone and other co-pollutant concentrations. In
the context of air pollution health effects, results from recent epidemiological studies
indicate the importance of determining the key sources and constituents of indoor,


-------
outdoor pollution, and personal exposures to PM, ozone and other air pollutants.
However, understanding the magnitude and nature of human exposure is clearly the first
step in assessing the occurrence of adverse effects that could follow upon contact with
environmental pollutants. One of the ways to access the human exposure is through the
use of exposure models such as EPA's Hazardous Air Pollutant Exposure Model
(HAPEM), the Air Pollutant Exposure Model (APEX), or Stochastic Human Exposure
and Dose Simulation (SHEDS).

Since predicted concentrations from air quality models are key drivers for human
exposure models, it is essential to improve the accuracy and precision of spatial and
temporal characterization of results from these models. However, complex interactions
between interventions over time can make it difficult to isolate the environmental
impacts and associated health effects of any one regulation. For example, a regulatory
action may have varying effects on emissions depending upon compliance and the real-
world effectiveness of the interventions applied. Often, the connection between
emissions and ambient air quality depends on complex atmospheric and chemical
transformations. Environmental public health protection requires a good understanding
of the types and locations of pollutant emissions of health concern and their relationship
to environmental public health indicators. Therefore, it is necessary to develop the
methodologies, data sources, and tools for assessing the public health impact of air
pollution reduction programs, and accountability analysis. Since air quality models are
the principal predictive tool for assessing the impacts of potential emissions control
strategies on future-year concentrations, we describe here a modelling approach to
support air accountability studies.

2. AIR QUALITY MODELING APPROACH

Environmental health studies require detailed information on air quality. Therefore, air
quality modelling should include local-scale features, long-range transport, and
photochemistry to provide the best estimates of air concentrations. There are several
available modelling approaches capable of assessing pollutant concentration gradients at
a fine resolution (Touma et. al., 2006) and these can be categorized into two major types
of air quality models: source-based dispersion models and Eulerian grid-based chemical
transport models. Chemical transport models, such as the Community Multi-scale Air
Quality (CMAQ, Byun and Schere, 2006), are used to simulate the transport and
formation of ozone, acid rain, particulate matter (PM) and other pollutants formed by
chemical reactions among precursor species that are emitted from hundreds or
thousands of emission sources. Such models may be set up to apply to a wide range of
scales ranging from global to urban. However, regional-scale grid-based models can
address photochemistry effects, but not local-level gradients. CMAQ provides volume-
average, hourly concentration values for each grid cell in the modelling domain.
Emissions are assumed to be instantaneously well-mixed.


-------
While grid-models are the model platform of choice for simulation of chemically-
reactive airborne pollutants, source-based dispersion models such as AERMOD
(Cimorelli et. al., 2005) that have been developed to simulate pollutant concentrations
within a few hundred meters or a few kilometers from the source are typically used for
local scales. These models generally do not take into account atmospheric chemical
reactions or they do so using simplified representations such as first-order pollutant
decay. They provide detailed resolution of the spatial variations in hourly-average
concentrations. It would be desirable to combine the capabilities of grid-models and
dispersion models into one model, but this is a yet evolving area of research and
development. One option is a hybrid approach (Isakov et. al., 2007), where a regional
grid model and a local plume model are run independently. To illustrate how air quality
models can be used to provide inputs to human exposure models, we focus on a 20 by
20 km area encompassing New Haven, Connecticut that includes many stationary
sources emitting toxic pollutants and several major roadways as indicated in Figure 1.
The city of New Haven, with population of approximately 125,000, is a recipient of one
of EPA's nationally funded Community Air Toxics projects. Through this project, New
Haven has implemented a comprehensive Clean Air Initiative, which includes a number
of voluntary air pollution programs. Along with local and state efforts, there are also
several Federal regulations that have either recently been or soon will be implemented
(e.g., Clean Air Interstate Rule). This project presents an opportunity to assess the
feasibility of using air quality and human exposure models that can be used with health
data to conduct an air accountability study.

