Jflll
Example Application of Modeling Toxic Air
Pollutants in Urban Areas
-------
EPA-454/R-02-003
JUNE 2002
Example Application of Modeling Toxic Air Pollutants in Urban Areas
U.S. Environmental Protection Agency
Office of Air Quality Planning and Standards
Office of Air and Radiation
Research Triangle Park, North Carolina
-------
DISCLAIMER
The information in this document has been reviewed in accordance with the U.S. EPA
administrative review policies and approved for publication. Mention of trade names or
commercial products does not constitute endorsement or recommendation for their use.
11
-------
TABLE OF CONTENTS
1. METHODOLOGY FOR URBAN AIR TOXICS MODELING 1
1.1 Introduction 1
2. MODELING METHODOLOGY 2
2.1 Model Features 3
2.2 Model Options 4
2.3 Averaging Period 4
2.4 Receptors 4
2.5 Terrain 5
2.6 Meteorological Data 5
2.7 Chemistry 7
2.8 Background Concentrations 7
2.9 Monitoring Data 8
2.10 Study Limitations 8
3. SOURCE CHARACTERIZATION 8
3.1 Modeling Domain 8
3.2 Emission Inventories 8
3.3. Processing emissions data into ISCST3 11
3.4 Source Characterization for ISCST3 12
3.5 Default Source Parameters 15
3.6 Pollutants 17
3.7 Source Grouping 17
3.8 Quality Assurance 17
4. HOUSTON CASE STUDY 17
4.1 Introduction 17
4.2. Model Methodology 18
4.2.1 Model Selection 18
4.2.2 Averaging Period 19
4.2.3 Receptor Selection Strategy 19
4.2.4 Treatment of Terrain Influences 19
4.2.5 Land Use Classification 19
4.2.6 Meteorological Data 20
4.2.6.1 Selection of Surface and Upper Air Stations 20
4.4.6.2 Meteorological Parameters for Deposition Calculations 21
4.2.6.3 Meteorological Preprocessing 21
4.2.6.4 Meteorological Statistics for Houston 22
4.2.7 Chemistry 22
4.2.8 Background 22
4.2.9 Model Evaluation Procedure 23
4.2.10 Study Limitations 23
4.3 Emissions 24
4.3.1 Processing of Emission Data for ISCST3 24
in
-------
TABLE OF CONTENTS (cont.)
4.3.1.1 Gridded emissions 24
4.3.1.2 Allocation of onroad mobile emissions to road segments 27
4.4 Results 34
4.4.1 Benzene 34
4.4.1.1 Emissions 34
4.4.1.2 ISCST3 Results 35
4.4.2 Other HAP's 37
4.4.3 Model to Monitor Comparisons 39
5. SUMMARY AND CONCLUSIONS 39
6. REFERENCES 41
APPENDIX A Estimating Background Concentration for Diesel PM A-l
IV
-------
LIST OF TABLES
Table 4.2-1. Surface and upper air stations used in Houston study 44
Table 4.2-2. 1996 and Climatological Wind Speed, Wind Direction, Average Daily Maximum
Temperature, Average Daily Minimum Daily Temperature, and Annual Total Rainfall 44
Table 4.3-1. Corrected Location Coordinates of Point Sources in Houston Domain 45
Table 4.3-2. Vehicle Split Table for Allocation of Emissions to Road Segments 46
Table 4.3-3. Types and Dimensions of Roadway Segments 47
Table 4.4-1. Maximum annual average total (from all sources) concentration and location for
each HAP in study 47
Table 4.4-2. Maximum Annual Average Concentrations and Locations by Source Category . . 48
Table 4.4-3. Benzene maximum total concentration and location for ISCST3, ISCST3 ROADS,
ISCST3 FINE GRID 48
Table 4.4-4. Formaldehyde Maximum concentrations and concentration by source category for
ISCST3 and ISCST3 ROADS 49
-------
LIST OF FIGURES
Figure 4.1-1. A) Houston study domain with key roads, airport, and b) ISCST3 FINE GRID
location 50
Figure 4.2-1. Location of urban and rural grid cells, ISC receptors, and monitors 51
Figure 4.2-2. Wind rose for Houston, 1996 52
Figure 4.2-3. Annual average background concentrations (jig m"3) for a) benzene and
b) lead 53
Figure 4.3-1. Flowchart of gridded emissions processing for Houston 54
Figure 4.4-1. Benzene 1 km gridded emissions (tons year"1) from all sources 55
Figure 4.4-2. Benzene 1 km gridded emissions for a) major and b) onroad mobile source
emissions (tons year"1) 56
Figure 4.4-3. Distribution of emissions for Houston and U.S. for a) benzene, b) cadmium, c)
chromium, d) formaldehyde, e) lead and g) complete distribution of emissions for all
HAPs 57
Figure 4.4-4. Benzene a) road segment emissions (tons yr"1 km"2) and b) remaining 1 km gridded
onroad mobile emissions (tons yr"1) after extracting road segment emissions 63
Figure 4.4-5. Benzene ISCST3 BASE annual average total concentrations (jig m"3) using
gridded emissions 64
Figure 4.4-6. Benzene annual average ISCST3 BASE concentrations (jig m"3) using gridded
emissions for a) major, b) area, c) onroad mobile, and d) nonroad mobile sources .... 65
Figure 4.4-7. Benzene annual average ISCST3 ROADS concentrations (jig m"3) for a) total, and
b) onroad mobile concentrations 67
Figure 4.4-8. Benzene percent differences for ISCST3 ROADS minus ISCST3 BASE for a) total
concentrations and b) onroad mobile concentrations 68
Figure 4.4-9. Largest source contributor at each receptor for benzene a) ISCST3 BASE
and b) ISCST3 ROADS 69
Figure 4.4-10. Benzene ISCST3 FINE GRID annual average total concentrations (jig m"3) a)
scaled to 98th percentile of concentrations and b) scaled to maximum concentration ... 70
VI
-------
LIST OF FIGURES (cont.)
Figure 4.4-11. Benzene ISCST3 FINE GRID annual average major source concentrations
(jig m"3) a) scaled to 98th percentile of concentrations and b) scaled to maximum
concentration 71
Figure 4.4-12. Benzene ISCST3 FINE GRID annual average arear source concentrations
(jig m"3) a) scaled to 98th percentile of concentrations and b) scaled to maximum
concentration 72
Figure 4.4-13. Benzene ISCST3 FINE GRID annual average onroad mobiler source
concentrations (jig m"3) a) scaled to 98th percentile of concentrations and b) scaled to
maximum concentration 73
Figure 4.4-14. Benzene ISCST3 FINE GRID annual average nonroad mobiler source
concentrations (jig m"3) a) scaled to 98th percentile of concentrations and b) scaled to
maximum concentration 74
Figure 4.4-15. Cadmium 1 km gridded emissions (tons yr"1) for a) all sources and b) area
sources 75
Figure 4.4-16. Cadmium ISCST3 annual average a) total concentrations (jig m"3) and b) largest
source contributor at each receptor 76
Figure 4.4-17. Chromium 1 km gridded emissions (tons yr"1) from all sources 77
Figure 4.4-18. Cadmium ISCST3 annual average a) total concentrations (jig m"3) and b) largest
source contributor at each receptor 78
Figure 4.4-19. Formaldehyde 1 km gridded emissions (tons yr"1) from all sources 79
Figure 4.4-20. Formaldehyde road segment emissions (tons yr"1 km"2) 80
Figure 4.4-21. Formaldehyde 1 km onroad mobile gridded emissions (tons yr-1) for a) ISCST3
ROADS onroad mobile gridded emissions after extracting road segment emissions and b)
ISCST3 BASE onroad mobile gridded emissions 81
Figure 4.4-22. Formaldehyde ISCST3 BASE annual average concentrations for a) all sources
and b) onroad mobile sources 82
Figure 4.4-23. Formaldehyde ISCST3 ROADS annual average concentrations for a) all sources
and b) onroad mobile sources 83
vn
-------
LIST OF FIGURES (cont.)
Figure 4.4-24. Formaldehyde percent differences for ISCST3 BASE minus ISCST3 BASE for
a) total concentrations and b) onroad mobile concentrations 84
Figure 4.4-25. Largest source contributor for each receptor for formaldehyde for a) ISCST3
BASE and b) ISCST3 ROADS 85
Figure 4.4-26. Lead 1 km gridded emissions (tons yr"1) for a) all sources receptor, b) area
sources, and c) nonroad mobile sources 86
Figure 4.4-27. Lead ISCST3 annual average concentrations for a) all sources b) area sources, and
c) nonroad mobile sources 88
Figure 4.4-28. Largest source contributor at each receptor for lead 90
Figure 4.4-29. ISCST3 BASE (star), ISCST3 ROADS (circle) and monitor (box) annual average
concentrations (g m-3) for benzene 91
Vlll
-------
1. METHODOLOGY FOR URBAN AIR TOXICS MODELING
1.1 Introduction
This document provides an example application of modeling toxic air pollutants in urban
areas. In preparing this document, it was necessary to revise and update techniques described in
Dispersion Modeling of Toxics Pollutants in Urban Areas, EPA-454/R-99-021 and incorporate
techniques developed in A Simplified Approach for Estimating Secondary Production of
Hazardous Air Pollutants (HAPs) Using the OZIPR Model, EPA-454/R-99-054.
The 1990 Clean Air Act Amendments (CAAA) Section 112(k) requires EPA to reduce
hazardous air pollutant (HAP) risks in urban areas. The EPA's strategy for reducing these risks
is discussed in the Integrated Urban Air Toxics Strategy (U.S. EPA, 1999b). In order to help
understand the air toxics problem in an urban area, it is necessary to know the concentrations of
air toxics to which people are exposed. Air monitoring data are scarce and limited. Another
means for estimating HAP concentrations is through the use of air quality models. Urban areas
can vary greatly in terms of the types of emission sources and the legal enforcement options
provided by state and local programs to control air toxic emissions and air quality models also
allow state and local agencies to test the effectiveness of alternative control measures in reducing
ambient concentrations.
The intent of urban wide air toxics modeling applications is to provide data inputs for use in
exposure and risk calculation and prioritization, obtain a higher degree of geographic resolution
than those obtained from national scale studies, identify data gaps and help allocate resources
toward particular issues of interest or concern, and support the planning and implementation of
ambient air monitoring programs. Regarding the higher geographic resolution of assessment
results, an important benefit of refinement by an urban scale application is illustrated by
considering the methods by which emissions data are applied to the models. To achieve the
objective of a national scale assessment within a feasible scope of time and resources,
assumptions about emissions allocation are typically made to simplify the modeling. An
example would be the allocation of unknown emission source locations to the centroid of the
census tract. While this approach allows the national scale assessment to broadly identify
pollutants of potential concern across large geographical areas, the assigning of source locations
or other sensitive parameters in such a manner limits resolution. The urban scale modeling effort
is more localized and can compliment the national scale assessment by increase in specificity.
However, at this time, the number of assumptions made in urban scale modeling precludes the
use in specifying the individual sources that contribute to the total concentrations or the impact of
specific sources in specific areas such as a neighborhood. For determining impacts of specific
sources, more detailed analysis than is warranted in an urban scale application is needed.
Urban areas contain major sources, numerous smaller, area sources, and mobile source. As
a result, modeling analyses for large numbers of air toxics sources possess special challenges.
Although most HAPs are emitted directly, some are produced and destroyed through reactions in
-------
the atmosphere. These issues, as well as receptor selection, meteorological data processing, and
background concentration selection pose significant technical challenges to the air quality
modeler. Although many air quality models can be used for estimating urban wide ambient
concentrations, this document deals with the applications of the Industrial Source Complex
(ISCST3) model, a model that can estimate close distance impacts from industrial facilities. This
model has been extensively used in analyzing impacts from a single or a few facilities and this
report should help provide transition to the more complex issues associated with urban-wide
applications.
Sections 2 and 3 provides recommendations on specific issues needed for urban-wide air
quality modeling of air toxic pollutants. Section 2 focuses on modeling methodology and section
3 focuses on emissions and source characterization issues. Section 4 provides an overview of an
application of a Gaussian model to an urban-wide study, i.e., the Houston, Texas urban area.
Section 5 provides study summary and conclusions.
2. MODELING METHODOLOGY
The extent to which a specific model is suitable for the evaluation of source impact depends
on several factors that include the meteorological and topographic complexities of the area; the
level of detail and accuracy of the data base (i.e., emissions inventory), and the resources
available.
There are a number of design criteria which need to be satisfied in order to yield an
acceptable modeling study of toxic pollutants. For the air dispersion model, for example, these
include:
1. readily available/public domain/recommended by EPA
2. represents state-of-modeling practice
3. applicable to urban areas and irregular terrain
4. capable of handling point, area and mobile sources
5. capable of accounting for dry and wet deposition of pollutants
6. capable of treating atmospheric chemical transformations - pollutant chemistry
7. capable of accounting for pollutant emissions that vary by season and hour-of-day
8. ability to group source types for assessing impact
9. capable of providing annual average concentration estimates (as well as shorter time
averages)
10. computationally efficient
11. demonstrate good performance with measurements - estimated vs. observed concentrations.
The Gaussian plume model is a widely used technique for estimating the impacts of
nonreactive pollutants because of its good performance against field measurements, and because
it is computationally efficient relative to other types of models, such as grid and puff models.
-------
The plume dispersion model used in estimating urban-wide concentrations of toxic air
pollutants is the ISCST3 model. This section describes some of the most important
considerations needed to apply ISCST3 for these types of applications. Other sources of detailed
information are listed below:
For general information on air quality modeling, consult Appendix W to CFR Part 51-
Guideline on Air Quality Models.
For information on how to use the ISCST3 model, consult the ISC3 model user's guide
(U.S. EPA, 1995b).
For information on how to preprocess the meteorological data for input in ISCST3, consult
the MPRM User's Guide (U.S. EPA, 1996a) and PCRAMMET User's Guide (U.S. EPA,
1996b).
For information on chemical parameters required for estimating deposition, consult Wesely
et al., 2002.
For information on estimating secondary production of hazardous air pollutants, consult
U.S. EPA, 1999c.
All of the items listed above can be obtained from EPA's SCRAM web site at
http ://www. epa.gov/ttn/scram.
Information on the "Integrated Urban Air Toxics Strategy" developed under the authority of
Section 112(k) and 112(c) of the Clean Air Act is obtained from EPA's web site at
http://www.epa.gov/ttn/atw/index.html.
2.1 Model Features
Key features of the ISCST3 dispersion model that are useful for urban-wide air toxics
applications include:
• handles multiple point, area, and mobile sources
incorporates building downwash effects
• includes an urban dispersion option
contains considerable flexibility for specifying receptor locations and for grouping of source
impacts
includes algorithms to treat the effects of elevated and/or complex terrain
• treats the effects of deposition of gaseous and particulate emissions
includes an option to vary emissions by season and hour-of-day
• includes an option to treat atmospheric transformations by exponential decay.
-------
2.2 Model Options
The regulatory default model option in ISCST3 should be selected for urban-wide
applications. More information about the default option parameters can be found in the ISC3
User's Guide (U.S. EPA, 1995b).
The option to vary the emissions by season and hour-of-day should be selected, unless the
objectives of the application or the form of the emissions inventory data dictate otherwise.
For best estimates, the use of the wet and dry deposition and plume depletion options should
also be selected. Deposition estimates are very useful in multi-pathway exposure assessments.
Note that the selection of the plume depletion option will increase model run time. To utilize the
wet and dry deposition option, the model requires additional data for the meteorological and
chemical parameters. For meteorological input, the user should consult the meteorological data
preprocessors' user's guides (see above).
To determine whether the modeling domain satisfies the criteria for an "urban" or "rural"
area, Section 8.2.8 of the Guideline on Air Quality Models (40CFR51) should be followed. The
domain can be subdivided into urban and rural areas based on land use data. Sources in these
areas are modeled separately and concentrations from each model run are then added at each
receptor.
2.3 Averaging Period
The ISCST3 model computes an hourly concentration for each receptor. Other averaging
periods, e.g., 3-hour, daily, seasonal and annual can also be aggregated (U.S. EPA, 1995b). The
averaging period selected is based on the intended use. Annual average air concentrations are
generally needed for use in chronic (long-term) exposure studies. Shorter term ambient
concentrations are usually needed for determining acute exposure. However, it should be noted
that sometimes the input data (i.e., emissions) may not have the temporal resolution needed for
short term concentrations.
2.4 Receptors
A receptor is any location where ambient concentration estimates are needed. Receptors are
usually placed in "ambient air" which is outside of inaccessible plant property. For point and
area sources, placement is usually at the fence line and for mobile sources placement is near
roadways. The ISCST3 model requires the coordinates of the specified receptors. Receptor
locations should be selected based on a case-by-case determination with expert judgement on the
needs of the study. Often, receptors are selected at coordinates provided in the census data
(census block, census block groups or census tracts). Other receptor locations include ambient
air quality monitoring site locations.
-------
Census data and urban land use information can be used to identify locations (potential
receptors) where individuals live, work, attend school, and spend time in recreation. Since model
run time is proportional to the number of receptors, a degree of care is needed to select the
optimal number of receptors. In the example study, receptor selection was based in part, on the
input needs of the Hazardous Air Pollutant Exposure Model (HAPEM4) exposure model (U.S.
EPA, 1999a). The receptor points were defined as the population weighted centroid of each
census tract.
2.5 Terrain
Terrain elevation at each source and receptor must be input into ISCST3. Digitized terrain
elevation data are available through the U.S. Geological Survey (USGS) web site at
http://edcwww.cr.usgs.gov/doc/edchome/ndcdb/ndcdb.html. Their l:24,000-Scale (7.5-minute)
Digital Elevation Model (DEM) data can be downloaded directly, at no cost, from the GeoComm
International Corporation (GCIC) at http://gisdatadepot.com/dem and from MapMart.com at
http://www.mapmart.com. Data can also be ordered and shipped on CD ROM for a nominal cost
from these vendors. The free and purchasable data from GCIC are in the new USGS Spatial Data
Transfer Standard (SDTS) format and must be converted to the old DEM file format to work in
the EPA terrain preprocessor AERMAP (U.S. EPA, 1998). Data in the old DEM format are
available for relatively nominal cost from MAPMART. MAPMART does offer a limited
number of the old DEM formatted data files to be downloaded for free. ISCST3 model users can
use AERMAP to enter the receptor locations and retrieve the elevation data from the AERMAP
output1. In the AERMAP download package, there is a conversion procedure for converting
SDTS formatted data to the old DEM format for input into AERMAP.
Source (stack) elevation is usually provided in the inventory. For many urban areas, the
majority of emission sources are near ground level. In these cases, terrain can be assumed to be
flat and source and receptor elevations set to zero. Where the urban area is in mountainous
terrain, terrain effects are important for sources with stacks. First, the impact of individual
plumes on elevated terrain results in higher air concentration (through placing the receptor at the
correct higher air concentration, vertical location within the plume and estimating the impaction
of the plume upon intervening terrain). Second, wind channeling due to terrain can cause higher
air concentrations. The ISCST3 model does not address wind channeling effects other than if
these effects are captured by the available meteorological data. If the urban area contains
complex terrain features that are expected to significantly affect the modeled concentrations, a
dispersion model that handles such situations should be selected from those listed in the
Guideline on Air Quality Models (40CFR51).
