United States
Environmental Protection
Agency
National Risk Management
Research Laboratory
CINCINNATI, OH 45268
Research and Development
EPA/600/SR-98/097 August 1998
Project Summary
A CIS-Based Modal Model of
Automobile Exhaust Emissions
William H. Bachman
Suburban sprawl, population growth,
and auto-dependency have been linked,
along with other factors, to air pollution
problems in U.S. metropolitan areas.
Addressing these problems becomes
difficult when trying to accommodate
the needs of a growing population and
economy while simultaneously lower-
ing or maintaining levels of ambient pol-
lutants. Growing urban areas must,
therefore, continually develop creative
strategies to curb increased pollutant
production.
The report presents progress toward
the development of a computer tool
called MEASURE, the Mobile Emission
Assessment System for Urban and
Regional Evaluation. The tool works to-
ward a goal of providing researchers
and planners with a way to assess new
mobile emission mitigation strategies.
The model is based on a geographic
information system (GIS) and uses
modal emission rates, varying emis-
sions according to vehicle technologies
and modal operation (acceleration, de-
celeration, cruise, and idle). Estimates
of spatially resolved fleet composition
and activity are combined with situa-
tion-specific emission rates to predict
engine start and running exhaust emis-
sions. The estimates are provided at
user-defined spatial scales. A demon-
stration of modal operation is provided
using a 100 km2 study area in Atlanta,
Georgia. Future mobile emissions mod-
eling research needs are developed
from an analysis of the sources of
model error.
This Project Summary was developed
by the National Risk Management Re-
search Laboratory's Air Pollution Pre-
vention and Control Division, Research
Triangle Park, NC, to announce key find-
ings of the research project that is fully
documented in a separate report of the
same title (see Project Report ordering
information at back).
Introduction
Suburban sprawl, population growth,
and auto-dependency have been linked,
along with other factors, to air pollution
problems in U.S metropolitan areas. Ac-
cordingly, the Clean Air Act and other
federal legislation and regulations require
metropolitan areas to develop strategies
for reducing air pollution where air quality
standards are exceeded. An emissions
'budget' is established in metropolitan ar-
eas that provides a benchmark for com-
paring new emission-generating activity,
and presumably not exceeded. Such a
goal becomes difficult when trying to ac-
commodate the needs of a growing popu-
lation and economy while simultaneously
lowering or maintaining levels of ambient
pollutants. Growing urban areas, there-
fore, must continually develop creative
strategies to curb increased pollutant pro-
duction. Because the largest contributor
of pollutant emissions in urban areas has
most often been transportation (or mobile)
sources, transportation is targeted for new
control strategies.
Developing measures of effectiveness
and subsequent predictions of overall
impact for control strategies require an
understanding of the relationship between
observable transportation system charac-
teristics and emission production. Quan-
tifying this effectiveness requires modeling
-------
these relationships. According to published
research, motor vehicle emission rates are
correlated to a variety of vehicle charac-
teristics (weight, engine size, emission con-
trol equipment, etc.), operating modes (idle,
cruise, acceleration, and deceleration), and
transportation system conditions (road
grade, pavement condition, etc.). Exhaust
emissions are produced when a vehicle is
started and while it is operating. Pollut-
ants produced from starting a vehicle can
be predicted using vehicle characteristics.
Running exhaust emissions additionally
require estimates of dynamic engine con-
ditions that result from how the vehicle is
driven. Estimating motor vehicle emissions
requires the ability to predict or measure
these parameters for an entire region at a
level of spatial and temporal aggregation
fitting the scope of control strategies. Cur-
rent modeling approaches, however, can-
not provide these estimates.
Emission Modeling
Today's motor vehicle emission model-
ing process is based on four separate
models: a travel demand forecasting
model, a mobile emission model, a photo-
chemical model (for emission inventory),
and a microscale model (for analyzing
transportation improvements). The travel
demand forecasting model uses charac-
teristics of the transportation system and
socioeconomic data to develop estimates
of road-specific traffic volumes and aver-
age speeds. Mobile emission models use
these travel demand estimates, operating
fleet model year distributions, and envi-
ronmental conditions to develop estimates
of mobile source pollutant production.
These estimates are fed into photochemi-
cal models (along with stationary source
estimates and data regarding atmospheric
conditions) and are used to predict ambi-
ent pollutant levels in space and time.
These mobile source estimates can then
be used by microscale models to predict
pollutant levels near specific transporta-
tion facilities.
