United States Office of Air Quality EPA-450/4-79-018
Environmental Protection Planning and Standards September 1979
Agency Research Triangle Park NC 27711
Air
&EPA Procedures for the
Preparation of Emission
Inventories for
Volatile Organic
Compounds
Volume II:
Emission Inventory
Requirements for
Photochemical Air Quality
Simulation Models
-------
EPA-450/4-79-018
Procedures for the Preparation of Emission
Inventories for
Volatile Organic Compounds
Volume II:
Emission Inventory Requirements for
Photochemical Air Quality Simulation Models
EPA Project Officer: Thomas F. Lahre
U.S. ENVIRONMENTAL PROTECTION AGENCY
Office of Air, Noise, and Radiation
Office of Air Quality Planning and Standards
Research Triangle Park, North Carolina 27711
September 1979
-------
This report is issued by the Environmental Protection Agency to report
technical data of interest to a limited number of readers. Copies are
available free of charge to Federal employees, current contractors and
grantees, and nonprofit organizations - in limited quantities - from the
Library Services Office (MD-35), U.S. Environmental Protection Agency,
Research Triangle Park, North Carolina 27711; or for a nominal fee,
from the National Technical Information Service, 5285 Port Royal Road,
Springfield,Virginia 22161.
This document has been reviewed by the Office of Air Quality Planning
and Standards, U.S. Environmental Protection Agency, and approved
for publication. Subject to clarification, the contents reflect current
Agency thinking.
Publication No. EPA-U50/4-79-018
M
-------
ACKNOWLEDGMENTS
This document was prepared by Tom Lahre of the Monitoring and
Data Analysis Division, OAQPS, EPA in conjunction with personnel
from Pacific Environmental Services (PES) and PEDCo Environmental.
Dr. Lowell Wayne of PES was primarly responsible for much of the
material in Chapters 6, 7, and the appendices, and provided inval-
uable review comments on the rest of the document. Mr. Keith
Rosbury of PEDCo assembled much of the material in Chapter 5.
The authors wish to thank the numerous persons involved
in the planning and review of this document as well as those
groups that pioneered many of the inventory concepts developed
herein.
-------
-------
TABLE OF CONTENTS
Section Page
1.0 INTRODUCTION 1-1
1.1 Purpose 1-1
1.2 Background 1-3
1.3 Contents of Volume II 1-6
2.0 INVENTORY PLANNING AND DESIGN CONSIDERATIONS 2-1
2.1 Determination of the Grid System 2-1
2.2 Data Collection 2-4
2.2.1 Existing Emission Inventories 2-4
2.2.2 Planning the Data Collection Effort 2-6
2.2.3 Inventories of Pollutants Other Than
VOC and NOX 2-8
2.2.4 Elevated Point Source Requirements 2-8
2.3 Spatial Resolution of Emissions 2-9
2.4 Temporal Resolution of Emissions 2-11
2.5 VOC (and NOx) Species Allocation 2-12
2.6 Emission Projections 2-14
2.7 Data Handling 2-17
2.8 Manpower Requirements 2-19
2.9 Overview of Emission Inventory Procedures 2-20
3.0 DETERMINATION OF THE GRID SYSTEM 3-1
3.1 Choosing An Appropriate Grid System 3-1
3.1.1 Area Covered by the Grid System 3-1
3.1.2 Grid Cell Size 3-4
3.2 Map Gridding Procedures 3-6
3.2.1 UTM Coordinate System 3-6
3.2.2 Orienting the Grid System 3-6
3.2.3 Problems in Gridding 3-7
4.0 POINT SOURCE EMISSIONS 4-1
4.1 Data Collection 4-1
4.2 Spatial Resolution 4-3
4.3 Temporal Resolution 4-4
4.4 Point Source Projections 4-6
-------
Section Page