2.6 Meteorological Data
Meteorological data must be preprocessed before use in ISCST3. The ISCST3 model
lrThe AERMAP format must be slightly modified to eliminate extraneous data.
5
-------
requires two meteorological data sets, surface and upper air. Hourly surface and twice-daily
upper air meteorological data files can be purchased online from the National Climatic Data
Center (NCDC) (http://www.ncdc.noaa.gov). The data can also be purchased on CDs. The CDs
with surface data are SAMSON (Solar and Meteorological Surface Observation Network),
HUSWO (Hourly U.S. Surface Weather Observations), INSWO (International Surface Weather
Observations), and ISHD (Integrated Surface Hourly Data). The CD with upper air data is titled
Radiosonde Data of North America. The EPA meteorological preprocessors are currently
designed to process meteorological data from the CDs. For data purchased online, reformatting
is required before use in the EPA preprocessors.
If wet deposition estimates from ISCST3 are required, precipitation data are necessary. The
ISHD data contains hourly precipitation measurements.
In urban areas, on-site meteorological data are not often available. The closest NWS
stations may not be the most representative due to the influences of terrain or water bodies.
Consult with the State/Regional meteorologist for the most applicable NWS stations for your
area.
Mixing heights are computed using surface and upper air data (Radiosonde Data of North
America) via the mixing height program provided on the SCRAM web site
(http://www.epa.gov/scram001).
The PCRAMMET and MPRM meteorological data preprocessors use surface and mixing
height data as input to create ISCST3 input files. PCRAMMET and MPRM can accept data
directly from the SAMSON and HUSWO CDs. In addition, MPRM can also accept INSWO and
ISHD data. MPRM should be used to prepare the input files necessary for applying the gas
deposition algorithm in ISCST3. Values for additional parameters needed in applying the gas
deposition algorithms for the case study city are presented in Section 4.2.6.2. MPRM can also be
used for setting up a meteorological data file for ISCST3 to use in estimating both particle dry
deposition, and gas and particle wet deposition.
Both MPRM and PCRAMMET meteorological data preprocessors can occasionally produce
very low mixing heights (less than 10 meters) based on the twice-daily values from the mixing
height data file and the interpolation scheme used to provide hourly values of mixing height.
Anomalously low calculated mixing heights may be associated with a midday cold frontal
passage. While the occurrence of very low mixing heights is more likely for the rural mixing
heights than for the urban mixing heights, due to differences in the interpolation routines, low
mixing heights may occur for both rural and urban conditions. The application of a very low
mixing height with a near-surface level area source can produce anomalously high air
concentrations due to the treatment of limited mixing effects in the ISCST3 model; expert
judgement is needed to determine the minimum mixing height for a given urban area. For the
example study, a minimum value of 100 meters was applied to the hourly mixing heights
produced by MPRM to avoid this anomaly from influencing the results. For urban areas,
-------
building heights will limit the lower mixing heights and the 100-meter value was considered the
upper limit to the minimum value for the depth of the well-mixed boundary layer in a large urban
area(Sutton, 1953).
2.7 Chemistry
The ISCST3 model provides concentration estimates due to primary emissions and has a
limited capability to consider atmospheric transformations by exponential decay (half-life).
Some pollutants (e.g., formaldehyde, acetaldehyde, and acrolein) are also formed in the
atmosphere due to reactions among other pollutants (i.e., formed by secondary production).
Thus, in addition to estimating concentrations due to primary emissions, an estimate of
concentrations based on secondary production is necessary and should be added to the ISCST3
output in order to avoid large underpredictions. EPA's OZIPR screening model (Gery and
Crous, 1991) may be used to estimate the secondary transformation of acetaldehyde,
formaldehyde and acrolein. U.S. EPA, 1999c describes an approach where secondary HAP
production is estimated with the stand-alone OZIPR model that incorporates only nondispersive
processes, such as photochemistry and the results from this model are then coupled with output
from the ISCST3 model, that accounts for dispersion but not for chemical transformation. The
study results were encouraging because, in comparisons with available monitoring data, this
simple approach seems to perform as well as more complex models.
2.8 Background Concentrations
Background air quality includes pollutant concentrations due to natural sources, nearby
sources other than those under consideration, and unidentified sources. For typical exposure
assessments, background concentrations should be added to the modeled concentrations to
provide total ambient air concentrations for estimating exposure. Air quality data from a HAP
monitoring network in the vicinity of the analysis area are often used to establish background
concentrations. Also, background concentrations of some air toxics may be found in the
literature.
The following approach for estimating background concentrations in the absence of
measured or other reported values can be used. An expanded point source inventory can be
obtained for an area surrounding each city from the National Toxic Inventory (NTI). The domain
for this expanded point source inventory should extend beyond the domain of the inventory being
explicitly modeled in the analysis. An estimate of background concentrations at each receptor
within the modeling domain is obtained by multiplying the point source emission rate by a
distance dependent factor; sources less than 50 km are excluded. The modeled background
concentration can be based on a summation of concentrations computed from a grid across the
modeling domain. These background concentrations can then be added to the modeled
concentrations. See Section 4.2.8 and Appendix A for details of estimating background
concentrations for the Houston study.
-------
2.9 Monitoring Data
Monitoring data can be used to check the validity of the modeled concentration estimates or
determine background concentrations. Ideally, the monitoring and modeling data should span the
same time period. Air toxics monitoring data are available from EPA's Aerometric Information
Retrieval System (AIRS) web site at http://www.epa.gov/airs. In most instances, ambient data
are collected at a frequency of one in six days. A variety of statistical tests can be used to
compare modeled with observed estimates. Statistical tests, such as root mean square (RMS)
errors, can be used to evaluate the model performance against monitor values. How the model
estimates compare to annual average monitored data is useful for determining the suitability of
the estimates. For comparisons in urban areas, there are many uncertainties in all facets of the
comparison effort. For model evaluation studies, a factor of two agreement between modeled
and observed values is generally considered to be acceptable.
2.10 Study Limitations
As part of the conclusions in an urban-wide air toxics modeling study report, the limitations
of the modeling effort should be clearly stated. The important limitations of the ISCST3 model
are provided in the User's Guide (U.S. EPA, 1995b) and the Guideline on Air Quality Models
(40CFR51). Limitations due to data availability and other factors should also be described.
3. SOURCE CHARACTERIZATION
3.1 Modeling Domain
The urban area domain should be selected based on case-by-case determination with expert
judgement. The urban area domain can include a city center or multiple counties. It should be
carefully defined because the larger the modeling domain, the greater the number of sources and
receptors to be considered and thus the greater the required computational resources. Guidance
in the Integrated Urban Air Toxics Strategy (U.S. EPA, 1999b online at
http://www.epa.gov/ttn/uatw/urban/frl9jy99.html) should be consulted.
3.2 Emission Inventories
The first step of the urban-wide dispersion modeling process is the assembly of the
emissions inventory with the specific air toxics emitted and the sources of their airborne
emissions. Ideally, the emission estimates are from direct measurements of representative source
emissions. Although such measurements are likely to provide the most accurate data for an
emission source, these data are typically not available because such sampling is often too time
and resource-intensive. When specific emission measurements are not feasible or available,
other emission estimation methods, including material balances and emission factors, are
sometimes used as an alternate method. Emission factors indicate the quantity of a pollutant
typically released to the atmosphere for a particular source operation, and are usually considered
-------
to be representative of an industry or emission type as a whole. Each approach to estimating
emissions, including use of direct measurement data, has an inherent level of uncertainty, which
adds to the overall uncertainty of a risk analysis.
A national emissions inventory of toxic pollutants developed by EPA is a good starting point
for gathering the necessary emissions data for an urban-scale assessment. The national toxics
inventories compiled by EPA contain emissions of the 188 air toxics listed in section 112(b) of
the CAA. The 1996 National Toxics Inventory (NTI)2 is the first national modeling emission
inventory constructed using state and local HAP inventory data and containing stationary and
mobile source data. EPA prepared the 1996 NTI using five primary sources of data: (1) state and
local air toxics inventories developed by state and local air pollution control agencies, (2)
existing databases related to EPA's Maximum Achievable Control Technology (MACT)
program which requires emission standards under Section 112(d) of the CAA.
(www.epa.gov/ttn/uatw/eparules.html) (3) Toxics Release Inventory (TRI) data
(www.epa.gov/tri/), (4) emissions estimated by using mobile source methodology developed by
experts in EPA's Office of Transportation and Air Quality, and (5) emission estimates for 30 of
500 non-point source categories generated using emission factors and activity data. Much of the
state/local, TRI, and EPA MACT emissions data may have been generated by the sources
themselves. Documentation for all emissions estimates in the 1996 NTI is available on
http://www.epa.gov/ttn/chief/nti/index.htmWnti. The following provides a brief summary of the
data contained in the NTI.
All of the raw inventory inputs in the NTI exist as estimates for point sources, non-point
stationary sources, and mobile sources. "Point" sources provide emissions data at the facility and
sub-facility level and include location coordinates (e.g., latitude and longitude). "Non-point"
stationary source and "mobile" source data exist as emissions estimates for an entire source
category aggregated to the county level. Inventory data files for these different types of sources
are generally maintained separately and include different data elements. For the purpose of
aggregating air toxics emission sources in the urban wide assessment in the example application,
all emissions inventory inputs were grouped into four sectors: "major," "area and other,"
"onroad," and "nonroad." Each sector is further defined as follows:
Major sources are large stationary sources that emit more than 10 tons per year of any listed
air toxic (CAA, section 112(b)) or a combination of listed air toxics of 25 tons per year or
more. Typical examples of major sources include electric utility plants, chemical plants,
steel mills, oil refineries, and large hazardous waste incinerators. These sources may release
air toxics from equipment leaks, when materials are transferred from one location to another,
or during discharge through emissions stacks or vents.
Area and Other sources are smaller stationary sources that emit less than 10 tons per year
of a single air pollutant or less than 25 tons per year of a combination of air toxics. The
2 At the time of writing this report, the NTI has been replaced with the National Emissions Inventory (NEI).
-------
emission inventory includes facility data for some area sources and aggregated emission
estimates at the county level for the remaining area sources. Typical examples of area
sources include neighborhood dry cleaners and gas stations. Though emissions from
individual area sources are often relatively small, collectively their emissions can be of
concern particularly where large numbers of sources are located in heavily populated areas.
"Other" stationary sources are sources that may be more appropriately addressed by other
programs rather than through regulations developed under certain air toxics provisions
(sections 112 or 129) in the Clean Air Act. Examples of other stationary sources include
wildfires and prescribed burning, which have emissions that are being addressed through the
burning policy agreed to by the EPA and the USDA. For this assessment, the "area" and
"other" sectors have been combined in the calculations and presentation of the current
national-scale assessment.
Onroad mobile sources comprise vehicles used on roads and highways (e.g., cars, trucks,
buses).
Nonroad mobile sources are all remaining mobile sources (e.g., trains, lawnmowers,
construction vehicles, farm machinery). Note that airport data are handled separately as
major sources.
In the NTI, major and area source facilities are drawn from the "point" source inventory
files, meaning those with known geographic locations (i.e., latitude and longitude). Area and
other source categories that are aggregated as county-level emissions are drawn from the "non-
point" source inventory files, meaning those stationary sources that do not have location
coordinates but instead exist as county-wide total emissions by source category. Onroad and
nonroad sources exist as distinct sectors in the "mobile" source inventories and are also
aggregated to the county level.
As explained earlier, a primary source of data in EPA's toxics emission inventory is an
inventory developed by state and/or local air pollution control agencies. Thus, the data in EPA's
inventory is, meant to be, at least in theory, locale-specific. However, a number of states and
local agencies do not submit inventories and therefore, the data are from the other sources and
thus to some extent, based on National estimates. In addition, the facility-specific data submitted
may be lacking geographic coordinates or have erroneous ones, and may also not have facility-
specific emission release source characteristics. Furthermore, the level of specificity may still
not be sufficient for an accurate urban scale modeling assessment. For example, large industrial
sources (e.g., paper mills, refineries, etc.) may be grouped together so that hundreds of individual
release points are assumed to exit from a few groups of stacks. If an inventory does not contain
the individual location and release parameters, the analyst has little choice but to model the
source as a group. Ambient concentrations from such facilities should be viewed with caution,
especially at nearby receptors. If one assumes that ground level fugitive releases (e.g., leaks from
pumps, seals or compressors, spilled liquids that form a puddle and then evaporate, lagoons, etc.)
exit through an elevated stack, ground level concentrations will be significantly underestimated.
10
-------
Thus the existing information should be analyzed and, to the extent possible, more detailed
information should be gathered at the local scale. Such additional information includes the
temporal pattern of emissions (e.g., periodic "puffs" vs. constant emission rates), the specific
locations and emission release characteristics of individual small sources such as dry cleaners
and gas stations (which for many local areas are represented by a county sum in the national
inventory), and more detailed information on the individual specific release characteristics for
large sources with hundreds of release points (e.g., whether release is from a stack or fugitive
source, and the associated required release parameters).
In summary, while the national inventory is a starting point, it should be used with
considerable analysis and supplemented with local data. It is important to understand the
national inventories developed by EPA, determine the extent that local data exists for the urban-
scale domain of interest, analyze the data for missing or erroneous features, and
supplement/correct it.
3.3 Processing emissions data into ISCST3
Before emissions are used as input to the ISCST3 dispersion model, the emissions data
require significant preparation. Some of this preparation occurs during the compilation of the
inventory and some occurs in the Emissions Modeling System for Hazardous Air Pollutants
(EMS-HAP), which is a series of computer programs that process emission inventory data for
subsequent air quality modeling (U.S. EPA, 2000a). The necessary inventory preparation steps
are described below:
• Compiling detailed, quality assured, air toxics emissions inputs for all known stationary
and mobile sources.
• Grouping individual pollutant species into compound groups. The NTI contains
approximately 400 different species representing the 188 air toxics listed in section
112(b) of the CAA. Many of the species belong to compound classes. Grouping of these
species is necessary for many reasons. One reason is that the individual chemical species
belonging to groups are not geographically representative. For example, "lead oxide"
may have been reported in just a few counties, whereas other counties aggregated their
lead oxide emissions into "lead compounds." Second, grouping allows for pollutants
with similar characteristics to be modeled together for purposes of efficiency. For
example, specific lead species and compounds reported as the broad group "lead
compound" are grouped to be subsequently modeled as "lead compounds-fine" and "lead
compounds-coarse." Third, grouping decisions made for urban-scale assessment reflect
"downstream" data needs, such as making the resultant concentration estimates reflect
compounds for which health benchmark information exists.
• Temporally allocating emission values by season and day of week to 24-hours emission
rates. Emissions are temporally allocated based on the type of source using a database of
temporal profiles by source classification code.
11
-------
• Grouping into desired source category groups to be able to determine relative contribution
of concentrations for each, e.g., major, area, onroad and nonroad.
• Spatially allocating county-level emissions tol km grid cells using surrogate data, such as
population, industrial land or roadway miles. Note that the appropriate surrogate data for
the urban area need to input into EMS-HAP (U.S. EPA, 2000a) and EPA has developed
available surrogates for national scale modeling assessments. For Urban wide analysis,
users should develop local scale surrogates using site-specific information. Emissions for
a source category are computed based on the percentage of the matching surrogate in the
grid cell for that county. For example, the consumer products usage source category is
matched to population. If 10 percent of the population of the county is in grid cell A,
then this grid cell gets 10 percent of the county's consumer products usage emissions.
However, allocation is unnecessary where local activity data is available (i.e., travel
demand models or local business surveys).
3.4 Source Characterization for ISCST3
Generating the source inventory for modeling is intertwined with the creation of the
pollutant inventory. Each emissions source and the constituents each source emits must be
specifically identified. For the ISCST3 dispersion model, each source will need to be classified
as a point, area, volume, or line source. Building the source inventory usually begins with
mapping the locations of emission sources, receptors and the study domain.
The selection of either urban or rural dispersion coefficients is based on commercial and
industrial land use classifications (Guideline on Air Quality Models). For large individual major
sources, the selection is based on land use classification within a 3 km radius. For other more
numerous area and mobile sources, the designation can be made based on the predominant land
use type in the 1 km grid cell in which the source resides3. Sources located in an area defined as
urban should be modeled using urban dispersion parameters and sources located in areas defined
as rural are modeled with rural dispersion parameters and resulting concentrations at each
receptor are added together.
The ISCST3 model can accommodate a large number of sources and receptors, however, an
optimum configuration is needed in order to minimize computer resources. Because source
inputs vary with the type of source modeled, an important first step in creating the inventory is to
identify each source of emissions as a point, area, volume, or line source. In ISC, line sources are
not modeled, so line sources are modeled as a series of area sources as explained below. With
the source types established, the appropriate model inputs can be determined. The following
subsections describe the various source types and associated inputs for modeling. EMS-HAP
3In making these determinations, the user should examine land use in adjacent grid cells and use judgement to create
a "broad brush" view of the land use in the urban area; a "checkerboard" pattern is not meaningful.
12
-------
(U.S. EPA, 2000a) has been designed to assist in the preparation of the NTI data as input in
ISCST3. Additional information is available from the EMS-HAP user's guide (U.S. EPA,
2000a).
ISCST3 Point Source Characterization
Point sources involve the release of emissions from a well-defined stack or vent, at known
physical stack parameters and operational conditions. Consequently, characterizing point sources
for modeling is fairly straightforward. The basic model inputs for any point source are: stack
location coordinates, the physical stack parameters (height above ground level and inside
diameter at stack exit); operational conditions (gas velocity and temperature at stack exit);
building dimensions (height, width, depth), and emission rate. In situations where a major source
has multiple stacks and buildings, the individual locations of each should be used in the model.
ISCST3 Area Source Characterization
The definition of an area source for ISCST3 modeling is not the same as the CAAA area
source definition in Section 3.2. Area sources are sources of air toxic pollutants that are emitted
at or near ground level (e.g., landfills, waste lagoons, evaporation and settling ponds, nonroad
mobile sources, etc.). Onroad mobile sources can also be characterized as area sources when
specific roadway emissions are not available. The sizes of these sources can range from a few
square meters in the case of settling ponds, to a few square kilometers or larger in the case of
landfills. Emissions from area sources are assumed to be of neutral buoyancy. Therefore, plume
phenomena such as downwash and impaction on elevated terrain features are not considered
relevant for modeling area sources. The emission rate for area sources is in units of mass per unit
time per unit area [e.g., g s"1 m2]. It is an emission flux rather than an emission rate. As an
example, assume the pollutant emission rate from a small lagoon is 150 g s"1. The dimensions of
the lagoon are 10m by 20 m (total area is 200 m2). If this source were modeled as a single,
square area source, then the modeled emission flux would be 0.75 g s^m"2 (150 g s"1 + 200 m2).
In ISCST3, area sources can be modeled in two ways: 1) with known locations and
dimensions (e.g., landfills, airports, etc.); and 2) allocated to 1 km grid cell locations in the
county when the actual location is not available (e.g., dry cleaners). Obtaining the actual location
of the latter type sources and modeling them as point sources will result in better air quality
estimates.