Several problems with the four-model
system limit effective evaluation of motor
vehicle emission control strategies. First,
the estimates of vehicle activity (vehicle
miles traveled and average speed) lack
the accuracy and spatial resolution needed
to evaluate control measures. Second, the
mobile source emission rate modeling pro-
cess uses highly aggregate fleet average
emission rates which are not specific for
the fleet in operation, mode of vehicle
operation, or grade of the highway facility.
As a consequence, the current modeling
system has limited capabilities for meet-
ing the modeling requirements of trans-
portation planners. Transportation planners
and environmental assessment and con-
trol officials have need for improved mod-
els that help identify the impacts of
standard transportation system improve-
ments (e.g., lane additions, signal timing,
peak-hour smoothing).
While many researchers agree that new
models and processes need to be devel-
oped to overcome these problems, they
disagree over the best approach. The U.S.
Environmental Protection Agency and the
Federal Highway Administration held a
workshop in Ann Arbor, Michigan, in May
1997, to identify and discuss current emis-
sion modeling research efforts. After the
workshop, it was clear that defining ap-
propriate model aggregation levels is im-
portant in defining how and what research
should be conducted. A point of departure
between the largest vehicle emissions re-
search efforts (University of California at
Riverside, and the Georgia Institute of
Technology) and the currently mandated
approach (MOBILESa) is the level of ag-
gregation required. Figure 1 shows the
spectrum of possible approaches. It shows
that highly aggregate approaches limit ex-
planatory power, but have reduced data
intensity. Disaggregate models have the
most explanatory power, but the highest
data needs. An added dimension of the
issue is that estimates must be spatially
and temporally resolved, suggesting that
an undefined level of spatial and temporal
aggregation must also be defined. In fact,
the level of spatial and temporal aggrega-
tion of mobile source emissions needed
by photochemical models may help define
the minimum level of model aggregation
currently being debated.
The report presents a research model
that can guide future mobile emissions
model development efforts. A major ob-
jective of the model is to incorporate the
latest transportation / air quality findings
at a low level of spatial aggregation (re-
stricted only by data availability). Creating
a model under these guidelines develops
information that leads to the maximum
level of disaggregation given user needs
and data availability. The research model
will be comprehensive, flexible, and user
oriented. It includes enhanced vehicle
activity measures: starts, idle, cruise, ac-
celeration, and deceleration. Vehicle tech-
Emission Rates
Average Fleet
(g/trip)
Vehicle Class
Average Speed
(g/mi)
Tech Groups'
Vehicle Mode
(g/sec)
Tech Groups'
Vehicle Mode
(g/sec)
Individual Vehicles
Vehicle Mode
(g/sec)
Individual Vehicles
Engine Mode
(g/sec)
Aggregate
Total Trips/Day
Vehicle Class
VMTb
Mean Link Speeds
Tech Group
Speed/Accel.
Profiles
Tech Group
Traffic Flow
Simulation
Vehicle Activity
a Tech groups refer to sets of vehicles with similar emission characteristics.
b Vehicle miles traveled.
Disaggregate
Vehicle
Activity
Simulation
Vehicle/Engine
Activity
Simulation
Figure 1. Spectrum of modeling approaches.
-------
nology characteristics (model year, engine
size, etc.) and operating conditions (road
grade, traffic flow, etc.) are developed at
a large scale (small zones and road seg-
ments). Flexibility is achieved through a
modular design that separates emission
production based on thresholds determined
in background research. Due to large gaps
in the state of knowledge, technology, and
practice regarding travel behavior, emis-
sion rates, and the urban system inven-
tory, the accuracy of the model results
remains unvalidated and therefore un-
known. However, the model contributes to
transportation and air quality research in
that it aids research and software devel-
opment.
The intended model users include emis-
sion science experts, model developers,
transportation planners, policy makers, and
government researchers. Each user group
has specific modeling interests that define
how the model should be designed and
presented. Central to the model design is
a geographic information system (CIS).
GISs are widely used computer tools
that allow geographically referenced data
to be organized and manipulated. Both
transportation and air quality vary in spa-
tial dimensions. Thus, GISs have the con-
ceptual capability to manage the
relationships between transportation ac-
tivity and resulting air quality changes
based on their spatial characteristics. Fur-
ther, GISs are already used by most plan-
ning organizations in government
institutions. Thus, a CIS-based emissions
modeling framework fits the character of
emission science as well as the technical
environment of the expected users.