4.4.1 Individual Facility Projections 4-7
4.4.2 Aggregate Point Source Projections 4-10
4.4.3 Point-Source Projection Review and
Documentation 4-14
5.0 HIGHWAY VEHICLES 5-1
5.1 Introduction 5-1
5.2 Nature of Automotive Emissions 5-1
5.3 Overview of Highway Vehicle
Inventory Procedures 5-3
5.4 Transportation Planning Data 5-5
5.5 Emission Factor Models 5-11
5.5.1 MOBILE1 5-12
5.5.2 Modal Model 5-23
5.5.3 Computing S0£ and Particulate
Emissions 5-23
5.6 Network Emission Models 5-24
5.6.1 Link Based Network Emissions Models 5-24
5.6.2 Hybrid Models 5-31
5.7 Highway Vehicle Emission Gridding Procedure 5-34
5.7.1 Gridding Procedures for Link-Based Models.... 5-34
5.7.2 Gridding Procedures for Hybrid Models 5-36
5.8 Transportation Control Measure Analysis 5-36
6.0 AREA SOURCES 6-1
6.1 General 6-1
6.2 General Methodology for Spatial Resolution 6-3
6.2.1 Direct Grid Cell Level Determination
of Emissions 6-3
6.2.2 Surrogate Indicator Approach 6-4
6.3 General Methodology for Temporal Resolution 6-20
6.4 Area Source Projection Procedures 6-23
6.5 Specific Area Source Data 6-27
6.6 Aircraft 6-41
6.6.1 Data Collection 6-41
6.6.2 Spatial Resolution 6-42
v1
-------
Section Page
6.6.3 Temporal Resolution 6-42
6.6.4 Projections 6-42
6.7 Other Off-Highway Fuel Usage 6-43
6.7.1 Data Collection 6-43
6.7.2 Spatial Resolution 6-43
6.7.3 Temporal Resolution 6-43
6.7.4 Projections 6-44
6.8 Railroads 6-44
6.8.1 Spatial Resolution 6-44
6.8.2 Temporal Resolution 6-44
6.8.3 Projections 6-44
6.9 Vessels 6-45
6.9.1 Spatial Resolution 6-45
6.9.2 Temporal Resolution 6-45
6.9.3 Projections 6-45
6.10 Gasoline Service Stations 6-46
6.10.1 Spatial Resolution 6-46
6.10.2 Temporal Resolution 6-46
6.10.3 Projections 6-46
6.11 Drycleaning 6-48
6.11.1 Spatial Resolution 6-48
6.11.2 Temporal Resolution 6-48
6.11.3 Projections 6-48
6.12 Degreasing 6-49
6.12.1 Spatial Resolution 6-49
6.12.2 Temporal Resolution 6-49
6.12.3 Projections 6-49
6.13 Nonindustrial Surface Coating 6-50
6.13.1 Spatial Resolution 6-50
6.13.2 Temporal Resolution 6-50
6.13.3 Projections 6-50
-------
Section Page
6.14 Cutback Asphalt 6-51
6.14.1 Spatial Resolution 6-51
6.14.2 Temporal Resolution 6-51
6.14.3 Projections 6-51
6.15 Pesticides Applications 6-51
6.15.1 Spatial Resolution 6-51
6.15.2 Temporal Resolution 6-52
6.15.3 Projections 6-52
6.16 Fuel Combustion 6-52
6.16.1 Spatial Resolution 6-52
6.16.2 Temporal Resolution 6-52
6.16.3 Projections 6-53
6.17 Other Miscellaneous Area Sources 6-53
6.17.1 Data Collection 6-53
6.17.2 Spatial Resolution 6-53
6.17.3 Temporal Resolution 6-54
6.17.4 Projections 6-54
7.0 VOC (AND NOX) ALLOCATION INTO CHEMICAL CLASSES 7-1
7.1 Allocation of VOC to Classes 7-1
7.2 Specification of NOX as NO and N02 7-6
7.3 Projecting VOC (and NOX) Split Factors 7-8
7.4 Compatibility with Inventory Data and Source
Categories 7-9
8.0 DATA HANDLING 8-1
8.1 General 8-1
8.2 Point Source Data Handling 8-2
8.3 Area Source Data Handling 8-7
8.4 Highway Vehicle Data Handling 8-10
8.5 Modeler's Tape Creation 8-12
APPENIDX A - URBAN SCALE PHOTOCHEMICAL AIR QUALITY SIMU-
LATION MODELS ,.. A-l
APPENDIX B - ACRONYMS AND ABBREVIATIONS, GLOSSARY OF TERMS. B-l
-------
LIST OF FIGURES
Figure
2-1. Overview of Emission Inventory Procedures
3-1. Tulsa Inventory Area
4-1. National Emissions Data System Point Source
Li sti ng
5-1. Overview of Processes for Compiling Detailed..