For dispersion modeling, the important parameters used to characterize area sources are
location, geometry (this includes SW corner, initial vertical dimension, and angle of rotation),
and release height. If the area source is not at ground level, a height for the source may be
entered (for example, a non-zero value would typically be entered for the height of a land fill). If
the release height of the source is greater than approximately 10 m, it should probably be
13
-------
modeled as a volume source.
ISCST3 Volume Source Characterization
There are two basic types of volume sources: surface-based or ground-level sources that may
also be modeled as area sources, and elevated sources. As with area sources, emissions from
volume sources are assumed to be of neutral buoyancy. The effective emission height of a
surface-based volume source, such as a surface rail line, is usually set equal to zero. An example
of an elevated volume source is an elevated conveyor with an effective emission height set equal
to the height of the conveyor. A source may be defined as a volume source for modeling when
its emissions can be considered to occur over a certain area and within a certain depth of space.
At refineries, fugitive exhaust from on-site structures such as tanks, or a treatment facility may be
modeled as a volume source. Release area, base elevation and area are needed for modeling. A
roadway over which contaminated soil is hauled may also be modeled as a series of volume
sources.
The important parameters used to characterize volume sources for dispersion modeling are
location, release height and initial lateral and vertical dimensions. The ISCST3 model user's
guide has instructions on defining the initial lateral and vertical dimensions of the source. The
length of the side of the volume source will need to be known, as will the vertical height of the
source, and whether it is on or adjacent to a structure or building. The north-south and east-west
dimensions of each volume source must be the same. For refined modeling, the location is
simply expressed by a single east-west (X) and north-south (Y) coordinate.
ISCST3 Line Source Characterization
Line sources are typically used to represent roadways. For specifically estimating
concentrations of nonreactive pollutants from highway traffic at adjacent receptors (hot spots) a
dispersion model that handles such situations should be selected from those listed in the
Guideline on Air Quality Models (40CFR51). Basic model inputs are the overall source length,
width, and height
In ISCST3, toxic pollutants from line sources are simply modeled as a series of area or
volume sources. In the case of a long and narrow line source, it may be impractical to divide the
source into N volume sources, where N is given by the length of the line source divided by its
width. Dividing the length of the line source by its width effectively splits the line source into a
string of squares (for example, if the length of the line source was 100 m, and the width was 5 m,
then the line source could be split into twenty, adjacent square volume sources). An approximate
representation of the line source can be obtained by placing a smaller number of volume sources
at equal intervals along the line source (for example, for the line source of length 100 m and
width 5 m, a total of 10 square volume sources separated from one another by 5 m could be
14
-------
defined). With this option, the spacing between individual volume sources should not be greater
than twice the width of the line source. A larger spacing can be used, however, if the ratio of the
minimum source-receptor distance and the spacing between individual volume sources is greater
than about three. Previous model evaluation studies with roadways have shown that in ISCST3
modeling roads as volume sources gave similar results to modeling the roads as area sources
(EPA, 1995c). However, modeling as area sources is more resource efficient. Additional
sensitivity studies (Personal Communication, Erode, 2001) showed that the aspect ratios (ratio of
roadway length to roadway width) can be increased from the present 1 to 10 up to 1 to 100
without degrading model performance. For the above reasons, in this example application, roads
are modeled as ISCST3 area sources with aspect ratios up to 100.
Typically, onroad mobile sources are considered line sources. However, the NTI mobile
source emissions are based on county-wide totals and allocating the emissions to all roads is
impractical. Onroad emissions are modeled in ISCST3 in two ways. The first was to assign
onroad mobile emissions to 1 km grid cells (see Section 4.3.1.1). A second method was to
allocate onroad mobile emissions to major road segments such as Interstate, U.S. and State
Highways using Geographical Information System (GIS) software. Onroad mobile emissions not
specifically allocated to these roads were interpolated to 1 km grid cells (see Section 4.3.1.2).
Nonroad mobile emissions, also reported as a county-wide total, are typically allocated to 1 km
grids based on surrogates and also modeled as area sources in ISCST3.
3.5 Default Source Parameters
Besides the emission rate, the parameters needed to model emissions from point sources
include source location coordinates, physical release height, stack diameter, exit velocity and
temperature. These parameters should available in the NTI which employs an extensive data
default assignment (see http://www.epa.gov/ttn/chief/net/nei_plan_feb2001.pdf).
Since most modeling analyses include a large number of sources over a relatively large area, it is
inevitable that there will be gaps in the data for some of the sources. It is necessary to determine
values for all the missing source characteristics, substitute them, and document the substitutions
before the sources can be modeled.
Latitude and longitude are necessary to correctly place facility release points and associated
emissions into specific geographic domains. Many instances have been reported where state and
county codes do not correspond to the latitude and longitude values and/or the zip code supplied
with each facility. The first stage in the verification of non-missing data is to use GIS overlays to
determine if the latitude and longitude of each release point is within the study domain (i.e., the
county). This ensures that sources supposed to be located within a physical representation
associated with a county are sited within the boundaries of that county. This step includes
coordination with local agencies.
15
-------
Valid parameters for the physical characteristics of each point release (stack height,
diameter, temperature and velocity) are necessary for proper air quality modeling. However, not
all of the physical characteristics of each release point are reported. Also, fugitive or vent release
locations are often not reported. Sometimes point source release values reported are physically
implausible, suggesting that a misunderstanding regarding the meaning of the data field or the
units of measure, or a mistranscription of data. The user should contact the local agency for
better data or replace these unreasonable parameters with conservative values that are related to
the type of emissions source.
For point sources with missing data, the following conservative values are recommended for
use in air toxics modeling analyses:
Stack height 10 meters
Stack diameter 1 meter
Exit temperature 295 K
Exit velocity 1 meter/second
If the NTI does not contain building dimension information, and since building wake effects
(building downwash) influences can significantly increase concentrations for receptors located
close to the point source, the following approach may be used to set default values of building
height and building width in the ISCST3 model. Default building dimensions of Hb = 0.625 * Hs
and Hw = 2 * Hb (where Hb is building height, Hw is building width and Hs is stack height) may be
used for stack heights of less than or equal to 65 meters, with a minimum building height of 3.05
meters, representative of a one-story structure. The value used for Hb places the stack height just
above the Schulman-Scire criterion, except for stack heights that are less than about 4.6 meters,
which is 1.5 times the minimum building height of 3.05 meters. The application of the
Schulman-Scire downwash algorithm is therefore limited to the shorter stacks for which it is
more likely to be applicable. The use of the Huber-Snyder downwash algorithm for stacks that
are taller than 4.6 meters also avoids the potential for unrealistically increasing predicted impacts
for these stacks based on relatively arbitrary building information, which could occur if the
Schulman-Scire algorithm were to be applied to those stacks. For stack heights of greater than
65 meters, assume no building downwash occurs, since stacks of that height are likely to satisfy
good engineering practice (GEP) stack height requirements to avoid building downwash
influences.
Non-buoyant sources are sources with plume height equal stack height (e.g., from isolated
vents), are likely to be from building vents or similar emission points. A building height equal to
the stack height should be assumed, with building width equal to twice the building height. This
automatically triggers the ISCST3 building downwash algorithm, and is a conservative approach.
16
-------
3.6 Pollutants
The ISCST3 model is run for one pollutant at a time. The number of pollutants should be
carefully defined to minimize resources. Section 112 of the Clean Air Act lists 188 hazardous air
pollutants (HAPs). The Integrated Air Toxics Strategy has identified 33 HAPs that are of
primary concern in urban areas (U.S. EPA, 1999b).
3.7 Source Grouping
From a post-analysis viewpoint, by grouping similar sources (e.g., mobile emissions), the
analyst can more easily look at the impact of different source types. These groupings can be
further subdivided into onroad and nonroad mobile source groupings. ISCST3 provides methods
for grouping sources for these purposes.
3.8 Quality Assurance
In a complex analysis such as urban area wide modeling, there are many opportunities for
error. Also, there many people are involved in the analysis and making decisions. It is
recommended that all decisions be documented.
4. EXAMPLE CASE STUDY
4.1 Introduction
This section documents the methodology and results of an example case study. For this
example, the Houston urban area is selected. The model domain covers several counties with the
Houston urban area and Harris County in the center of the domain. The area covered by the
study, along with key roadways and the location of the surface meteorological data site, is shown
in Figure 4.1-la. Emissions data for 1996 are used in this example. For illustration, example
estimated concentrations are presented for five HAPs: benzene, cadmium, chromium,
formaldehyde, and lead. There are three sets of model results/illustrations. The first set shows
ambient concentrations when all highway emissions are allocated to 1 km grid cells (see section
4.3.1.1 for more detail on how emissions are allocated). For the second set, benzene and
formaldehyde emissions from onroad mobile emissions are allocated to road segments (see
43.1.2 for an explanation of emission allocation). A third set was created to examine the effects
of receptor placement on concentrations. In this example, benzene emissions using the latter
emissions are used to calculate concentrations at receptors on a 500 m in a subset of the Houston
domain (Fig. 4.1-lb). Since benzene is the most extensive of the HAPs, benzene's model results
are presented in Section 4.4.1. A summary of results for the remaining HAPS, cadmium,
chromium, formaldehyde, and lead are then presented in Section 4.4.2.
17
-------
4.2 Model Methodology
4.2.1 Model Selection
The model used for this study was the EPA Industrial Source Complex Short Term
(ISCST3) dispersion model. The ISCST3 model is a steady-state Gaussian plume model which
can be used to assess pollutant impacts from a wide variety of sources such as multiple point,
area and mobile sources. This model was selected for this application to demonstrate what can
be done with off the shelf modeling tools. For this modeling application annual average, daily
average, and by hour-of-day concentrations were calculated. This selection for the temporal
resolution of the modeling results was based on the type of data that might be needed for use in a
typical long term human exposure assessment. The ISCST3 model is applicable to receptors
within about 50 km from the source and does not directly simulate the effects of pollutant
chemistry (i.e., chemical transformation and reactivity).
The ISCST3 dispersion model includes the capability of handling multiple point, area, and
mobile sources, incorporates building downwash effects, includes an urban dispersion option,
and also contains considerable flexibility for specifying receptor locations and for grouping of
source impacts. The ISCST3 model also includes algorithms to treat the effects of elevated
and/or complex terrain, and the effects of dry and wet deposition of gaseous and particulate
emissions. The ISCST3 model includes an option to vary emissions by season and hour-of-day,
which was useful in meeting one of the design criteria for this modeling analysis, since the
available emissions inventories reflect variations in emission rates by season and hour-of-day as
inputs. This temporal resolution has also been selected for the model outputs based on the needs
of a typical long term exposure assessment.
At the time of report preparation, EPA is developing and testing another steady state plume
model, AERMOD which could be used as an alternative to the ISC3 model. AERMOD is
actually a modeling system with three separate components: AERMOD (AERMIC Dispersion
Model), AERMAP (AERMOD Terrain Preprocessor), and AERMET (AERMOD Meteorological
Preprocessor). Special features of AERMOD include its ability to treat the vertical
inhomogeneity of the planetary boundary layer, special treatment of surface releases,
irregularly-shaped area sources, a three plume model for the convective boundary layer,
limitation of vertical mixing in the stable boundary layer, and fixing the reflecting surface at the
stack base. A treatment of dispersion in the presence of intermediate and complex terrain is used
that improves on that currently in use in ISCST and other models, yet without the complexity of
the Complex Terrain Dispersion Model-Plus (CTDMPLUS). To the practicable extent, the
structure of the input or the control file for AERMOD is the same as that for the ISCST3.
18
-------
4.2.2 Averaging Period
In order to ascertain the long term exposure to the pollutants being modeled, annual average
concentrations were calculated. Hourly and daily averages were also calculated for further study.
4.2.3 Receptor Selection Strategy
Receptors in the Houston domain were selected to coincide with census tract centroids as
shown in Figure 4.2-1. This would allow the receptors to represent the impact concentrations
would have on urban population areas. The selection of the locations of the receptors was made
so that modeled results can be input into the EPA's HAPEM exposure model (U.S. EPA, 1999a).
In the present analysis, our receptors are located at the centroid of each census tract;
typically several kilometers apart. There is interest in determining if there are any geographical
variations in concentrations within the census tract. Since adding additional receptors to the entire
domain increases model run time and analysis resources, we selected a smaller subdomain for
testing. This subdomain encompassed sources of high emissions, which could result in localized
high concentrations and may not be represented by the largely spaced census tract receptors. In
this case, the emissions are stationary sources in industrial areas located in the southeast section of
Houston where there were fewer census tracts (receptors) to calculate concentrations. In this
subdomain, we added receptors at 500 m intervals in order to see the locally high concentrations.
4.2.4 Treatment of Terrain Influences
The ISCST3 model may be run without terrain influences, i.e., flat terrain, or alternatively,
the ISCST3 model will adjust the plume heights by the receptor elevation above or below stack
base to account for the effects of elevated and complex terrain. The ISC3 User's Guide (U.S.
EPA, 1995b) contains information for handling terrain. The flat terrain option was used for the
Houston analysis. The terrain within the Houston modeling domain is relatively flat with
maximum height variations of about 50 feet. Given that a significant portion of the emissions for
these pollutants is from area sources, and the ISCST3 model ignores terrain influences for area
sources, the flat terrain assumption is considered adequate for Houston.
4.2.5 Land Use Classification
Since the Houston study domain is large, it encompasses both rural and urban regions.
Emissions input into ISCST3 must be assigned either all rural or all urban dispersion "flags" for
use in ISCST3. The "flag" was based on the presence of man-made objects likely to affect the
surface roughness characteristics. For efficiency, the Houston domain was divided into urban and
rural grid cells (1 km resolution). Grid cells were classified as urban or rural based on
19
-------
Commercial/Industrial Land Use Classification. A grid cell was classified urban that contained
Commercial/Industrial surrogates. Otherwise, it was classified as rural. All sources in an urban
grid cell are modeled with urban dispersion and all sources in a rural grid cell are modeled with
rural dispersion. Once the land use is classified, separate ISCST3 model runs were made for the
rural dispersion and rural emissions and for urban dispersion and urban emissions. The results
from the two separate runs were then combined to give concentrations at receptors from both
urban and rural sources. Figure 4.2-1 shows the land use classification used in ISCST3 for
Houston.
The effects of the land use was noticeable for near-source low-level sources during stable
meteorological conditions. Under these conditions, because of the reduced mechanical mixing in
the lower atmosphere with rural dispersion, a source assigned rural dispersion would yield higher
near source concentrations than for a similar source with urban dispersion.
4.2.6 Meteorological Data
4.2.6.1 Selection of Surface and Upper Air Stations
The ISCST3 model requires hourly surface observations of wind speed, wind direction,
ambient temperature, and stability category, in addition to mixing heights derived from twice-
daily upper air soundings as meteorological inputs. The mixing height data was calculated by the
Mixing Heights program on the SCRAM web site. The hourly surface data for major National
Weather Service (NWS) stations were obtained from NCDC.
Houston is located in flat coastal plains about 70 km from the Gulf of Mexico and about 40
km from Galveston Bay. The climate is predominantly marine and is influenced by land/sea
breezes. This effect is likely to decrease across the city. Meteorological data are collected at two
NWS sites in Houston: George Bush Intercontinental Airport (IAH) located north of the city and
Hobby Field (HOU) located south of the city. In a large urban area such as Houston
meteorological conditions are likely to vary across the city and ideally, emission sources should be
modeled with the most representative meteorological data. In this case, Hobby Field reported a
large amount of missing meteorological data and for computational efficiency and simplicity, only
the George Bush airport data were used. It is assumed that annual average concentrations based
on the two data sets are likely to be very similar and any effects due to land/sea breeze circulation
are negligible. The location of IAH can be seen in Figure 4.1-la.
The selection of the upper air stations for deriving mixing heights was based on the station
considered to be the most representative for the city. For Houston, the Lake Charles upper air
station is located about 135 miles away, while the upper air station at Victoria, TX is about 120
miles away. However, the Victoria station was relocated to Corpus Christi, TX in January 1990.
Since Victoria is located about the same distance inland from the Gulf of Mexico as both Houston
20
-------
and Lake Charles, and the distance from Houston to Victoria is comparable to the distance from
Houston to Lake Charles, both stations would be equally representative for use with the Houston
surface data. However, given the fact that the Victoria station was moved to Corpus Christi, and
the fact that Corpus Christi is located nearer to the Gulf coast, Lake Charles was considered to be
the better choice for use with Houston. The selections of the upper and surface stations also
corresponds with the recommendation of the Texas Natural Resource Conservation Commission
(TNRCC) for modeling in Harris County, where Houston is located (TACB, 1992). Table 4.2-1
gives the names and locations of the stations used for the study.
4.2.6.2 Meteorological Parameters for Deposition Calculations
Several additional meteorological parameters are needed as inputs to the Meteorological
Processor for Regulatory Models (MPRM) in order to implement the dry deposition algorithms in
the ISCST3 model for paniculate and gaseous emissions. For this study, the TOXICS option was
selected in ISCST3 which included dry deposition. Additional parameters related to wet
deposition were not needed, since wet deposition was not included in the analysis. The additional
dry deposition parameters are listed below with values based on guidance in Section 3.3 of the
MPRM User's Guide (U.S. EPA, 1996a):
Houston:
Albedo
Bowen Ratio
Roughness Length (measurement site) (m)
Roughness Length (application site) (m)
Minimum Monin-Obukhov Length (m)
Surface Heat Flux (fraction of net)
Anthropogenic Heat Flux (Wm"2)
Leaf Area Index
Winter
0.20
1.5
0.15
1.00
50.0
0.25
10.0
1.0
Summer
0.16
2.0
0.15
1.00
50.0
0.25
10.0
1.0
Fall
0.18
2.0
0.15
1.00
50.0
0.25
10.0
1.0
4.2.6.3 Meteorological Preprocessing
The MPRM program was used to preprocess the meteorological data for use with the
ISCST3 model. The source of the surface meteorological data used in MPRM was the Integrated
Surface Hourly Data (ISHD), available from NCDC, for the year 1996. Both the MPRM (U.S.
EPA, 1996a) and PCRAMMET (U.S. EPA, 1996b) meteorological preprocessors can be used to
preprocess NWS surface and mixing height data for use with the ISCST3 model. Only MPRM
was used. PCRAMMET does not allow for specifying temporal (e.g., seasonal) or spatial
variations of the surface parameters identified in the previous section and does not support the
additional parameters needed to utilize the dry deposition algorithm for gaseous pollutants. These
additional parameters are leaf area index (input by the user), and incoming solar radiation
(calculated by MPRM). An estimate of minimum mixing depth for both study areas was
21
-------
determined based on guidance in Section 2.6.
4.2.6.4 Meteorological Statistics for Houston
Meteorological statistics were calculated for the Houston domain in order to further
understand model results. A wind rose shown in Figure 4.2-2 was created for Houston using the
WRPLOT program which is available on the SCRAM web site. The wind rose showed that in
1996, the wind direction for Houston was predominantly from the south-southeast. There were
1,599 calm hours reported in the data or about 18% of the hours in a year (8,784 hours for a leap
year).