The variables included in the proposed
research model are those whose relation-
ship to vehicle activity and emission rates
has been defined in research and is avail-
able to public agencies. They can be cat-
egorized as:
Spatial Character:
• U.S. Census block boundaries
• Land use boundaries
• Traffic analysis zone boundaries (from
travel demand forecasting model)
• Grid cell boundaries (defined by user)
• Road segments (by classification)
• Travel demand forecasting network
links
• Grade school and university locations
Temporal Character:
• Hour of the day
Vehicle Technology:
• Model year
• Engine size
• Vehicle weight
• Emission control equipment
• Fuel injection type
Modal Activity:
• Idle
• Cruise
• Acceleration
• Deceleration
Trip Generation:
• Home-based work trips
• Home-based shopping trips
• Home-based university trips
• Home-based grade school trips
• Home-based other trips
• Non-home-based trips
Road Geometries:
• Number of lanes
• Grade
Socioeconomic Characteristics (for spa-
tial allocation only):
• Housing units
• Land use (residential, nonresidential,
and commercial)
Summary of Contributions to
Research
• An automobile exhaust emissions
model is developed maximizing com-
prehensiveness, flexibility, and user
friendliness.
Comprehensiveness is ensured by in-
cluding variables and procedures identi-
fied in the literature as significant to
emission rate modeling. Flexibility is
achieved by organizing the model com-
ponents by geographic location, and by
maintaining a modular program design.
User friendliness is achieved by includ-
ing only current data available to plan-
ning agencies, and by using a CIS
framework.
• A research tool is provided that al-
lows for testing variable levels of mo-
tor vehicle emission model spatial
aggregation.
By having the flexibility to use a variety
of spatial entities, the model can become
a testbed for determining the spatial reso-
lution needed for future models. This in-
formation is valuable in identifying future
research needs, costs of emission esti-
mation, model development, maintenance,
and operation. A question this model could
be used to help answer would be, "Given
the current state of research, does a 1
km2 aggregation of ozone precursors pro-
vide enough resolution to predict ozone
formation, or would a 4 km2 aggregation
be better?"
• The benefits of using GIS for emis-
sions modeling are demonstrated.
GISs provide the ability to organize data
by location, in turn providing the capability
to develop relationships with new or exist-
ing spatial datasets. This allows the de-
velopment of creative alternatives to model
construction and provides the ability to
prioritize emission control strategies based
on location.
• Research and data needs for im-
proved spatial and temporal emissions
modeling are identified.
A study of background research into
emissions modeling, coupled with an
analysis of data available in Atlanta, will
determine gaps in important emission-
specific variables. Further, a prioritization
of the data needs based on balancing
explanatory power and cost will guide
future model development.
Report Organization
Chapter 1 presents an introduction to
the report and provides an overview of
the research being presented to the reader.
Chapter 2 discusses background research
significant to automobile exhaust emis-
sion modeling, vehicle activity modeling,
and GISs. It also identifies a research
foundation of knowledge that is used to
develop model parameters. Chapter 3 pre-
sents a conceptual model design that
serves as the foundation of the research
approach. Accuracy, comprehensiveness,
user needs, and enterprise awareness are
important considerations in developing this
conceptual model. Chapter 4 provides a
physical model structure that can be used
as a research tool. The model will reside
in a UNIX operating system and use Make,
the C programming language, and ARC/
INFO. A step-by-step guide to model use
is also provided. Chapters 5 and 6 present
and analyze a model implementation for a
100 km2 area in Atlanta. Each module of
the system is studied using sensitivity
analysis or a comparison of observed data.
Chapter 7 discusses data needs and pre-
sents final conclusions. An expanded
model diagram demonstrates how future
vehicle types and operating modes can
be added to the system. Chapter 8 lists
references cited in the report, and Appen-
dix A is a data dictionary.
-------
I/I/ Bachman is with the School of Civil and Environmental Engineering, Georgia
Institute of Technology, Atlanta, GA 30332.
Carl T. Ripberger is the EPA Project Officer (see below).
The complete report, entitled "A GIS-Based Modal Model of Automobile Exhaust
Emissions," (Order No. PB98-165145; Cost: $44.00, subject to change) will be
available only from:
National Technical Information Service
5285 Port Royal Road
Springfield, VA 22161
Telephone: 703-487-4650
The EPA Project Officer can be contacted at:
Air Pollution Prevention and Control Division
National Risk Management Research Laboratory
U.S. Environmental Protection Agency
Research Triangle Park, NC 27711
United States
Environmental Protection Agency
CenterforEnvironmental Research Information
Cincinnati, OH 45268
BULK RATE
POSTAGE & FEES PAID
EPA
PERMIT No. G-35
Official Business
Penalty for Private Use
$300
EPA/600/SR-98/097
------- |