Resolving fine scale pollutant gradients to identify local concentration hot spots from
both stationary and mobile sources is critical for exposure assessments. For example,
individuals who spend more time near busy highways are likely to be exposed to higher
levels of air pollution. To account for this near-road exposure we modelled ambient air
quality concentrations for multiple pollutants resulting from roadway emissions. There
are multiple modelling techniques to simulate near-road dispersion from mobile sources
(Borrego et. al., 2006, Cook et. al., 2006). In this study, we used the AERMOD
dispersion model which treats individual road links as area sources to simulate hourly
concentrations of various pollutants near the road. AERMOD also simulates near-
source impacts from stationary sources. Contributions to photochemical interactions are
provided as a background concentration level from CMAQ, a regional grid model.


-------
Modeling Domain, New Haven, CT

locations of emission sources, roads and census block group ceniroids

West-East (infanttt (km UTM, zano 19)

Figure 1. Modelling domain showing locations of emission sources and model receptors.

A hybrid approach (Isakov et. al., 2007) is a logical and efficient way to combine
regional grid and local plume models. Results of both model simulations are combined
to provide the total ambient air pollutant concentrations. The hybrid approach uses the
appropriate modelling tools to describe different types of sources, making its
application computationally efficient. Furthermore, since local dispersion models are
not resource intensive, this methodology allows the study of local concentration
variability due to changes in several model inputs and physical parameters, helping
gain confidence in the simulation results by encompassing a range of model outcomes.
This constitutes a clear advantage of the hybrid approach, since performing a local
concentration variability estimation using a nested grid model alone would be an
impractical task, especially over larger urban areas. A schematic of the hybrid approach
is shown in Figure 2, where the CMAQ model was used to estimate regional
background concentrations and AERMOD was used to estimate local-scale details for
stationary and mobile sources.


-------
Figure 2. Application of the hybrid approach in New Haven, Connecticut.

Combined model results
for multiple pollutants for
each of the 380 census
block group centroids in
New Haven area

Local impact from stationary sources

Near-road impact from mobile sources

In this study, CMAQ was used to simulate ambient concentrations of several air toxics
(Lueken et. al,. 2006). The CMAQ modeling system was run for an annual period in a
nested mode at 36 and 12 km horizontal grid dimensions using the 1999 National
Emission Inventory and meteorological outputs from 2001 using the MM5
meteorological model. The CMAQ results were extracted for the New Haven
modelling domain to provide regional background concentration values. This regional
background was combined with local concentrations predicted by the AERMOD
dispersion model. The application of this hybrid approach is illustrated in Figure 3
which displays modelled annual average outdoor carbon monoxide concentrations in
New Haven, Connecticut. Predicted CMAQ concentrations at 12 by 12 km resolution
are combined with estimates at 200m receptor resolution from AERMOD.


-------
Stationary sources
(AERMOD)

Mobile sources
(AERMOD)

Regional transport and
chemistry (CMAQ)

Combined
(hybrid approach)







HIPWrl

J***"

Figure 3. Annual average CO concentrations in New Haven, Connecticut: (a) impact of stationary sources; (b)
impact of mobile sources; (c) regional background; (d) combined concentrations using hybrid approach.

3. MODELING APPROACH FOR ACCOUNTABILITY STUDIES

The methodologies developed under this project can be applied to future projects in
other areas to simulate air quality impacts for various controls scenarios. For example:
1) what happens if emissions from some specific stationary sources are reduced by "x"
percent? 2) what happens if emissions from mobile sources could be reduced by "y"
percent? 3) what is the impact of local controls? 4) what is the impact of
regional/national controls resulting in reduction of regional background? Figure 4
provides a hypothetical example of the relative impacts of various control strategies on
ambient concentrations. These examples include: reducing emissions from mobile
sources, controlling emissions from stationary sources, and reducing impact of the
regional background. This example helps determine which control options are most
effective in reducing ambient concentrations. In order to link theSe air quality estimates
to health effects associated with human exposures to environmental pollutants, we will
use exposure modelling. When combined with exposure models, the control strategies
can be assessed to optimize the impacted population, or population subset, such as
children, people with respiratory problems, etc. In this study, we are evaluating
alternative techniques for estimating cumulative exposures to selected air toxics, PM,
and ozone using probabilistic cumulative exposure models: HABEM 6 and SHEE)S-Air
Toxics, time-series based models, using human activity pattern data, modeled/measured
concentrations, and exposure factors.