Table 4.2-2 shows the average daily maximum and minimum temperatures, averages hourly
wind speed, average hourly wind direction, and annual accumulated rainfall for 1996 and
climatology. Climatological values for wind speed and temperature were obtained from the EPA
SCRAM web site for the years 1984 through 1992. Thirty year (1961-1990) climatological annual
rainfall was obtained from http://www.met.utah.edu/jhorel/html/wx/climate/normrain.html. From
Table 4.2-2, it can be seen that 1996 did not deviate from climatology for wind speed and
direction, and temperatures. However, it did appear that 1996 was a dry year when compared to
climatology.
4.2.7 Chemistry
Estimating ambient concentrations of pollutants that undergo secondary transformation such
as formaldehyde requires three steps. In the first step, the ISCST3 model is used to estimates
concentrations from sources that directly emit formaldehyde. Decay can be estimated by using a
half life for the modeled HAP. A value of 155,520 seconds was used for the half life of
formaldehyde and for exponential decay. See Section 3 of the ISC3 User's Guide (U.S. EPA,
1995b) for decay calculations in ISCST3. Concentration estimates are obtained at all receptors in
the domain. In the second step, estimates from the screening level photochemical model (OZIPR)
are obtained. In the third step, estimates for steps 1 and 2 are added to obtain the total
formaldehyde concentration. For this analysis, OZIPR was not run, and instead tabular values in
U.S. EPA, 1999c were used. This is consistent with recommendations in the report. The tables in
U.S. EPA, 1999c are organized by season, for the hours between Sam and 8pm local time.
Secondary contributions during overnight hours, 9pm to 7am, are obtained by linearly
interpolating 8pm to Sam secondary concentrations when the 8pm value exceeds the Sam value;
when the 8pm value is smaller than the Sam value, all overnight values are assigned the 8pm
value.
4.2.8 Background
The methodology used to calculate background concentrations is described in Appendix A.
22
-------
The methodology described is for diesel PM but the same approach was used for benzene and
lead.
Annual average background concentrations were calculated for benzene and lead. Benzene
background concentrations are presented in Figure 4.2-3a. The higher background concentrations
are southwest, northwest, and northeast of Houston. The high background concentrations in these
areas are from sources in Houston and Galveston. Note that lower concentrations are in Houston
because the large sources located there do not impact within Houston but their impact is seen
elsewhere in the domain. In the southeast part of the domain, there is overlap of lower
background concentrations. This is because the 50 km radius ring for both Houston and
Galveston, TX overlap. This is an area where sources less than 50 km away from Houston are not
considered and sources less than 50 km from Galveston are not considered.
Lead annual average background concentrations are shown in Figure 4.2-3b. Higher
background concentrations are located west of the Houston area. The lowest background
concentrations are located in the northwest part of the Houston downtown area. The two circular
regions are representative of the 50 km ring associated with the two airports in the Houston
region.
4.2.9 Model Evaluation Procedure
In order to evaluate the results for the Houston study, benzene, formaldehyde, and lead
modeled annual average concentrations are compared to observed annual average concentrations
from monitors in the domain. The locations of the monitors are shown in Figure 4.1-1. There are
no monitors in the Houston area for cadmium and chromium. Note that Monitor 3 is missing.
This monitor had sparse data and appeared to have been dropped from the monitor dataset for
Houston.
4.2.10 Study Limitations
Some limitations of the study involve the following:
1. Data availability/reliability. Emissions data can have uncertainties in magnitude of
emissions but also in other parameters such as location. These issues are addressed by user
quality assurance of the data and EMS-HAP quality assurance programs. Meteorological
data, terrain inputs and site selection also contribute to uncertainty in model results. Care
should be taken in selecting the meteorological data and sites as outlined in Section 2.6.
Meteorological preprocessors, such as MPRM should be used to perform additional quality
assurance on data as well.
23
-------
2. Computational resources. Computational resources can limit the number of receptors
chosen for model simulations (See Section 2.4) since increasing the number of receptors can
increase runtime. Also, the number of sources as well as averaging times for concentrations
can also increase runtime.
3. Model Limitations. ISCST3 model formulation also contributes to uncertainty in modeled
results. See ISC3 User's Guide (U.S. EPA, 1995b) for ISCST3 model limitations.
4.3 EMISSIONS
4.3.1 Processing of Emission Data for ISCST3
Emissions data was processed with two methods. The first method was to process all
sources, major, area/other, onroad mobile, and nonroad mobile, and interpolate to 1 km grid cells.
A second method was to interpolate major, area/other, and nonroad mobile emissions to 1 km grid
cells as in the first method. However, in order to better model onroad mobile emissions, the
emissions were allocated to road segments in the Houston domain. Onroad mobile emissions not
specifically allocated to road segments were allocated to 1 km grid cells. Emission processing for
the 1 km gridded emissions (method one) are discussed in 4.3.1.1 while the allocation of onroad
emissions to road segments is discussed in Section 4.3.1.2. Emissions for ISCST3 were processed
using EMS-HAP (U.S. EPA, 2000a). The processing steps of the gridded emissions can be seen
in Figure 4.3-1. Once processed by EMS-HAP, the emissions data were put into a format
necessary for input to the source (SO) pathway of ISCST3 (see U.S. EPA, 1995b for source input
format). Also, emissions were split into urban and rural sources for the urban and rural dispersion
in ISCST3. Details follow in Section 4.3.1.1. It should be noted that the emissions inventories
processed through EMS-HAP contained emissions for several pollutants, including those not
presented in this report (i.e., butadiene, diesel particulate matter, etc.).
4.3.1.1 Gridded emissions
Mobile Sources Processing
For mobile sources, the first preprocessing step was to separate the airport emissions from
the mobile inventory using AirportProc (Chapter 2 of the EMS-HAP User's Guide, 2000a). This
program separated the airport emissions from the mobile inventory and prepared the airport
emissions for input into the point source processing programs. AirportProc allows for modeling
airport emissions as point sources instead of spatially allocated mobile sources. This capability
was built into the program because airport locations are readily available.
Once the airport emissions are split from the mobile inventory, the remaining mobile
24
-------
inventory is processed through a series of programs to create ISCST3 ready emission source files.
The first step is that the sources within the Houston domain are extracted from the mobile
inventory. These sources are then processed through the EMS-HAP MobilePrep program (Ch 9
of EMS-HAP User's Guide, 2000a). This program splits the mobile inventory into onroad and
nonroad mobile inventories and also creates variables needed for the AMProc program.
After the onroad and nonroad inventories are created, they are then processed through
Mobile_addDPM.SAS® which concatenates diesel-PM emissions with the onroad inventory and
the nonroad inventory, resulting in onroad and nonroad inventories containing diesel-PM
emissions. Next, each inventory, onroad and nonroad mobile, are processed through AMProc
(Ch. 10, EMS-HAP User's Guide, 2000a). AMProc selects pollutants, groups and assigns
characteristics to each pollutant, spatially allocates county level emissions, temporally allocates
emissions, determines model parameters, and other functions (see Ch. 10 of EMS-HAP User's
Guide, 2000).
Output from AMProc is processed through AMFinalFormat which creates SAS® datasets
and ISCST3 emission source files (SO pathway) to be included in the ISCST3 runstream input
file.
Area Sources Processing
The 1996 area source inventory initially includes landfills. The area source inventory was
then split into landfills and remaining county level area sources. Landfills were to be modeled as
ISCST3 area sources. This was because ISCST3 area sources are used to model low level or
ground level emissions with no plume rise. See Section 3.4 for a review of ISCST3 area source
characterization.
The first step in processing the landfills was to obtain location and size data about the
landfills. TNRCC provided landfill data for Texas. Landfills were selected by county and area
converted from acres to square meters. Since counties contained more than one landfill,
allocation factors were calculated as the ratio of the area of a specific landfill and the total area of
all landfills in a county. This was done by the program landfill2point.SAS® (Figure 4.3-1). The
landfill inventory was then used as input into the ISCST3 major sources.
The remaining county level area sources were processed by EMS-HAP. The first step was
the AreaPrep program (Ch. 8 of EMS-HAP User's Guide) which prepares the area source
inventory for AMProc. AreaPrep assigns spatial surrogates for each area source category for
subsequent spatial allocation of county level emissions. The program also assigns codes to source
categories for matching to temporal profiles and creates inventory variables needed by AMProc.
After processing by AreaPrep, the area source inventory was processed by AMProc and
AMFinalFormat resulting in 1 km gridded area source emissions. These emissions were written
25
-------
to source (SO pathway) files ready for ISCST3.
Major Sources Processing
The 1996 point source inventory was first processed through a program
(extract_Houston.SAS®) that extracted sources within the Houston domain (state county FIPS
48201, 48473, 48339, 48291, 48071, 48167, 48039, 48157) that have defaulted site locations or
questionable FIPS. The data were from the output from PtDataProc (see description below) when
it was used to prepare data for the ASPEN model. Sites extracted had the LFLAG variable set to
"county", indicating a defaulted site. A site's location is defaulted when the location coordinates
are missing or incorrect. See the EMS-HAP User's Guide, Chapter 3 for a more detailed
explanation (U.S. EPA, 2000a). Once the sites were extracted, they were then manually checked.
Sites in which the correct location could be determined readily were corrected. Remaining
defaulted sites were sent to TNRCC for correct locations. A list of corrected sites is shown in
Table 4.3-1.
The next step was to process the entire major source inventory, airport emissions inventory,
and landfill emissions through the program Houston_ISCpreproc.SAS®. This program combined
the three inventories, initialized certain ISCST3 variables (ISCtype, arelhght, axlen, etc.),
corrected the suspect sites, and performed quality assurance on the inventories. Output included
the corrected point sources (no landfills or airports) and data ready for PtDataProc. The
PtDataProc ready data contained point sources, landfills, and airport emissions.
PtDataProc (Ch.3 EMS-HAP User's Guide) performed quality assurance on point source
locations and stack parameters. The program also removed inventory variables not needed for
further processing. After processing by PtDataProc, the point source inventory was processed by
PtModelProc (PtAspenProc in EMS-HAP User's Guide, Ch. 4). PtModelProc selected pollutants,
grouped or partitioned pollutants and determined their characteristics. Rural/urban dispersion
parameters were assigned and vent type and building parameters were also assigned. Output from
PtModelProc was used as input into PtTemporal.
PtTemporal temporally allocated annual emissions to hourly profiles. These hourly profiles
are then used to produce eight three-hour emission rates. Output from PtTemporal was input into
PtFinalJSC (PtFinalFormat in EMS-HAP User's Guide, Ch. 7) which created the ISCST3 ready
source files (SO pathway).
Final Processing of Gridded Emissions
The source files created by EMS-HAP contained both rural and urban sources. Since
ISCST3 is run for urban and rural separately, the source files had to be split into all rural
26
-------
emissions and all urban emissions. The splitting was done by a SAS® program that processed the
source files and split them into rural or urban files based on the locations of the sources in the
Houston domain. The cells were designated urban or rural based on the Commercial/Industrial
land use classification. After the splitting, the source files were ready for input into ISCST3. The
source files consisted of the emissions, building dimension files, hourly emission files, and gas
and particle deposition information files. See the ISC3 User's Guide (U.S. EPA, 1995b) for
information about the format of these files. Values for particle deposition (particle diameter, etc.)
were obtained from EMS-HAP, Appendix E (U.S. EPA, 2000a). Fine particle diameters were
1.58 (j,m and coarse particle diameters were 6.93 //m. Mean particle density for cadmium and
chromium was 1.0 g cm"3. Values for other deposition parameters for both particles and gases can
be found in Wesely et al., 2002.
4.3.1.2 Allocation of onroad mobile emissions to road segment emissions
Previous methodologies for preparing ISCST3 model-ready mobile emissions inventories
are based on using spatial surrogates to allocate the county level emissions to each grid cell.
These surrogates are meant to represent areas where mobile source emissions are likely to occur.
Population is used as the surrogate for neighborhood roads and roadway types are used for the
major roads. Surrogate values are calculated for each grid cell by using Geographic Information
System (GIS) software to spatially overlay a grid onto census block areas and TIGER roads.
Census blocks boundaries and roads are obtained from the U.S. Census TIGER/Line data
(http://www.census.gov/geo/www/tiger)and population counts from the U.S. Census of
population and housing (http://www.census.gov/mp/www/rom/msrom6ae.htm). Census block
areas are joined with population data to determine the number of people living within each block.
A modeling grid is overlayed onto census blocks and the total population in each grid cell is
calculated. The ratio of the cell population to the total county population is then applied to the
county emissions to obtain grid cell level emissions. Roads consist of contiguous arcs which are
coded according to road type. The total length of each road type is calculated for each cell.
Ratios to the total length of each road type in the county are calculated and applied to the county
emissions for that road type. The emissions from the population and roads are then summed to
obtain the total mobile emissions for each grid cell.
A drawback to this methodology is that the spatial distribution of emissions are not always
represented accurately. First, surrogates may not adequately represent the census block's mobile
source activity correctly. Secondly, distributing mobile source emissions throughout an area may
underestimate emissions density on transportation features such as roadways or parking lots.
Road emissions are spread evenly throughout a cell instead of occurring along actual road
locations. As a result, air dispersion modeling can not capture high concentrations that often
occur next to roadways and at intersections. This study demonstrates an alternative methodology
using link-based emissions generated for the major roadways in Harris County, Texas. Harris
County emissions are calculated for each major roadway link using traffic counts and vehicle
emission data. Roads are then modeled as ISCST3 area sources with aspect ratios (ratio of
27
-------
roadway length to roadway width) up to 100. Road emissions for local and neighborhood streets
are processed using the previous gridding methodology due to the lack of traffic counts on the
local facilities. The process for creating link based emissions for Harris County is described
below.
Data Sources
The data required for generating link based emissions include 1) city specific traffic counts
2) a base map of road locations and 3) mobile emissions factors.
Local Traffic Counts - City specific traffic counts may be obtained from the State
Department of Transportation (DOT) or from local sources such as the Metropolitan Planning
Organization (MPO) or a Regional Planning Council/Council of Governments. State DOTs' may
also provide access to local travel data. These data are usually part of the Highway Performance
Monitoring System (HPMS), administered by the U.S. Department of Transportation
(http://www.fhwa.dot.gov/ohim/hpmspage.htm). It is preferable that the traffic counts are in a
spatial database, suitable for import into a GIS. Traffic counts will usually be provided as Annual
Average Daily Traffic (ADT). The ADT represents the total number of vehicles crossing a
measured point during an average day. The traffic counts for the Houston study are from the
Houston/Galveston Area Planning Council.
As an alternative to traffic counts for inventory calculations, many MPOs use travel demand
models (TDMs) to predict the number of vehicle trips in a transportation network TDMs use
local economic and demographic data to determine trip originations and destinations in travel
analysis zones (TAZs), often the same as Census blocks. TDMs also calculate the optimum
number of trips along each roadway in a simulation network.
TDMs have several advantages in modeling air toxics, including placement of vehicle starts
in TAZs. Start emissions from mobile sources may constitute the majority of toxic releases.
Because TAZs often use Census-defined boundaries, vehicle start emissions and their locations
can be estimated with greater accuracy and imported into air quality models as area sources using
GIS software. Users familiar with emission budgeting and conformity applications of the
MOBILE model will find that geographically resolved toxic emissions inventories can be created
using many techniques already in place.
Road Locations - A base map of road locations in the study area can usually be acquired
from the sources identified above. These should be in electronic form and, at a minimum, contain
geographic coordinates and road names. If possible, these data should contain information on
roadway attributes including road width, number of lanes, median width and surface type. If an
electronic map is not available, then a paper map can be converted to digital form by digitizing
which will assign geographic coordinates to the road links. Ideally, the digital map would be in a
28
-------
format that can be easily imported into a GIS. Examples are Arclnfo® export files or Arc View®
shape files that are used by those software packages. If local sources are not available, U.S.
Census TIGER roads can be used, although they may not be as accurate and up to date.
Emission Factors - In the Houston modeling analysis, emission factors for benzene, 1,3-
butadiene and formaldehyde are obtained from the EPA toxic emission factor model,
MOBTOXSb, developed by the Office of Transportation and Air Quality (Cook et al., 1998; U.S.
EPA, 1999d; Cook et al., 2000). The emission factors are expressed in grams emitted per vehicle
mile traveled (g mi"1). MOBTOXSb generates emissions factors for total organic gas (TOG) and
speciates TOG into individual air toxics based on vehicle and fleet parameters. The TOG
speciation fractions in the model depend on technology types, driving cycles, and normal versus
high emitters. The toxic fraction of TOG is also highly dependent on fuel parameters included in
the Complex Model for reformulated gasoline and a draft fuel effects model for MTBE. Emission
factors are generated for the vehicle classes shown in Table 4.3-2. They are based on an average
vehicle speed of 19.6 miles per hour for all vehicles. In reality, speeds will vary significantly
among different roadway links, and can have a large impact on emission factor estimates (U.S.
EPA, 2000b). These emission factors also assume a distribution of operating modes with 20.6%
of VMT assigned to cold starts, and 27.3% to hot starts4. Using parameters more appropriate for
individual roadway links will result in more accurate link specific emissions. In the Houston
analysis, for instance, a higher average vehicle speed and assignment of more VMT to running
mode would have been more representative of major roadways. Of course, these parameters also
vary by time of day, which is not reflected in this analysis. However, the use of default
parameters are adequate for the purposes of methodology development and evaluation of the
impacts of a link based approach on dispersion modeling results.
EPA has recently integrated gaseous toxic and PM emission factor estimations into the
MOBILE6 model (U.S. EPA, 2000c; U.S. EPA, 2002). MOBILE6 estimates emission factors by
highway functional system (freeway, arterial collector, local roadway, freeway ramp). When the
user specifies an average speed for roadway functional type, MOBILE6 applies a speed
distribution for that roadway type. Thus, MOBILE6 can be used to develop emission factors more
appropriate for specific links.
MOBILE6 calculates toxic emission factors as a fraction of total organic gas (TOG). The
toxic fraction of TOG for each species is highly dependent on fuel properties. Therefore,
MOBILE6 requires more detailed fuel parameter descriptions for calculating toxic emission
factors than for criteria pollutant emission factors. Detailed fuel composition data is available for
4A cold start is defined as the first 3.5 miles traveled by vehicles after a "cold-start", and a hot start as the first 3.5
miles after a hot start."
29
-------
some areas through the Alliance of Automobile Manufacturers' North American Fuel Survey5.
Where Alliance survey data is unavailable, TRW Petroleum Technologies Survey gathers regional
fuel data which may be substituted for local survey data6. In particular, the fuel properties
required by the MOBILE6.2 and MOBTOXSb models are:
• % Aromatic of gasoline on volume basis
• % Olefin of gasoline on volume basis
• % Benzene of gasoline on volume basis
E200 - % of vapor a gasoline produces at 200 °F
• E300 - % of vapor a gasoline produces at 300° F
• Oxygenate type and content on volume basis
Emission factors for diesel PM are based on modeling done for EPAs' recent regulation
promulgating 2007 heavy duty vehicle standards (U.S. EPA, 2000c). The emission factors used
are as follows:
LDDV - 253 mg/mi
LDDT - 309 mg/mi
HDDV (urban interstate/ freeway) - 985 mg/mi
HDDV (other urban facilities) - 921 mg/mi)
Generally, paniculate emission factors can be estimated using the MOBILE6 emissions
model, which calculates particulate emission rates using similar fleet and fuel parameters as
discussed above. However, particulate emissions do not require fuel parameter specifications of
the same detail as the gaseous toxics component of the model.