Highway Emission Inventories
6-1. Segment of Land Use Map for Tampa Bay, Florida
8-1. Example of File Structure for Elevated Point
Sources in Modeler's Tape
8-2. Example of File Structure for Ground Level
Emissions in Modeler's Tape
LIST OF TABLES
Table
4-1. Projected Employment by Sector in Hillsborough
County 4-12
4-2. Selected OBERS Industrial Categories 4-13
5-1. Link Data for PLANPAC/BAGKPAC Urban
Transportation Planning Batterv 5-7
5-2. Effect of some MOBILE1 Variables on 1978
Emission Factors 5-15
5-3. Assumed Percentages of Cold and Hot Transient
Vehicle Operation in APRAC-2 and HWYEMIS1 5-19
5-4. Hourly Distributions of Average Daily Traffic
for Large Cities as a Percentage of the Total. 5-28
5-5. Hourly Distributions of Directional Split for
Large Cities as a Percentage of Hourly Volume. 5-29
5-6. Example Transportation Control Measures 5-37
6-1. Land Use Categories for Figure 6-1 6-8
6-2. Demographic Parameters Used in San Francisco
Bay Area for Making Zonal Allocations of Area
Sources 6-13
6-3. Excerpt From ABAG Cross-Classification Table
Used in San Francisco Bay Area for Subcounty
Allocation of Area Source Activities 6-14
6-4. Illustrative Excerpts From Zone-to-Grid-Cell
Correspondence Table for Determining
Apportioning Factors 6-18
6-5. Diurnal Patterns for Gasoline Stations in
Tampa Bay, in Percent of Daily Operation 6-22
6-6. Example Methodologies for Spatial and Temporal
Resolution and Projection of Countywide
Emission Totals for Area Source Categories 6-28
-------
Table Page
7-1. Estimated Composition of Exhaust Gases From
Highway and Off Highway Vehicles Using
Gasoline Fuel/Composite of All Grades 7-3
7-2. Example VOC Allocation Factors by Source
Category for a 3-Class Scheme 7-5
8-1. Example Temporal Factor File for Point
Sources 8-4
8-2. Example "Split Factor" File 8-6
8-3. Example File of Grid Cell Apportioning
Factors for Area Sources 8-8
-------
1.0 INTRODUCTION
1.1 PURPOSE
This document is a companion volume to Procedures for the
Preparation of Emission Inventories for Volatile Organic Compounds,
Volume I.1 Volume I describes procedures for compiling annual
inventories of volatile organic compound (VOC) emissions for rela-
tively large areas, such as counties. The less detailed emission
inventory that results from the procedures in Volume I is primarily
intended to be used as input to simple source/receptor relation-
ships, such as the Empirical Kinetic Modeling Approach (EKMA)^.
Volume II describes procedures for compiling VOC emission in-
ventories in sufficient detail for use in photochemical air quality
simulation models. Because photochemical models can simulate the
hour-by-hour photochemistry occurring over numerous, small sub-
county areas (i.e., grid cells), more temporal and spatial resolu-
tion of input data must be included in the inventories that provide
emission data to these models. Total VOC (and NOX) emissions
must be apportioned into species classes, and information may be
required on other pollutants such as carbon monoxide, sulfur
dioxide, and particulates. Further, in applying photochemical air
quality simulation models to the evaluation of proposed control
strategies, it is necessary to provide "projection inventories,"
i.e., anticipated emission inventories for future years, at the
same level of detail as required for current inventories. Methodol-
ogies for providing this added detail are presented in this docu-
ment. In each case, the requirements for projection inventories
are equivalent to those for current or "base year" inventories.