-------
Figure 4. Example of modelling application to investigate the effect of various control strategies.

Reducing regional background '

Hypothetical Example: Modeling impact of
various controls for stationary sources,
mobile sources, and regional background

Reducing emissions from mobile sources

Reducing emissions from
I "'j mobile and some stationary sources

4. SUMMARY

At the present, most Federal and State air quality implementation plans rely heavily on
ambient modeling study results for targeting emissions reductions. However, the
complexity in the spatial and microenvironmental variation of exposures among the
different population subgroups, especially in the context inter- and intra-urban analysis
of air pollution health effects, could pose several challenges. Thus, integrated air
quality - human exposure modeling provides the means to evaluate the potential health
risks from air pollution exposures and the basis to determine optimum risk management
strategies, while considering scientific, social and economic factors. Ideally, emission
control strategies not only aim at reducing the emissions from principal sources of
targeted pollutants but also to identify those sources and microenvironments that
contribute to greatest portion of personal or population exposures. Recent advances in
exposure modeling tools and better information on time-activity, commuting and
exposure factors data provide unique opportunities for improving the assignment of
exposures during the course of future accountability and community health studies.
Moreover, the combination of sophisticated air quality and exposure models will
improve the accuracy of present air quality and exposure forecasts, and help us better
quantify the health and economic benefits of emissions reductions programs, as part of
air accountability studies.


-------
ACKNOWLEDGEMENTS

Disclaimer: The research presented here was performed under the Memorandum of
Understanding between the U.S. Environmental Protection Agency (EPA) and the U.S.
Department of Commerce's National Oceanic and Atmospheric Administration (NOAA) and
under agreement number DW13921548. This work constitutes a contribution to the NOAA Air
Quality Program. It does not necessarily reflect Agency policies or views.

REFERENCES

Borrego, C., A. Tchepel, A.M. Costa, H. Martins, J. Ferreira, A.I. Miranda, 2006,
Traffic-related particulate air pollution exposure in urban areas. Atmospheric
Environment 40, 7205-7214.

Byun, D.W. and K. Schere, 2006, Review of the Governing Equations, Computational
Algorithms, and Other components of the Models-3 Community Multiscale Air Quality
(CMAQ) Modeling System. Applied Mechanics Review 59, 51-77.

Cimorelli, A.J., S.G. Perry, A. Venkatram, J.C. Weil, R.J. Paine, R.B. Wilson, R.F. Lee,
W.D. Peters, and R.W. Brode, 2005, AERMOD: A Dispersion Model for Industrial
Source Applications. Journal of Applied Meteorology 44, 682-693.

Cook, R., J.S. Touma, A. Beidler, M. Strum, 2006, Preparing Highway Emissions
Inventories for Urban Scale Modeling: A Case Study in Philadelphia. Transportation
Research Part D: Transport and Environment 11, 396-407.

Isakov, V., J.S. Irwin and J. Ching, 2007, Using CMAQ for exposure modeling and
characterizing the sub-grid variability for exposure estimates. JAMC (in press).

Luecken, D.J., W.T. Hutzell, and G.J. Gipson, 2006, Development and analysis of air
quality modeling simulations for hazardous air pollutants. Atmospheric Environment
40, 5087-5096.

Touma, J.S., V. Isakov, J. Ching, and C. Seigneur, 2006, Air Quality Modeling of
Hazardous Pollutants: Current Status and Future Directions. J. Air & Waste Manage.
Assoc. 56, 547-558.


-------