Users with access to a TDM should make use of roadway specific speeds, informed by TDM
results. Since some TDMs often produce speeds only to ensure optimum distribution of traffic
volume, speeds in inventory calculations should be based on other means of calculating speed
(For more detailed discussion of travel demand modes in, consult "Procedures for Emission
Inventory Preparation. Volume IV: Mobile Sources." EPA report number EPA420-4-92-009.
Available at http://www.epa.gov/otaq/invntory/r92009.pdf)
In the Houston modeling study, toxic emissions factors for each pollutant are calculated
using local data on Houston fuel and fleet parameters. These emissions factors are applied to the
5The Alliance's fuel survey can be accessed via http://www.autoalliance.org/fuelquality.htm or (202)326-5533
6The TRW fuel survey data can be accessed from: TRW Petroleum Technologies, Attn: Cheryl L. Dickson, P.O.
Box 2543, Bartlesville, OK 74005. Telephone: (918)338-4419
30
-------
annual Vehicle Miles Traveled (VMT) for each road segment as described later in this section.
Traffic counts by vehicle class are determined by a vehicle split table that apportions the total
traffic into 12 vehicle types as seen in Table 4.3-2. This type of table is specific to every locality
as each city has its own particular vehicle mix. Emission rates are also specific for each area.
Those for the Houston study are estimated as described above.
Data processing
Road segments
The Houston study uses roads extracted from TIGER data because road data from local
sources do not contain the necessary geographic information. Once the data are acquired they are
imported into the Arclnfo® GIS. The road data must then be edited to obtain a layer of major
roads which meet the following ISCST3 model criteria. First, local and neighborhood roads are
deleted so that only the major roads for which VMT will be calculated remain. Refer to Figure
4.1-1 for road locations. Next, road segments must be processed to meet minimum and maximum
length requirements for use in the ISCST3 model. A roadway link cannot have less than a 1:100
ratio or greater than a 100:1 ratio with the road width. Table 4.3-3 shows the number of lanes
assumed for each road class and the associated length parameters. In order to create road
segments with correct lengths, the segments are first joined together to form pieces of the
maximum possible length, thereby eliminating segments shorter than the minimum threshold.
Vertices, x and y locations along the line, are then added at specified intervals along the segments
and used to split them into sections that do not exceed the maximum length threshold.
Traffic counts
Each road segment must be associated with a traffic count in order to calculate the traffic
volumes and related mobile emissions. In the Houston study area there are discrepancies between
the traffic count locations and the road segments; this becomes apparent when these two data
sources are overlayed using the GIS. The traffic counts are collected at discrete point locations
which do not always align with the roads. The points contain no attribute information, such as
address location, to link them to road segments. In order to link traffic counts with the closest
road segment, a buffer is generated around each traffic count point. In the Houston study a 100
meter buffer proved the optimum size but this may vary depending on the area modeled. The
traffic count buffers serve to convert point locations to polygon areas which can then be overlayed
with road segments. The overlay process joins traffic counts to individual road segments.
Traffic counts assigned using the method described above are linked to discrete locations
along a road. Counts must be interpolated along the entire length of the road in order to calculate
continuous traffic flows. To do this, the dynamic segmentation capabilities of the GIS are used to
create a file containing ordered road segments and their ADT counts. Dynamic segmentation is a
31
-------
layered system of pointers which can associate attributes to geographic locations along a linear
feature. In this study, it is used to determine the order of segments along each road by creating a
route. Each route is comprised of sections which are identified by their starting and ending
position along the road. The starting point is the "from node" and the ending point is the "to
node"; therefore each section is a vector which imparts a direction of flow. As long as all of the
sections along a road run or flow in the same direction, it is possible to create an ordered table of
road segments with attached ADT value. This ordered table can then be used to interpolate the
ADT along the route by creating ratios of distances between segments with known traffic counts.
The ratios are used to apportion ADT based on distance and direction from known values
assuming a constant rate of change between traffic count locations. The road segments containing
road name, unique ID number, segment length, traffic count, road type and State/County FIPs
codes are then written out into an ASCII file and imported into SAS® software for calculation of
emissions.
Calculation of link based emissions
The ASCII records generated in the steps above are read into SAS® software for calculation
of road emissions. First, the vehicle split factors shown in Table 4.3-2 are applied to each record
by road type. These distributions are national defaults used in the EPA Emission Trends report to
characterize the vehicle types present on the 12 different roadway types included in the HPMS.
Applying the vehicle split table divides the traffic along each road among the 12 vehicle
types and assigns a proportionate amount of the total ADT to each road segment. Once the
vehicle split ADT counts are determined, the emissions factor table is used to calculate emissions
by vehicle type, pollutant and road type by performing the following operations:
Annual VMT = ADT * Roadway Length * 365
Emissions = (VMT * emissions factor)
After the emissions along each road segment are calculated, the data are brought back into
Arclnfo and the emissions are merged with the road segment locations. The results for benzene
and formaldeyde are presented in Section 4.4. A final step is to write out a formatted file for input
into the ISCST3 model. This file contains the UTM coordinates which define the starting and end
points of each segment as well as the information needed to model the road segments which
include emissions,total segment length, road classification code, and road width. Road segment
emissions are modeled in ISCST3 as rural sources while the remaining onroad mobile emissions
not specifically allocated to road segments are processed through EMS-HAP as for gridded
emissions and modeled in ISCST3 as rural or urban sources.
32
-------
In applications in which a TDM is used, it may also be necessary to match vehicle traffic
counts to specific roadways, as transportation models often express vehicle counts at nodes
between roadways, rather than on specific roadways. TDM users may also need to generate a
number of emission factors for each vehicle type, corresponding to different transportation facility
categories and vehicle speeds predicted through use of the model.
One problem with the approach described here is that county level VMT estimates from the
Highway Performance Modeling System (HPMS) used in the National Emissions Inventory and
National Toxics Inventory do not always match local traffic count data. For example, the link
based approach was tried on several adjoining counties in the Houston metropolitan area. The
link based emissions calculated for major roadways in these counties exceeded the total county
emissions in the NTI. This may be due to disparities in the way in which traffic data are collected
but further investigation is necessary.
One consideration in implementing link-based roadway emissions inventories are
parameters for the "release height" above the roadway of interest. In the Houston analysis, a zero
release height was used. However, for some heavy-duty trucks, a higher release height may be
warranted to reflect greater stack height. Due to uncertainties introduced by turbulence on
roadways, however, it is unclear which release height is most appropriate. For elevated highways,
a higher release height may also be warranted. However, highway elevation data were not
available.
Another concern is the treatment of link emissions as area sources or as volume sources.
The Houston model analysis implemented roadways as rectangular area sources. Common air
quality models such as ISCST3 also allow specification of volume sources. Given the turbulent
roadway meteorology, it is unclear whether volume or area source treatment of mobile sources is
most appropriate. It will improve understanding of model output stability for users to employ
both approaches for sensitivity analysis.
Users employing TDMs may also use starts per TAZ to locate start emissions more
accurately. In this case a map of the TAZs, including number of starts, can be imported into a GIS
database system. The distribution of vehicle types in each TAZ throughout the modeling domain
are likely to be similar to those in the link based methodology. However, it is anticipated that
local registration databases will more accurately reflect the composition of "start" fleets.
MOBILE6 can determine start emissions for each hour of the day. The distribution of vehicle
soak emissions can be determined by the ends of trips as predicted by travel demand models. The
total emissions for each TAZ can be gridded using a GIS, or imported directly into air quality
model input files, however, it is anticipated that grid cells will be easier to program as area
sources into air quality models.
33
-------
4.4 HOUSTON DOMAIN EMISSIONS AND MODELING RESULTS
This section describes the results of emissions preprocessing and ISCST3 simulations for
Houston. Section 4.4.1 will describe the results for benzene and Section 4.4.2 will give overviews
of results for the other HAPs. Simulations using all mobile sources modeled as area sources are
referred to as ISCST3 BASE. For simulations using the road segment emissions (benzene and
formaldehyde), results will be referred to as ISCST3 ROADS. ISCTST3 results using the 500 m
receptor density are referred to as ISCST3 FINE GRID. For each simulation, modeling is
performed for all source categories at one time, but model results can be output for each source
category separately, so that concentration estimates can be attributed to each category.
For all HAPs, concentrations were calculated at 711 receptors (700 census tract centroids
and 11 monitors sites) by ISCST3. For purposes of displaying ISCST3 results, the receptor
concentrations were averaged over a 1 km grid cell containing that receptor. If there was more
than one receptor in a grid cell, concentrations among the receptors were summed and then
averaged within the grid cell; if there were no receptors in a cell, then the cell average was zero.
This process resulted in 692 grid cells. See Figure 4.4-1 as an example showing the 1 km grid
cells.
Table 4.4-1 gives the maximum annual averaged total (all sources) concentration and
southwest corner of the grid cell in which it is located for each pollutant. The concentrations for
each source category in that grid cell are also given. Table 4.4-2 gives the maximum
concentration and location of each source category for each pollutant.
4.4.1 Benzene
4.4.1.1 Emissions
One kilometer gridded benzene emissions for total emissions, major, and onroad sources are
shown in Figures 4.4-1 and 4.4-2. The emissions are high along the roadways leading into and
around Houston, showing the contribution of onroad emissions. Other roadways can be seen as
long segments in which the emissions are higher than the surrounding areas. The major emissions
(Figure 4.4-2a ) are mostly located in eastern Harris County. Most of the emissions range from 1-
25 tons year"1. However, there are five sources with emissions greater than 80 tons year"1.
Figure 4.4-3a shows the contributions from the different source categories to the emissions
for Houston. Also shown are the distributions for the 1996 national inventory, the Houston major
source emissions, Houston area source emissions, and Houston mobile emissions. Onroad mobile
emissions comprised most of the emissions (41%). A more detailed breakdown of the emissions
by source category can be seen in Figure 4.4-3f for Houston.
34
-------
Road segment emissions for benzene are shown in Figure 4.4-4a. Some of the higher
emissions can be seen along the interstate highways (> 0.4 tons year"1 km"2). Once the emissions
on the road segments were allocated to those segments, the road segment emissions could be
subtracted from the NTI onroad mobile gridded emissions. The remaining onroad gridded
emissions are those onroad emissions not specifically allocated to the roadways shown in Figure
4.4-4a. These gridded remaining onroad mobile emissions are shown in Figure 4.4-4b. The
highest emissions are clustered around the center of the city. Compared to Figure 4.4-2b, the
roads were no longer as obvious in the remaining onroad gridded emissions as in the onroad
mobile emissions shown in Figure 4.4-2b.
4.4.1.2 ISCST3RESULTS
ISCST3BASE
Annual average concentrations (background concentrations not included) from all sources
were highest north of the city center (Fig 4.4-5). These higher concentrations correspond to the
areas where there were high emissions (see Figure 4.4-1). Other high values were scattered
around the region, northwest Galveston County and in eastern Harris County. The concentrations
for each source category are presented in Figure 4.4-6. For major source concentrations, values
increased from west to east. The highest concentrations are located near the major source
emissions. The higher onroad concentrations appeared to be concentrated near high onroad
emissions. Nonroad concentrations were mostly located within the city.
ISCST3 ROADS
The total concentrations for the ISCST3 ROADS run are shown in Figure 4.4-7a. The
higher concentrations were mostly located within the city urban area. The onroad concentrations
(Figure 4.4-7b) were also higher in the city than the surrounding areas. This pattern differs from
the ISCST3 BASE onroad mobile concentrations (See Figure 4.4-6c) in which the onroad mobile
concentrations are more widespread in the city. In order to compare the concentrations from the
two different model runs, percent differences between ISCST3 ROADS and ISCST3 BASE were
calculated. Figure 4.4-8 shows the percent differences between ISCST3 ROADS and ISCST3
BASE for total and onroad concentrations respectively (major, area, and nonroad mobile
concentrations were not affected by the road segment emissions allocation). Allocating the
onroad mobile emissions to road segments resulted in higher concentrations for most receptors.
In north and northwest Harris County, the ISCST3 ROADS total concentrations are lower than
ISCST3 BASE total concentrations. This same pattern also holds for percent differences between
the onroad source concentrations. The lower ISCST3 ROADS concentrations in north Harris
County were a result of the allocation of emissions to road segments. There were actually lower
traffic counts in this area, resulting in lower emissions, and subsequently lower concentrations.
35
-------
Figure 4.4-9 shows the largest source contributor to total concentrations for each receptor for
ISCST3 BASE (Figure 4.4-9a) and ISCST3 ROADS (Figure 4.4-9b). For ISCST3 BASE
concentrations, the total concentrations in each grid cell were composed in most part to the
contribution from onroad mobile source concentrations (Figure 4.4-9a). The location of the
maximum total concentration was an exception, in that the major source concentration contributed
the most to the total concentration. As for the ISCST3 BASE run for benzene, the onroad
concentrations contributed most to the total concentration at a majority of receptors. A
comparison of maximum total concentrations for ISCST3 BASE and ISCST3 ROADS is shown
in Table 4.4-3. The maximum concentration for ISCST3 ROADS is higher than for ISCST3
BASE. Also, the location of maximum concentration for ISCST3 ROADS was southwest of the
maximum concentration of ISCST3 BASE.
ISCST3 FINE GRID
ISCST3 was run for benzene using a set of receptors spaced 500 m apart in an area shown in
Figure 4.1-1. Onroad segment emissions and gridded emissions for all other sources (major, area,
nonroad, and nonallocated onroad) were input into ISCST3. As previously noted, this area was
chosen because of the location of major and onroad emissions. Concentrations from the model
were analyzed in two ways: 1) scaling the color bar on the plots to the 98th percentile of the
concentrations to see the concentration gradient; 2) scaling the color bar on the plots to the
maximum concentration so that higher concentrations, "hot spots" could be detected.
The total concentrations are presented in Figure 4.4-10. Figure 4.4-10a shows the
concentrations scaled to the 98th percentile. The most noticeable concentration gradients can be
seen along the roadways. Figure 4.4-10b shows the concentrations scaled to the maximum value.
Most concentrations were 5 //g m"3 or lower. The maximum concentration was over 40 //g m"3.
Figure 4.4-11 through Figure 4.4-14 show the concentrations for major, area, onroad, and
nonroad concentrations. For major concentrations, there appear to be three local maxima in the
concentrations (Figure 4.4-1 la). Figure 4.4-1 Ib shows a similar pattern as for total
concentrations, the highest concentration was in the western part of the domain with a
concentration over 40 //g m"3, although the local maxima can still be seen. Area source
concentrations were low, less than 1 //g m"3, with a maximum concentration at the eastern edge of
the domain (Figure 4.4-12). Local maximum onroad concentrations are located along the
roadways with the maximum concentration located near the interstate highways (Figure 4.4-13).
As for area concentrations, nonroad concentrations are less than 1 //g m"3 (Figure 4.4-14). Several
local maxima are evident from the concentration with the maximum concentration, approximately
0.8 //g m"3 was located in the southwest corner of the domain.
36
-------
4.4.2 Other HAPs
For the other pollutants, cadmium, chromium, formaldehyde, and lead, the methodology for
modeling and displaying used for benzene was applied to each pollutant.
Cadmium
Cadmium emissions were less than 2 tons year"1. Area emissions were the largest
contributors to the emissions by source categories, with incinerators being the largest sources
(Figure 4.4-3b). Figure 4.4-15 shows the gridded total (all sources) emissions and the area source
emissions. The pattern between both emission categories was similar. Emissions that appeared in
the total emissions (for example, in northeast Harris County) but not in area source emissions
were due to nonroad and major source emissions, which were few in number.
Total concentrations for cadmium (Figure 4.4-16a) were composed of concentrations from
three source categories: major, area/other, and nonroad mobile. The highest total concentrations
were located near the city center and in Galveston County. The area source concentrations were
the largest components of the total concentration at most locations as seen in Figure 4.4-16b with
the exception of the maximum total concentration (Table 4.4-1) where the major source
concentration was the largest component.
Chromium
Figure 4.4-3c shows that major source emissions are the largest of the total emissions for
chromium. Analysis of the gridded emissions for total emissions (Figure 4.4-17), major
emissions, area emissions, onroad, and nonroad emissions (not shown) show that for the most
part, the total emissions at each 1 km cell were composed of onroad, nonroad, and area sources.
In other words, the major sources were few in number but were large emitters.
Highest total concentrations were in eastern Harris County (Figure 4.4-18a) near high
emissions. Also, high total concentrations were modeled in Galveston County. Figure 4.4-18b
shows the largest source contributors for each grid cell. In eastern Harris County, the major
sources contributed the most to the total concentration. This was also true for the location of
highest total concentration (Table 4.4-1).
Formaldehyde
Gridded total emissions for formaldehyde are shown in Figure 4.4-19. The higher emissions
are along and around the roadways. Also, high emissions are located at the airports in northern
and southern Harris County. These are primarily due to aircraft emissions which are large
37
-------
nonroad mobile emitters. Figure 4.4-3d shows that nonroad mobile sources are in fact, the largest
contributor to the total emissions.
Road segment emissions for formaldehyde are presented in Figure 4.4-20. A pattern in the
location of the higher emissions is seen that is similar to the benzene road segment emissions
pattern. As with benzene, the higher gridded remaining onroad mobile emissions are near the
center of the city (Figure 4.4-2 la). Compared to the all gridded onroad emissions (Figure 4.4-
21b), more detail of the emissions can be seen in the city and in northwest Harris County, as well
as northern Brazoria County.
ISCST3 BASE total concentrations for formaldehyde (including secondary concentrations of
8.8 //g m"3) are shown in Figure 4.4-22a. As with other pollutants, high concentrations were
modeled in northern Houston. Also high concentrations were calculated in the southern part of
the city. The high secondary concentrations gave the total concentrations a more uniform
distribution. Onroad mobile concentrations are shown in Figure 4.4-22b. High onroad
concentrations were calculated in the northern part of the city and also in the southwest part of the
city near high emission sources.
For ISCST3 ROADS, highest total (including secondary) concentrations for formaldehyde
were mostly located within the city (Figure 4.4-23a) and in northern Harris County. The onroad
concentrations also show a similar pattern (Figure 4.4-23b). The percent differences between
ISCST3 ROADS and ISCST3 BASE are shown in Figures 4.4-24. Total concentration percent
differences were positive at a majority of receptors with a few negative differences. For onroad
concentrations, at most locations, the differences are positive, indicating that ISCST3 ROADS
concentrations were higher than ISCST3 BASE. The exceptions were mainly in northern Harris
County, in northeast Harris County and, southwest Harris County.
Figure 4.4-25 shows the largest source category contributing to the total concentrations for
ISCST3 BASE (Figure 4.4-25a) and ISCST3 ROADS (Figure 4.4-25b). The pattern is similar for
both model results but along some of the roadways within Houston, the onroad concentrations
contribute more than nonroad, due in part to the increased onroad concentrations from ISCST3
ROADS. Table 4.4-4 compares the locations of the maximum total concentrations for ISCST3
and ISCST3 ROADS. Only the value of the total concentration changed as did the value of the
onroad mobile concentration. Note that the position of the maximum total concentration did not
change.
Lead
Figure 4.4-3e shows the breakdown of emissions for lead. The nonroad emissions are the
largest contributors to the emissions (44%). The nonroad mobile emissions are due to piston
engine aircraft. The gridded emissions for lead are shown in Figure 4.4-26. For the most part, the
38
-------
spatial distribution of the gridded total emissions (Figure 4.4-26a) appears similar to the gridded
area emissions (Figure 4.4-26b). The nonroad emissions (Figure 4.4-26c) show only a few
sources but they are large emissions relative to area emissions. The nonroad emissions consist
mainly of the two major airports in the domain.
Total concentrations for lead are shown in Figure 4.4-27a. The higher concentrations are in
southern Harris County with another area of high concentrations in the northern part of the county.
The area, and nonroad mobile concentrations are shown in Figure 4.4-27a-b. Higher area
concentrations are located in the center of Houston eastward to the Harris County line. High
nonroad concentrations are clustered in southern Harris County. For most receptors, the nonroad
concentrations contribute the most to total concentrations (Figure 4.4-28).
4.4.3 Model to Monitor Comparisons
In order to ascertain the performance of the model concentrations, the annual average model
concentrations at monitor sites were compared to annual average concentrations from the
monitors. Only three of the HAPs had reported monitor values: benzene, formaldehyde, and lead.
However, there was only one monitor for lead and four for formaldehyde. Only benzene will be
discussed. Figure 4.4-29 gives the annual average concentrations for individual monitors and
ISCST3 and ISCST3 ROADS for benzene. Also shown on the graph are the average
concentration among all monitors for ISCST3 BASE, ISCST3 ROADS, and the monitor values.
Root mean square (RMS) errors were calculated for ISCST3 BASE and ISCST3 ROADS.
Allocating the onroad mobile emissions to road segments increased the average model
concentration and decreased the RMS errors. Since benzene is dominated by onroad mobile
sources, these findings suggest that allocating onroad emissions to road segments improves the
model results. At the individual monitors, ISCST3 ROAD concentrations are higher than ISCT3
BASE concentrations. Observed concentrations are higher than ISCST3 BASE at all monitors
except one monitor. ISCST3 ROADS concentrations are higher than observed concentrations at
four of the monitors.
5. SUMMARY AND CONCLUSIONS
Emissions for the Houston region were processed through EMS-HAP and annual average
concentrations were calculated using the ISCST3 model. Emissions were processed with two
methods: 1) all sources were modeled as 1 km grid cells, and 2) onroad mobile emissions were
allocated to road segments. In method two, onroad mobile emissions not specifically allocated to
road segments were modeled to 1 km grid cells. Other sources (major, area/other, and nonroad
mobile) were modeled as in method one. After emission processing by EMS-HAP, sources were
divided into rural and urban sources. ISCST3 was then run for each source type and
concentrations added together at each receptor during postprocessing. ISCST3 was run for five
pollutants: benzene, cadmium, chromium, formaldehyde, and lead. There were 711 receptors
39
-------
used in the model (700 census tract centroids and 11 monitors). For benzene and formaldehyde,
ISCST3 was also run using the road segment emissions. Additionally, for benzene, ISCST3 was
run in an area of high major/onroad emissions with receptors every 500m.
Several conclusions can be drawn from the results of this study:
1. Increasing the receptor density near high emission sources changed the location of
maximum concentrations between ISCST3 BASE and ISCST3 FINE GRID. The ISCST3
FINE GRID results also illustrated the concentration gradients that can occur near high
emission sources. These findings illustrate the importance in of the receptor placement and
density to model performance.
2. Allocating the onroad mobile emissions to road segments can improve the model predicted
concentrations when compared to monitor observations. The benzene ISCST3 BASE
underpredicted the average concentrations at the seven monitors. Road segment allocation
(ISCST3 ROADS) resulted in better model-monitor agreement and also changed the
location of maximum concentrations when compared to ISCST3 BASE. Allocating onroad
mobile emissions to road segments also increased the maximum total concentration for
formaldehyde but as seen in Table 4.4-2, the nonroad mobile concentrations still dominated
the maximum total concentration.
3. It can be seen that higher concentrations are located near the higher emissions for the five
HAPS presented. It appeared that a majority of the high emissions were located in eastern
and northern Harris County, as were the higher concentrations. Also among the five HAPs,
the trend is that the HAPs with higher emissions also have higher maximum concentrations.
4. It is important that emissions inventory development continue to be refined in order to
define emission parameters, sources, emissions amounts, and locations for input into
dispersion models. This will aid in predicting accurate model concentrations for assessing
exposure to toxic pollutants.
40
-------
6. REFERENCES
40CFR51. Guideline on Air Quality Models. Appendix W to 40CFR51.
http ://www. epa.gov/ttn/scram
Cook, R., P. Brodowicz, D. Brzezinski, P. Heirigs, and S. Kishan, 1998. Analysis of In-Use
Motor Vehicle Toxic Emissions Using a New Emission Factor Model, MOBTOXSb.
Presented at AWMA Specialty Conference, Emission Inventory: Living in a Global
Environment, December 8-10.
Cook, R., P. Brodowicz, P. Heirigs, S. Kishan, and M. Weatherby, 2000. Assessment of
Emissions and Exposure from Selected Motor Vehicle Air Toxics. Paper No. 77,
Presented at 93rd Annual AWMA Conference, Salt Lake City, UT, June 18-22.
Gery, M. W. and R.R. Grouse, 1991. User's Guide for Executing OZIPR, U.S. Environmental
Protection Agency, Research Triangle Park, NC.
Sutton, O. G., 1953. Micrometeorology. McGraw-Hill, New York, NY.
TACB (Texas Air Control Board), 1992. Air Quality Modeling Guidelines. Texas Air Control
Board, Austin, Texas.
U.S. EPA, 1995b. User's Guide for the Industrial Source Complex (ISC3) Dispersion Models.
Office of Air Quality Planning and Standards, Research Triangle Park, NC.
EPA-454/B-95-003b. http://www.epa.gov/scram001/userg/regmod/isc3vl.pdf
U.S. EPA, 1995c. Modeling Fugitive Dust Impacts from Surface Coal Mining Operations- Phase
HI, Evaluating Model Performance. Office of Air Quality Planning and Standards, Research
Triangle Park, NC. EPA-454/R-96-002.
U.S. EPA, 1996a. Meteorological Processor for Regulatory Models (MPRM) User's Guide.
Office of Air Quality Planning and Standards, Research Triangle Park, NC.
EPA-454/B-96-002. http://www.epa.gov/scram001/userg/relat/mprmd.zip
U.S. EPA, 1996b. PCRAMMET User's Guide. Office of Air Quality Planning and Standards,
Research Triangle Park, NC. EPA-454/B-96-001.
http://www.epa.gov/scram001/userg/relat/pcramtd.pdf
41
-------
U.S. EPA, 1998. User's Guide for the AERMOD Terrain Preprocessor (AERMAP), Revised
Draft. Office of Air Quality Planning Standards, Research Triangle Park, NC.
http://www.epa.gov/scram001/7thconf/aermod/aermapug.pdf
U.S. EPA, 1999a. The Hazardous Air Pollution Exposure Model (HAPEM4 ) User's Guide,
Draft, Office of Air Quality Planning and Standards, Research Triangle Park, NC.
U.S. EPA, 1999b. National Air Toxics Program: The Integrated Urban Air Strategy; Federal
Register; 64FR38705, July 19, 1999. http://www.epa.gov/ttn/uatw/urban/frl9jy99.html
U.S. EPA, 1999c. A Simplified Approach for Estimating Secondary Production of Hazardous Air
Pollutants (HAPs) using the OZIPR Model. Office of Air Quality Planning and Standards,
Research Triangle Park, NC. EPA-454/R-99-054.
http://www.epa.gov/ttn/scram/guidance/reports/oziprpt/oziprhps.pdf
U.S. EPA, 1999d. Analysis of the Impacts of Control Programs on Motor Vehicle Toxics
Emissions and Exposure in Urban Areas and Nationwide. Prepared for U.S. EPA, Office
of Transportation and Air Quality, by Sierra Research, Inc., and Radian International
Corporation/Eastern Research Group. Report No. EPA 420 -R-99-029/030.
http://www.epa.gov/otaq/regs/toxics/r99029.pdf
U.S. EPA, 2000a. User's Guide for the Emissions Modeling System for Hazardous Air Pollutants
(EMS-HAP, Version 1.1). United States Environmental Protection Agency, Office of Air
Quality Planning and Standards, Research Triangle Park, NC, 27711. EPA-4541R-00-018.
http://www.epa.gov/ttn/scram/userg/other/emsug.pdf
U. S. EPA, 2000b. AP-42: Compilation of Air Pollutant Emission Factors - Volume II: Mobile
United States Environmental Protection Agency, Office of Air Quality Planning and
Standards, Research Triangle Park, NC, 27711. Sources, http://www.epa.gov/otaq/ap42.htm
U.S. EPA, 2000c. Procedures for Developing Base Year and Future Year Mass and Modeling
Inventories for the Heavy Duty Engine and Vehicle Standards and Highway Diesel Fuel
(HDD) Rulemaking. Prepared for U.S. EPA, Office of Air Quality Planning and
Standards, by E. H. Pechan and Associates. Report No. EPA420-R-00-020.
U.S. EPA, 2002. User's Guide to MOBILE6.0: Mobile Source Emission Factor Model. Office of
Transportation and Air Quality, Ann Arbor, MI January, 2002.
http://www.epa.gov/otaq/m6.htm
42
-------
Wesely, M.L., P.V. Doskey., and J.D. Shannon, 2002. Deposit! on Parameterizations for the Industrial
Source Complex (ISC3) Model, Draft, ANL/ER/TM-nn, DOE/xx-nnnn, Environmental
Research Division, Argonne National Laboratory, Argonne, IL.
43
-------
Table 4.2-1 Surface and upper air stations used in Houston study.
Station, Identifier
George Bush Airport, TX
Lake Charles, LA
Type
surface
upper air
Identifier, Number
IAH, 722430
LCH, 3937
Lat(°N), Lon(°W)
29.99; 95.36
30.12; 93.22
Elevation (m)
33
5
Table 4.2-2. 1996 and climatological wind speed, wind direction, average daily maximum
temperature, minimum daily temperature, and annual total rainfall.
Variable
Wind speed (ms"1)1
Wind direction (°)'
Average maximum daily temperature(°C)1
Average minimum daily temperature(°C)1
Annual accumulated rainfall (mm)2
1996
o o
J.J
138
26.2
15.1
748.5
Climatology
3.6
153
25.9
15.3
1,170
1. Climatology based on 1984-1992.
2. Climatology based on 1961-1990.
44
-------
Table 4.3-1. Corrected Location Coordinates of Point Sources in Houston Domain
Site Name
Simpson Paper
Champion International
BASF Corporation
Ethyl Corporation
Ethyl Corporation
Hoescht Celanese
Occidental Chemical
Texaco Chemical Company
Rolling Environmental Services
Huntsman Petrochemical Corporation
Ethyl Corporation
Exxon Comp USA
Citgo Pipeline
ACT-ID
48201-12359
48201-12405
48039-12765
48201-50029
48201-50027
48201-12733
48201-12749
48339-12759
48201-15980
48201-53638
48201-50000
48201-50253
48201-47361
Corrected Point
Source Inventory
Location Variables
Coordinate
System
UTM
UTM
UTM
UTM
UTM
UTM
UTM
UTM
UTM
NC
NC
NC
NC
UTM
Zone
15
15
15
15
15
15
15
15
15
NC
NC
15
15
X
285900
296363
266202
290143
290143
300334
298400
270737
297414
NC
NC
NC
NC
Y
3289700
3307528
3210415
3291068
3291068
3278837
3290970
3355860
3290047
3275.68
3292.29
NC
NC
NC= No Change
UTM = Universal Transverse Mercator
45
-------
Table 4.3-2. Houston/Vehicle Split Table for Allocation of Road Segment Emissions.
National 1 997 VMT (millions of Miles Traveled)
Vehicle Type1
Road Type2
Rural INT
Rural OF
Rural MA
Rural MJC
Rural MNC
Rural LOG
Urban INT
Urban OF
Urban OP
Urban MA
Urban COL
Urban LOC
LDGV
55.06%
55.06%
55 06%
55 06%
55.06%
55.06%
60 39%
60 39%
60.39%
60.39%
60 39%
60 39%
LDGT1
21.00%
21.00%
21 00%
21 00%
21.00%
21.00%
21 95%
21 95%
21.95%
21.95%
21 95%
21 95%
LDGT2
10.70%
10.70%
10 70%
10 70%
10.70%
10.70%
11 18%
11 18%
11.18%
11.18%
11 18%
11 18%
HDGV
2.81%
2.81%
2 81%
2 81%
2.81%
2.81%
1 63%
1 63%
1.63%
1.63%
1 63%
1 63%
MC
0.42%
0.42%
0 42%
0 42%
0.42%
0.42%
0 38%
0 38%
0.38%
0.38%
0 38%
0 38%
1.
LDGT1 - Light Duty Gasoline Trucks (<= 6,000 Ibs.)
LDGT2 - Light Duty Gasoline Trucks (6,000 - 8,500 Ibs.)
HDGV - Heavy Duty Gasoline Vehicles
MC - Motorcycles
LDDV - Light Duty Diesel Vehicles
LDDT - Light Duty Diesel Trucks
2B HDDV - Heavy Duty Diesel Vehicles (6,001 - 10,000 Ibs.)
Light HDDV - Heavy Duty Diesel Vehicles (10,001 - 19,500 Ibs.)
Medium HDDV - Heavy Duty Diesel Vehicles (19,501 - 33,000 Ibs.)
Heavy HDDV - Heavy Duty Diesel Vehicles (> 33,000 Ibs.)
LDDV
0.76%
0.76%
0 76%
0 76%
0.76%
0.76%
0 83%
0 83%
0.83%
0.83%
0 83%
0 83%
LDDT
0.26%
0.26%
0 26%
0 26%
0.26%
0.26%
0 27%
0 27%
0.27%
0.27%
0 27%
0 27%
2B HDDV
0.01%
0.01%
0 01%
0 01%
0.01%
0.01%
0 00%
0 00%
0.00%
0.00%
0 00%
0 00%
Light HDDV
0.24%
0.24%
0 24%
0 24%
0.24%
0.24%
0 09%
0 09%
0.09%
0.09%
0 09%
0 09%
Medium HDDV
0.91%
0.91%
0 91%
0 91%
0.91%
0.91%
0 34%
0 34%
0.34%
0.34%
0 34%
0 34%
Heavy HDDV
7.37%
7.37%
7 37%
7 37%
7.37%
7.37%
2 76%
2 76%
2.76%
2.76%
2 76%
2 76%
Buses
0.46%
0.46%
0 46%
0 46%
0.46%
0.46%
0 17%
0 17%
0.17%
0.17%
0 17%
0 17%
2.
INT - Interstate
OF - Other Freeway or Expressway
MA - Minor Arterial
MJC - Major Collector
MNC - Minor Collector
OP - Other Principal
LOC - Local
COL - Collector
46
-------
Table 4.3-3. Types and Dimensions of Roadway Segments.
Roadway Type
Interstate Roads
U.S. Highways
State Highways
Lanes
8+5 median/shoulder
6+3 median/shoulder
4+2 median/shoulder
Width (Lane=3.65m)
47.45
32.85
21.90
Min. Length(m)
0.4745
0.3285
0.2190
Max Length(m)
4745
3285
2190
Table 4.4-1. Maximum Annual Average Total (From All Sources) Concentration and Southwest Corner of Grid Cell For Each
HAP in Study1.
HAP
Benzene
Cadmium
Chromium
Formaldehyde
Lead
Maximum
Concentration
Cug nr3)
5.245
0.002935
0.01578
5.41(14.2)3
0.3651
(X,Y) (UTM)
(285,000,3,290,000)
(314,000,3,252,000)
(314,000,3,252,000)
(279,000,3,281,000)
(279,000,3,281,000)
Major Source
Concentration
Cug nr3)
4.317
0.00283
0.01551
0.0516
0.00011
Area/Other
Source
Concentration
Cug nr3)
0.193
0.0001
0.00008
0.0865
0.00346
Onroad Mobile
Concentration
Cug nr3)
0.4848
N/A2
0.00001
0.251
0.00006
Nonroad Mobile
Concentration (/^g m 3)
0.2633
0
0.00018
5.06
0.3618
1. Background not included.
2. Not/Applicable
3. Value in Parentheses includes secondary contribution
47
-------
Table 4.4-2. Maximum Concentrations and Southwest Corner of Location For Each Category and HAP.
HAP
Benzene
Cadmium
Chromium
Formaldehyde
Lead
Major Concentration (/^g m 3) (x;y)
4.32 (285,000;3,290,000)
0.003(314,000;3,252,000)
0.0155 (3 14,000;3,252,000)
1.01 (293,000;3,281,000)
0.0083 (3 14,000;3,252,000)
Area/Other Concentration
Cug m-3) (x;y)
0.505 (305,000;3,290,000)
0.0014 (273,000;3,299,000)
0.0025 (231,000; 3,276,000)
0.328 (288,000;3,268,000)
0.0204 (289,000;3,293,000)
Onroad Mobile Concentration
Cug m-3) (x;y)
3.68 (274,000;3,304,000)
N/A
0.0003 (274,000; 3,304,000)
1.84 (274,000;3,304,000)
0.0004 (274,000;3,304,000)
Nonroad Mobile
Concentration (/^g m 3)
(x;y)
1.03 (279,000; 3,281,000)
0.0001 (306,000;3,284,000)
0.004 (306,000; 3,284,000)
5.06 (279,000;3,281,000)
0.3618 (279,000;3,281,000)
Table 4.4-3. Benzene Maximum Total Concentration and Southwest Corner of Location for ISCST3, ISCST3 ROADS,
ISCST3 FINE GRID.
Model
ISCST3
ISCST3 ROADS
Maximum
Concentration
(Mgm-3)1
5.245
7.09
(X,Y) (UTM)
(285000,3290000)
(262000,3286000)
Major Source
Concentration
Cug m-3)
4.317
0.0761
Area/Other
Source
Concentration
Cug m-3)
0.193
0.159
Onroad Mobile
Concentration
Cug m-3)
0.4848
6.43
Nonroad Mobile
Concentration (/^g m 3)
0.2633
0.466
48
-------
Table 4.4-4. Formaldehyde Maximum Concentrations and by concentration source category for ISCST3 and ISCST3 ROADS.
Model
ISCST3
ISCST3 ROADS
Maximum
Concentration
(Mgm-3)1
5.41(14.2)
5.59(14.4)
(X,Y) (UTM)
(279000,3281000)
(279000,3281000)
Major Source
Concentration
Cug nr3)
0.0516
0.0516
Area/Other
Source
Concentration
Cug nr3)
0.0865
0.0865
Onroad Mobile
Concentration
Cug nr3)
0.251
0.413
Nonroad Mobile
Concentration (/^g m 3)
5.06
5.06
1. Value in Parentheses includes secondary contribution
49
-------
3305
3300 -
, 3290 -
280
285 290
UTM Zone 15 West-East Distance (km)
295
300
240 260 280 300
UTM Zone 15 West-East Distance (km)
Figure 4.1-1. a) Houston domain with key roads and location of airport and b) ISCST3 FINE GRID with key roads.
-------
ISCST3 Urban (gray) and Rural (white) 1x1 km cells and Receptors
*
*
Receptors
Monitors
220
240
260 280 300
UTM Zone 15 West-East Distance (km)
Figure 4.2-1. Locations of urban and rural grid cells, ISC receptors, and monitors.
320
-------
N
to
3i
Mldnfgh-t-11 PM
NOTE: Frequencies
indicate direction
froM which the
wind is blowing.
/ / /
\ \ \ \
\ \ \ \
ill/
/•/'//
\ \ \ \
i i i /
\ \ \ \
MIND SPEED (KNOTS)
7_10 11-16
CALM MINDS 18.19X
Figure 4.2-2. Wind rose of winds for Houston, 1996.
-------
Annual Average Benzene Background Concentrations ^ig/m ], Houston, TX, 1996
Annual Average Lead Background Concentrations ^ig/m ], Houston, TX, 1996
240 260 280 300
UTM Zone 15 West-East Distance (km)
220 240 260 280 300
UTM Zone 15 West-East Distance (km)
0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45
Figure 4.2-3. Annual average background concentrations (jig nr3) for a) benzene and b) lead.
-------
Houston
^
areaPREprepl.sas p
Regular Area
Houston domain
Landfills
CAreaPrepJ /* i Ic-ri~i -""IX
-— (. Iandfills2pomt.sas )
AMFinalFormat
"1
SO files SASdata
hou Iandfill2
Houston domain
corrected point
source: manually
corrected "need xy
data"
ouston ISCprepro
— -^
Point,landfills,
1
<^PtDataProT^>
Figure 4.3-1. Flowchart of gridded emissions processing for Houston.
-------
ISC-ready Benzene Total 1x1 km Emissions [tons/yr], Houston, TX, 1996
3340 -
3320 -
CD
O
c
ro
"55
b
3300 -
o
i
o
o
N
3280 -
3260 -
220
240 260 280 300
UTM Zone 15 West-East Distance (km)
320
0.2 0.4 0.6 0.8
1.2
1.6
Figure 4.4-1. Benzene 1 km gridded emissions (tons yr1) from all sources.
55
-------
ISC-ready Benzene Major Source 1x1 km Emissions [tons/yr], Houston, TX, 1996
ISC-ready Benzene Onroad Mobile Source 1x1km Emissions [tons/yr], Houston, TX, 1996
240 260 280 300
UTM Zone 15 West-East Distance (km)
220 240 260 280 300 320
UTM Zone 15 West-East Distance (km)
0.2 0.4 0.6 0.8 1 1.2 1.4
Figure 4.4-2 Benzene 1 km gridded a) major source emissions, and b) onroad mobile emissions. Emissions are in tons yr1.
-------
Houston Benzene Emissions: 4,368 tons
U.S. Benzene Emissions: 336,544 tons
Chemical &
AlliedProducts
5%
Petroleum/Coal
Products
4%
Remaining
Sources
6%
Oil & Nat.
Gas Prod
2%
Ethylene Processes
POTW 4o/0
5%
Remaining
Sources
11%
Petroleum/Coal
Products
12%
Chemical &
AlliedProducts
Fires
(Forest/Wildfires
8%
Chemical & Allied
Products
5%
Gasoline
Dist. Stages
1&2
2%
Onroad
Mobile
Sources
50%
Nonroad
Mobile
Sources
28%
17%
Houston Benzene
Major Emissions: 1,255 tons
Houston Benzene
Area Emissions: 229 tons
Houston Benzene
Mobile Emissions: 2,883 tons
Figure 4.4-3a. Distribution of emissions for Houston and U.S. inventories for benzene.
-------
Houston Cadmium Emissions: 1.4 tons
U.S. Cadmium Emissions: 157 tons
-------
Houston Chromium Emissions: 6.4 tons
Haz. Waste
Incin.
3%
U.S. Chromium Emissions: 1,037 tons
Chemical &
Allied
Products
13%
Petroleum
Refineries
29%
Nonroad
Mobile
Sources
16%
Ship Building Metal Parts
3% 2%
Remaining
Sources
6%
Chemi/al & Allied
oducts
20%
Houston Chromium
Major Emissions: 4.3 tons
Industrial Equip.
13%
Fabricated Metal
Products
13%
Hard Chromium
Electroplating
14%
Primary
Metal
Products
7%
Misc. Metal
Parts
10%
Hard
Chromium
Electroplating
12%
Utility
Services
5%
Houston Chromium
Mobile Emissions: 1.1 tons
Houston Chromium Area Emissions: 1 tons
Figure 4.4-3c. Distribution of emissions for Houston and U.S. inventories for chromium.
-------
Stationary
Recip.
Internal
Comb.
Engines
Houston Formaldehyde Emissions: 2,811 tons
Chemical &
Oil and Gas Allied
Extraction Products
4% 2%
U.S. Formaldehyde Emissions: 301,708 tons
5%
Remaining
Sources
8%
Nonroad
Mobile
Sources
46%
On road
Mobile
Sources
35%
Remaining
Sources
Stationary 6%
Recip.
Internal
Comb.
Engines
7%
Petroleum/
Coal
Products
1%
Utility Boilers
3%
Utility Services
3%
Chemical & Allied
Remaining
Sources
4%
Remaining
Sources
3%
Stationary
Comb. Turbines
22%
Fires(Forest/Wild/
Presc.)
Utility Boilers
2%
tationary Recip.
Internal Conjb.
Engines I
Houston Formaldehyde
Major Emissions: 252 tons
23% Houston Formaldehyde
Area Emissions: 300 tons
Houston Formaldehyde
Mobile Emissions: 2,259 tons
Figure 4.4-3d. Distribution of emissions for Houston and U.S. inventories for formaldehyde.
-------
Houston Lead Emissions: 29 tons
U.S. Lead Emissions: 2,613 tons
Utility Boilers
6%
Portland
Cement Manu.
5%
Primary
Lead
Smelting
12%
Primary
Metal
Industries
Nonroa
Mobile
Sources
20%
Remaining Sources
7%
Houston Lead
Major Emissions: 7.7 tons
Remaining Sources
6%
Fabricated
Metal Products
6%
Autobody
Re finishing-
7%
Houston Lead
Area Emissions: 8.1 tons
Onroad Mobile
2%
Houston Lead
Mobile Emissions: 13 tons
Figure 4.4-3e. Distribution of emissions for Houston and U.S. inventories for lead.
-------
Emissions (tons)
Source Category
Nonroad Mobile Sources
Onroad Mobile Sources
Hazardous Waste Incineration
Petroleum Refineries
Utility Boilers: Coal, Oil, and Natural
Chemicals and allied products
Petroleum Refineries - Other Sources
Portland Cement Manufacturing
Chromium Metal Plating
Primary metal industries
Stationary Reciprocating Internal Combustion Engines
Petroleum and coal products
Fabricated metal products
Oil and gas extraction
Industrial/Commercial/ Institutional Boilers
Stationary Combustion Turbines
Pulp and Paper
Industrial machinery and equipment
Hard Chromium Electroplating
Autobody Refinishing Paint Application
Stone, clay, glass, and concrete products
Shipbuilding & Ship Repair (Surface Coating)
Oil & Natural Gas Production
Open Burning: Forest and Wildfires
Pubicly Owned Treatment Works (POTW)
Residential Heating: Distillate Oil
Miscellaneous Metal Parts & Products
Open Burning: Prescribed Burnings
Ethylene Processes
Residential Heating: Wood/Wood Residue
Medical Waste Incinerators
Natural Gas Transmission & Storage
Gasoline Distribution (Stage 1)
Asphalt roofing and Processing
Steel Foundries
Motor freight transportation and warehousing
Hydrogen Fluoride Production: GENERIC
Industrial Cooling Towers
Marine Vessel Loading Operations
Gasoline Distribution Stage II
Electric, gas, and sanitary services
Petroleum Refining and Natural Gas Support
Surface Coatings: Architectural
Iron Foundries
Structure Fires
MON - combined
Residential Heating: Bituminous and Lignite Coal
Municipal Solid Waste Landfills - NSPS&
Residential Heating: Natural Gas
Consumer Products Usage
Lumber and wood products
Primary Copper Smelting
Open Burning: Scrap Tires
Off-Site Waste and Recovery Operations
Refractories Products Manufacturing
Rubber and miscellaneous plastics products
Transportation equipment
Chromic Acid Anodizing
Residential Heating: Anthracite Coal
Transportation services
Metal Can (Surface Coating)
Asphalt Concrete Manufacturing
Primary Lead Smelting
Wood Furniture (Surface Coating)
Metal Coil (Surface Coating)
Pipelines, except natural gas
Automotive repair, services, and parking
Aviation Gasoline Distribution: Stage 1 & II
TYPE
MACT
MACT
MACT
SIC
MACT
MACT
CAT
SIC
MACT
SIC
SIC
SIC
MACT
MACT
MACT
SIC
MACT
CAT
SIC
MACT
MACT
CAT
MACT
CAT
MACT
CAT
MACT
CAT
MACT
MACT
MACT
MACT
MACT
SIC
MACT
MACT
MACT
CAT
SIC
CAT
CAT
MACT
CAT
MACT
CAT
MACT
CAT
CAT
SIC
MACT
CAT
MACT
MACT
SIC
SIC
MACT
CAT
SIC
MACT
CAT
MACT
MACT
MACT
SIC
SIC
CAT
CODE
0801
0502
1808
28xx
0503
0410
9069
33xx
0105
29xx
34xx
13xx
0107
0108
1626
35xx
1615
9027
32xx
0715
0501
9306
0803
9380
0710
9307
1635
9382
1801
0504
0601
0418
0309
42xx
1409
1619
0603
9166
49xx
9327
9427
0308
9425
1640
9379
0802
9381
9087
24xx
0203
9308
0806
0406
30xx
37xx
1607
9378
47xx
0707
9023
0204
0716
0708
46xx
75xx
9032
BENZENE
4368.00
23.55%
42.46%
0%
0.11%
0.13%
5.08%
14.65%
0%
0%
0%
0.11%
3.55%
0%
0.02%
0.03%
0.02%
0.17%
0.20%
0%
0%
0.01%
0.02%
1.86%
0.40%
1.56%
0%
0.02%
0.25%
1.07%
0%
0%
0.99%
0.98%
0.01%
0.02%
0.69%
0.40%
0.43%
0.31%
0.30%
0.02%
0.22%
0.18%
0%
0%
0.02%
0%
0.10%
0%
0%
0%
0%
0.04%
0%
0%
0%
0.04%
0%
0%
0.01%
0%
0%
0%
0%
0%
0.01%
0%
0%
CADMIUM
1.36
1.30%
0%
43.81%
25.28%
16.19%
0%
0%
4.69%
4.41%
0%
0.03%
0%
0%
0%
1.71%
0%
0.64%
0%
0%
0%
0%
0%
0%
0%
0%
0.98%
0%
0%
0%
0.29%
0.60%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0.01%
0.04%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0.02%
0%
0%
0%
0%
0%
0%
0%
0%
0%
CHROMIUM
6.38
15.69%
2.21%
3.42%
29.29%
17.61%
13.35%
0%
0.66%
2.51%
0.40%
0.01%
0%
2.00%
0%
0.93%
0%
0.29%
1.96%
2.15%
0%
1.88%
1.73%
0%
0%
0%
0.21%
1.16%
0%
0%
0.41%
0.01%
0%
0%
0.83%
0.90%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0.08%
0.04%
0%
0%
0%
0.08%
0%
0.03%
0%
0.07%
0.02%
0%
0.06%
0.02%
0%
0%
0%
0%
0%
0%
0%
0%
0%
FORMALDEHYDE
2811.00
45.65%
34.73%
0%
2.28%
0.27%
2.48%
0%
0%
0%
0%
5.12%
0%
0.01%
3.54%
0.18%
2.37%
0.16%
0%
0%
0%
0%
0%
0%
1.42%
0%
0.05%
0.01%
0.88%
0%
0.05%
0%
0%
0%
0%
0%
0%
0.07%
0%
0%
0%
0.26%
0%
0%
0%
0.17%
0.02%
0%
0%
0.09%
0.08%
0%
0%
0%
0%
0%
0%
0.03%
0%
0%
0.01%
0.01%
0%
0%
0.01%
0.01%
0%
0.01%
0%
LEAD
28.57
43.97%
0.70%
15.82%
3.53%
6.35%
12.20%
0%
5.11%
0%
4.97%
0%
0.42%
1.90%
0%
0.35%
0%
1.12%
0%
0%
1.97%
0.03%
0.12%
0%
0%
0%
0.18%
0.02%
0%
0%
0.28%
0.39%
0%
0%
0.12%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0.18%
0%
0.01%
0.02%
0%
0%
0%
0%
0.08%
0%
0.07%
0%
0.05%
0%
0%
0.01%
0%
0%
0.01%
0.01%
0%
0%
0%
0%
0.01%
Figure 4.4-3f. Percent contribution of individual source categories for each pollutant in
the Houston domain.
62
-------
ISC-ready Benzene Harris County Onroad Mobile Segments Emissions [tons/(yr-km )], Houston, TX, 1996
ISC-ready Benzene Remaining Onroad Mobile 1x1 km Emissions [tons/yr] after Extracting Roads, Houston, TX, 1996
ON
220 240 260 280 300 320
UTM Zone 15 West-East Distance (km)
240 260 280 300
UTM Zone 15 West-East Distance (km)
0.2 0.4 0.6 0.8 1 1.2 1.4 1.6
Figure 4.4-4. Benzene a) road segment emissions (tons yr1 km"2) b) remaining 1 km gridded onroad mobile emissions (tons yr1)
after extracting road segment emissions.
-------
3340 -
220
240 260 280 300
UTM Zone 15 West-East Distance (km)
320
Figure 4.4-5 Benzene ISCST3 BASE annual average total concentrations (|lg m~3).
64
-------
\J 0.00701-0.059)ig;m
0.0592-0.0871 }ig/m
0.0872-0.138 n;g/m3
0.139-0.239 ng/m3
0.24-4.32 ng/rrt3
\J 0.0205-0.0761 (ig/m
0.0762-0.10
0.106-0.122
Q 0.122-0.14 ng/m
• 0.14-0.505
240 260 280
UTM Zone 15 West-East Distance (km)
240 260 280
UTM Zone 15 West-East Distance (km)
Figure 4.4-6. Benzene annual average ISCST3 BASE concentrations (|ig m"3). a) major sources, b) area and other sources, c)
onroad mobile, and d) nonroad mobile sources.
-------
Oi
[] 0.0358-0.23
| 0.232-0.394 |ig/m
0.395-0.487
0.487-0.591
0.592-1.03 iig/m
240 260 280 300
UTM Zone 15 West-East Distance (km)
240 260 280 300
UTM Zone 15 West-East Distance (km)
Figure 4.4-6. Continued.
-------
240 260 280 300
UTM Zone 15 West-East Distance (km)
240 260 280 300
UTM Zone 15 West-East Distance (km)
Figure 4.4-7. Benzene annual average ISCST3 ROADS concentrations (|lg m"3) for a) total (all sources) concentrations, and b) onroad
mobile concentrations.
-------
oo
240 260 280 300
UTM Zone 15 West-East Distance (km)
240 260 280 300
UTM Zone 15 West-East Distance (km)
Figure 4.4-8. Benzene percent differences for ISCST3 ROADS minus ISCST3 BASE for a) total (all sources)
concentrations and b) onroad mobile concentrations.
-------
Benzene Largest Concentration Contributions by Source Category
Benzene Largest Concentration Contributions by Source Category
ON
VO
240 260 280 300
UTM Zone 15 West-East Distance (km)
220 240 260 280 300 320
UTM Zone 15 West-East Distance (km)
Figure 4.4-9. Largest source contributor at each receptor for benzene total concentrations for a) ISCST3 BASE and
b) ISCST3 ROADS.
-------
Annual Average Concentrations Jig/m ], ISCST3 ROADS, Houston, TX, 1996
Benzene, with Near-Source Receptors, All Sources- Incl. Background
COLORBAR SCALED TO 98TH-PERCENTILE TO SHOW CONCENTRATION GRADIENTS, NOT BIGGEST HOTSPO"
3305, 1 1
Annual Average Concentrations Jig/m ], ISCST3 ROADS, Houston, TX, 1996
Benzene, with Near-Source Receptors, All Sources- Incl. Background
COLORBAR SCALED TO MAX CONCENTRATION SHOWS BIGGEST HOTSPOTS
ar-2002
285 290
UTM Zone 15 West-East Distance (km)
285 290
UTM Zone 15 West-East Distance (km)
Figure 4.4-10. Benzene ISCST3 FINE GRID annual average total concentrations (|ig m"3). a) scaled to 98th percentile of
concentrations and b) scaled to maximum concentration.
-------
Annual Average Concentrations Jig/m ], ISCST3 ROADS, Houston, TX, 1996
Benzene, with Near-Source Receptors, Major Sources
COLORBAR SCALED TO 98TH-PERCENTILE TO SHOW CONCENTRATION GRADIENTS, NOT BIGGEST HOTSPOTS
3305 r
Annual Average Concentrations Jig/m ], ISCST3 ROADS, Houston, TX, 1996
Benzene, with Near-Source Receptors, Major Sources
COLORBAR SCALED TO MAX CONCENTRATION SHOWS BIGGEST HOTSPOTS
285 290
UTM Zone 15 West-East Distance (km)
285 290
UTM Zone 15 West-East Distance (km)
Figure 4.4-11. Benzene ISCST3 FINE GRID annual average major source concentrations (|lg m"3) a) scaled to 98th
percentile of concentrations and b) scaled to maximum concentration.
-------
Annual Average Concentrations Jig/m ], ISCST3 ROADS, Houston, TX, 1996
Benzene, with Near-Source Receptors, Area and Other Sources
COLORBAR SCALED TO 98TH-PERCENTILE TO SHOW CONCENTRATION GRADIENTS, NOT BIGGEST HOTSPO"
Annual Average Concentrations Jig/m ], ISCST3 ROADS, Houston, TX, 1996
Benzene, with Near-Source Receptors, Area and Other Sources
COLORBAR SCALED TO MAX CONCENTRATION SHOWS BIGGEST HOTSPOTS
to
285 290
UTM Zone 15 West-East Distance (km)
285 290
UTM Zone 15 West-East Distance (km)
Figure 4.4-12. Benzene ISCST3 FINE GRID annual average area source concentrations (|ig m"3) a) scaled to 98th
percentile of concentrations and b) scaled to maximum concentration.
-------
Annual Average Concentrations Jig/m ], ISCST3 ROADS, Houston, TX, 1996
Benzene, with Near-Source Receptors, Onroad Mobile Sources
COLORBAR SCALED TO 98TH-PERCENTILE TO SHOW CONCENTRATION GRADIENTS, NOT BIGGEST HOTSPOTS
3305
Annual Average Concentrations Jig/m ], ISCST3 ROADS, Houston, TX, 1996
Benzene, with Near-Source Receptors, Onroad Mobile Sources
COLORBAR SCALED TO MAX CONCENTRATION SHOWS BIGGEST HOTSPOTS
ar-2002
285 290
UTM Zone 15 West-East Distance (km)
285 290
UTM Zone 15 West-East Distance (km)
Figure 4.4-13. Benzene ISCST3 FINE GRID annual average onroad mobile source concentrations (jig nr3) a) scaled to
98th percentile of concentrations and b) scaled to maximum concentration.
-------
Annual Average Concentrations tig/m ], ISCST3 ROADS, Houston, TX, 1996
Benzene, with Near-Source Receptors, Non-road Mobile Sources
COLORBAR SCALED TO 98TH-PERCENTILE TO SHOW CONCENTRATION GRADIENTS, NOT BIGGEST HOTSPOT
Annual Average Concentrations Jig/m ], ISCST3 ROADS, Houston, TX, 1996
Benzene, with Near-Source Receptors, Non-road Mobile Sources
COLORBAR SCALED TO MAX CONCENTRATION SHOWS BIGGEST HOTSPOTS
8
.ffl 3295
N 3290
N 3290
285 290
UTM Zone 15 West-East Distance (km)
280 285 290
UTM Zone 15 West-East Distance (km)
Figure 4.4-14. Benzene ISCST3 FINE GRID annual average nonroad mobile source concentrations (|lg m"3) a) scaled to
98th percentile of concentrations and b) scaled to maximum concentration.
-------
ISC-ready Cadmium Total 1x1km Emissions [tons/yr], Houston, TX, 1996
ISC-ready Cadmium Area Source 1x1 km Emissions [tons/yr], Houston, TX, 1996
240 260 280 300
UTM Zone 15 West-East Distance (km)
240 260 280 300
UTM Zone 15 West-East Distance (km)
Figure 4.4-15 Cadmium 1 km gridded emissions (tons yr1) for a) all sources, and b) area/other sources.
-------
Cadmium Largest Concentration Contributions by Source Category
2e-05-0.0001
0.00016-0.00025 ng/m
0.00025-0.00031 ng/m
0.000315-0.00041
0.00041 -0.002935 iig/m
240 260 280 300
UTM Zone 15 West-East Distance (km)
240 260 280 300
UTM Zone 15 West-East Distance (km)
Figure 4.4-16. Cadmium ISCST3 annual average a) total concentrations (jig nr3) and b) largest source contributor at each receptor.
-------
ISC-ready Chromium Total 1x1 km Emissions [tons/yr], Houston, TX, 1996
3340 -
220
240 260 280 300
UTM Zone 15 West-East Distance (km)
320
x10
Figure 4.4-17. Chromium 1 km gridded emissions (tons yr1) from all sources.
77
-------
Chromium Largest Concentration Contributions by Source Category
OO
\J 0.00016-0.00077 ng/m3
| 0.00078-0.00099 ng/m3
0.00099-0.00121 ng/m3
0.00121-0.00161 ng/m3
0.00161 -0.01578 iig/m3
240 260 280 300
UTM Zone 15 West-East Distance (km)
240 260 280 300
UTM Zone 15 West-East Distance (km)
Figure 4.4-18. Chromium ISCST3 annual average a) total concentrations (jig nr3) and b) largest source contributor at each receptor.
-------
ISC-ready Formaldehyde Total 1x1 km Emissions [tons/yr], Houston, TX, 1996
3340 -
3320 -
CD
O
c
ro
"55
b
3300 -
o
z
I
"
o
in
CD
o
N
3280 -
3260 -
220
240 260 280 300
UTM Zone 15 West-East Distance (km)
320
0.2
0.4
0.6
0.8
Figure 4.4-19. Formaldehyde 1 km gridded emissions (tons yr1) from all sources.
79
-------
ISC-ready Formaldehyde Harris County Onroad Mobile Segments Emissions [tons/(yr-km )], Houston, TX, 1996
3340
3320
CD
O
I
3300
o
W
in
o
N
3280
3260
> 0 - 0.006
0.006 - 0.032
0.032-0.115
0.115-0.243
0.243- 1.02
Monitors
220
240 260 280 300
UTM Zone 15 West-East Distance (km)
320
Figure 4.4-20. Formaldehyde road segment emissions (tons yr1 km"2).
80
-------
ISC-ready Formaldehyde Remaining Onroad Mobile 1x1km Emissions [tons/yr] after Extracting Roads, Houston, TX, 19:
ISC-ready Formaldehyde Onroad Mobile Source 1x1km Emissions [tons/yr], Houston, TX, 1996
OO
f 3300
I
220 240 260 280 300 320
UTM Zone 15 West-East Distance (km)
240 260 280 300
UTM Zone 15 West-East Distance (km)
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8
Figure 4.4-21. Formaldehyde 1 km onroad mobile gridded emissions (tons yr1) for a) remaining onroad emissions after extracting
road segment emissions and b) ISCST3 BASE onroad mobile gridded emissions.
-------
oo
to
[J 8.9-9.46 n
| 9.46-9.88
10.1 ng/rrr3
10.1-10.3 ng/rrf
10.3-14.2 iig/rri3
[] 0.0388-0.207
0.207-0.296
0.297-0.377
Q 0.378-0.508
• 0.51-1.84
240 260 280 300
UTM Zone 15 West-East Distance (km)
240 260 280 300
UTM Zone 15 West-East Distance (km)
Figure 4.4-22. Formaldehyde ISCST3 BASE annual average concentrations (|lg nr3) for a) all sources, and b) onroad mobile
sources.
-------
oo
[] 8.903-9.505
9.505-10.016 ra/m3
10.026-10.276iig/m
10.286-10.466tig/m
10.466-14.376 iig/m
; 3300 -
240 260 280 300
UTM Zone 15 West-East Distance (km)
240 260 280 300
UTM Zone 15 West-East Distance (km)
Figure 4.4-23. Formaldehyde ISCST3 ROADS annual average concentrations (|lg nr3) for a) all sources, and b) onroad mobile
sources.
-------
oo
240 260 280 300 320
UTM Zone 15 West-East Distance (km)
240 260 280 300
UTM Zone 15 West-East Distance (km)
Figure 4.4-24. Formaldehyde percent differences for ISCST3 ROADS minus ISCST3 BASE for a) total
concentrations and b) onroad mobile concentrations.
-------
Formaldehyde Largest Concentration Contributions by Source Category
Formaldehyde Largest Concentration Contributions by Source Category
OO
240 260 280 300
UTM Zone 15 West-East Distance (km)
240 260 280 300
UTM Zone 15 West-East Distance (km)
Figure 4.4-25. Largest source contributor for each recptor for formaldehyde for a) ISCST3 BASE and b) ISCST3 ROADS.
-------
ISC-ready Lead Total 1x1 km Emissions [tons/yr], Houston, TX, 1996
3340 -
220
240 260 280 300
UTM Zone 15 West-East Distance (km)
320
x10"
Figure 4.4-26. Lead 1 km gridded emissions (tons yr1) for a) all sources,
b) area/other sources., and c) nonroad mobile sources.
86
-------
ISC-ready Lead Area Source 1x1 km Emissions [tons/yr], Houston, TX, 1996
ISC-ready Lead Nonroad Mobile Source 1x1km Emissions [tons/yr], Houston, TX, 1996
OO
240 260 280 300
UTM Zone 15 West-East Distance (km)
240 260 280 300
UTM Zone 15 West-East Distance (km)
0.5 1 1.5 2 2.5
0.05 0.1 0.15 0.2 0.25
Figure 4.4-26. Continued.
-------
3340 -
3320 -
CD
o
I
o
Z
I
si
o
in
o
N
3300 -
3280
3260 -
220
240 260 280 300
UTM Zone 15 West-East Distance (km)
320
Figure 4.4-27. Lead annual average ISCST3 concentrations (jig nr3) for a) all sources, b)
area/other sources, and c) nonroad mobile sources.
-------
oo
VO
[] 0.0003-0.00176 ng/m3
0.00177-0.00299 ng/m3
0.00303-0.00394 ng/m3
0.00394-0.00526 ng/m3
0.00526-0.02035 iig/m3
[] 0.00017-0.00144 ng/m3
0.00144-0.00243 3
0.00244-0.00396 ng/m
Q 0.004-0.00759tig/m3
• 0.00761-0.3618 |ig/m3
240 260 280
UTM Zone 15 West-East Distance (km)
240 260 280 300
UTM Zone 15 West-East Distance (km)
Figure 4.4-27. Continued.
-------
Lead Largest Concentration Contributions by Source Category
3340 -
3320 -
CD
O
c
ro
"55
b
3300 -
o
N
3280 -
3260 -
220
240 260 280 300
UTM Zone 15 West-East Distance (km)
320
Figure 4.4-28. Largest source contributor at each receptor for lead.
90
-------
"E5
1
O>
o
53
ISCST3 BASE (star), ISCST3 ROADS (circle), and Monitors(box)
Monitor Average Cone.: 3.11|ig/m
ISCST3 BASE Average Cone.: 2.22|ig/m3
ISCST3 BASE RMS error: 0.99
ISCST3 ROADS Average Cone.: 2.36|ig/m3~'
ISCST3 ROADS RMS error: 0.78
5
Monitor
8
Figure 4.4-29. ISCST3 BASE (star), ISCST3 ROADS (circles) and monitor (box) annual average concentrations
(jig nr3) for benzene.
-------
Appendix A
Estimating Background Concentration for Diesel PM
-------
APPENDIX A - TABLE OF CONTENTS
1. INTRODUCTION A-l
2. APPROACH TO DEVELOP CONCENTRATION VS. DISTANCE A-l
3. ESTIMATING BACKGROUND CONCENTRATIONS A-l
4. STUDY LIMITATIONS A-2
5. REFERENCES A-3
A-ii
-------
LIST OF FIGURES
Figure A-l. Annual average CALPUFF concentrations for Pittsburgh, Boise,
and Medford A-4
Figure A-2. Average concentration curves averaged among Pittsburg, Medford, and Boise for
release heights of 2 and 35 meters A-5
Figure A-3. Census tract centroids (dots) and rings of 50 and 300 km centered over Houston, TX
(star) used for calculation of background concentrations A-6
A-iii
-------
1. INTRODUCTION:
Background concentrations are an essential part of the total air quality concentration to be
considered in determining source impacts. Background air quality includes pollutant
concentrations due to: 1) natural sources; 2) nearby sources that are unidentified in the
inventory; and 3) long range transport into the modeling domain. Typically, monitored air
quality data should be used to establish background concentrations.
The ISCST3 model calculates concentrations at receptors with a maximum distance of 50
km. Gaussian type models are not applied for distances greater than 50 km. However, sources at
distances more than 50 km from the receptor contribute to the total concentration at the receptor
location.
For diesel PM, a modeling based approach was developed to provide a rough approximation
of concentrations due to transport from sources located between 50 km and 300 km from the
receptor. This approximation was based on results from existing CALPUFF simulations from an
elevated source (35 m) and a surface release (2 m) for three geographical areas: Boise, ID,
Medford, OR, and Pittsburgh, PA. These simulations were made as part of a series of
simulations to compare ISC results with CALPUFF results (U.S. EPA, 1993). CALPUFF is a
Lagrangian puff model, which was originally designed for mesoscale applications, and it can
operate in a range of 0-300 km from the source (U.S. EPA, 1995). For these CALPUFF
simulations, CALPUFF was run using ISC meteorology. Therefore, these CALPUFF results are
not the result of a full-scale refined analysis, in which the meteorological conditions are allowed
to vary in space and time.
2. APPROACH TO DEVELOP CONCENTRATION VS DISTANCE
The annual average CALPUFF concentration estimates, normalized by the emission rate, are
shown in Figure A-l as a function of distance from the source for 3 cases. A spline polynomial
approximation was used to get analytical representation for the results shown in Figure A-l.
These parameterizations provide annual average concentrations in (jig m"3) at a distance 50 km <
x < 300 km from a low release source (Eq.la) and an elevated source (Eq.lb).
C=6.18022xlO-10xx4-5.2255xlO-7xx3+1.61998xlO-4xx2-2.22567xlO-2xx+1.215630 (la)
C=3.37367x 10-10xx4-2.91373 x lQ-7xx3+0.323 IQx 10-5xx2-l .3411 x lQ-2xx+0.784964
Average curves for all 3 geographical areas are shown in Figure A-2. The approximations
l(a-b) are also shown in the figure for a low level release and a release from the elevated source.
The source emission rate is assumed to be equal 100 g s"1.
3. ESTIMATING BACKGROUND CONCENTRATION
We introduce a method to calculate the "background" concentrations due to contribution
from emission sources located farther then 50 km. The method is based on a simplistic
A-l
-------
approach: first, a receptor grid is set up for the entire U.S. with a spatial resolution of 0.2 degree
latitude by 0.5 degree longitude. Second, for each receptor, all emission sources located at a
distance greater than 50 km and less than 300 km from the center of a grid box are considered.
These census tract emissions are based on the 1996 NTI. The emissions from each census tract
located from 50 to 300 km away from the center of the grid box are multiplied by a distance
dependent factor defined in equation la and summed up to obtain a concentration at the center of
the grid box. In this analysis, diesel PM emissions are from onroad and nonroad mobile sources,
which are released at ground level. Therefore, equation la is applicable. In these estimates, no
adjustment has been made to account for the variation in transport due to the climatology of wind
direction for the area being modeled.
A schematic plot showing the relationship between the census tract centroids at a distance
50 - 300 km and the grid box centroid is shown in Figure A-3. Here the center of the grid box is
shown as a star and a contribution from emission sources within a ring of 50 - 300 km is
considered. The "background" concentration at each grid box center is the sum of concentrations
resulting from all sources within the 50-300 km radius.
4. STUDY LIMITATIONS
The approach described above has several limitations. The estimates assume a complete and
accurate inventory. Use of the ISC meteorology in CALPUFF does not account for wind flow in
rivers and valleys as in mountainous terrain. The local wind flow patterns could cause
concentrations to be significantly different at specific locations. Some uncertainty is introduced
when averaging results over grid boxes instead of specific tracts. Using three specific locations
to obtain a national average parameterization is simplistic. Finally, using CALPUFF with site
specific information on emission release height, stack parameters, wet and dry deposition,
meteorological wind field, etc. would give different estimates. Thus, these estimates of the
impact of emissions located greater than 50 km but less than 300 km are considered as an
approximation of "background" concentration until more reliable estimates can be obtained from
monitoring data or when improved modeling techniques are developed.
This analysis suggests that the limitations of the ISCST3 model to calculate dispersion not
farther than 50 km model may cause underestimates of concentrations in certain areas, where
many sources with a high emission rate are located close to each other. Using a constant value
for the "background" concentrations does not seem to be accurate enough and these results
suggest a value for "background" should be computed for each receptor.
A-2
-------
5. REFERENCES
U.S. EPA, 1993. Interagency Workgroup on air Quality Modeling (IWAQM), Phase 1 Report:
Interim Recommendation for Modeling Long Range Transport and Impacts on Regional
Visibility, EPA-454/R-93-015; U.S. Environmental Protection Agency; Research Triangle
Park, 1993.
U.S. EPA, 1995. A User's Guide for the CALPUFF Dispersion Model, EPA-454/B-95-006; U.S.
Environmental Protection Agency; Research Triangle Park, 1995.
A-3
-------
"S
ctf
'
PQ
fin
:_
o
(L)
O
O
o
PH
fin
h-1
<:
O
(L)
2
(L)
"
op
P-H
SUO!)EJ)U90UOO
A-4
-------
t«
(L)
(N
<+H
O
00
'S
(L)
SUOIJBJJU9OUOO
'
PQ
ctf
(L)
fin
00
o
ctf
T3
(L)
00
oj
(L)
(L)
O
O
'I
-4— »
(L)
O
O
o
(L)
00
e
(L)
(N
A-5
-------
>
38
36
34
32
CD
T3
co
30
28
26
24
-108
• • '• •
• • T' •:'» • -:
.'. \t.:\
••-•v..
m * »-• •
« •• • ,«•• •:
* • •.«.«
-106
-104
-96
-94
-102 -100 -98
Longitude (W)
Figure A-3. Census tract centroids (dots) and rings of 50 and 300 km centered over Houston, TX (star)
used for calculation of background concentrations.
-92
-------
TECHNICAL REPORT DATA
(Please read Instructions on reverse before completing)
1. REPORT NO.
EPA-454/R-02-003
3. RECIPIENT'S ACCESSION NO.
4. TITLE AND SUBTITLE
Example Application of Modeling Toxic Air Pollutants in Urban
Areas
5. REPORT DATE
June 2002
6. PERFORMING ORGANIZATION CODE
7. AUTHOR(S)
8. PERFORMING ORGANIZATION REPORT
NO.
9. PERFORMING ORGANIZATION NAME AND ADDRESS
10. PROGRAM ELEMENT NO.
11. CONTRACT/GRANT NO.
12. SPONSORING AGENCY NAME AND ADDRESS
U.S. Environmental Protection Agency
Office of Air Quality Planning and Standards
Emissions, Monitoring and Analysis Division
Research Triangle Park, NC 27711
13. TYPE OF REPORT AND PERIOD COVERED
Final Report
14. SPONSORING AGENCY CODE
15. SUPPLEMENTARY NOTES
EPA Work Assignment Manager: Jawad S. Touma
16. ABSTRACT
Urban areas contain major sources and numerous smaller area sources and, as a result, air quality modeling analyses posses special
challenges. This document deals with the applications of the Industrial Source Complex (ISCST3) model which can estimate close
distance impacts from industrial facilities. ISCST3 has been extensively used in analyzing impacts from a single or a few facilities
and this document provides a transition to the more complex issues associated with urban-wide applications. This document: 1)
provides a demonstration of a methodology for modeling air toxics for use in city-specific analyses and an example application, 2)
updates techniques described in Dispersion Modeling of Toxics Pollutants in Urban Areas, EPA-454/R-99-021 and, 3) incorporates
techniques described in A Simplified Approach for Estimating Secondary Production of Hazardous Air Pollutants (HAPs) Using the
OZIPR Model, EPA-454/R-99-054.
17.
KEY WORDS AND DOCUMENT ANALYSIS
DESCRIPTORS
b. IDENTIFIERS/OPEN ENDED TERMS
c. COSATI Field/Group
Air Pollution, Air Quality Dispersion Models,
Meteorology, Air Toxics, Urban Area Modeling
18. DISTRIBUTION STATEMENT
Release Unlimited
19. SECURITY CLASS (Report)
Unclassified
21. NO. OF PAGES
109
20. SECURITY CLASS (Page)
Unclassified
22. PRICE
EPA Form 2220-1 (Rev. 4-77) PREVIOUS EDITION IS OBSOLETE
------- |