1-1
-------
The basic emission inventory requirements are in many respects
quite similar for photochemical models and the less data-intensive
source/receptor relationships. For both, much of the same informa-
tion is obtained from the same sources. Moreover, the resulting
inventories are used by air pollution control agencies for the same
general purpose, i.e., for developing control strategies that will
assure the achievement and maintenance of the National Ambient Air
Quality Standard for ozone.3 Hence, there are many activities
(e.g., data collection, emission calculations) to which the same
considerations and techniques apply, regardless of whether the
inventory is being developed for a photochemical model or a simple
source/receptor relationship. In general, where procedures are
similar to those already described in detail in Volume I, they are
not repeated here. Thus, it is important that the reader be famil-
iar with the content of Volume I to thoroughly understand the pro-
cedures described in this document.
Several different photochemical simulation models exist or are
being developed whose specific emission inventory input require-
ments differ in some respects. Volume II does not describe the
exact detailed specifications and formats of the emission inventor-
ies needed as input to each model. This would be extremely diffi-
cult and tedious because many versions of these models are current-
ly evolving. Nor is it the intent of Volume II to set forth cri-
teria to help the user select an appropriate photochemical model.
Instead, this document summarizes the steps that must be carried
out to provide inventories that are generally suitable for use in
any photochemical model. To complement this general discussion, a
brief description of the emission requirements of currently avail-
able photochemical models is presented in Appendix A. In each
case, the requirements for projection inventories are equivalent to
those for current or "base year" inventories.
1-2
-------
It is assumed at the outset that an annual, countywide emis-
sion inventory is available as a starting point for the photochemi-
cal modeling effort. This basic type of inventory is useful both
in planning the more detailed emission inventory effort and as a
source of certain data. For most urban areas, some sort of basic
emission inventory has already been performed.
1.2 BACKGROUND
As described in Volume I, the emission inventory is an essen-
tial element in an ozone control program. It tells the agency what
sources are present in an area and how much of each ozone precursor
pollutant is emitted by each source. The emission inventory is
ultimately utilized in conjunction with a source/receptor relation-
ship of some kind for the development of an ozone control strategy.
Two basic approaches may be used in a control program for
relating photochemical oxidant (expressed as ozone) to VOC and
NOX. The first involves the use of empirical relationships such
as EKMA or rollback to relate ambient ozone with precursor emissions
over fairly broad geographical areas. These simple models provide
answers to questions such as "what level of overall volatile organic
compound emission control is needed to attain the ozone standard in
an urban area?" or "what reduction in maximum ozone concentration
will accompany a specified reduction in ambient levels of volatile
organic compounds?" For these simple source/receptor relation-
ships, it is generally sufficient to use annual or seasonally ad-
justed estimates of total reactive VOC emissions for the entire
area of concern. This type of emission estimate can be obtained
using the procedures outlined in Volume I.
The second basic approach for relating ozone to precursors
involves the use of photochemical air quality simulation models.
1-3
-------
These models offer a more theoretically sound approach for control
strategy development in that they attempt to simulate the photo-
chemical reactions that occur over an urban region during each hour
of a typical day for which a model is being applied. Because of
their ability to provide detailed spatial and temporal information
on both ozone levels and precursor pollutants and because they can
directly relate emissions to ozone concentrations, photochemical
simulation models offer considerable potential for use in control
strategy design and evaluation.
In addition to answering the limited questions that the simp-
ler empirical source/receptor relationships may address, photochem-
ical models enable strategists to make more sophisticated determin-
ations relating to control program development. For example,
photochemical models enable control agencies to judge whether it is
more effective to control only certain precursor sources within an
urban area rather than all sources, or where (and when) benefits
from various control options are most likely to occur within an
urban area.' Another application is the development of environ-
mental impact assessments, since photochemical models allow an
agency to evaluate the impact of new precursor sources (e.g.,, a
major highway) at various receptor locations. Photochemical models
are also useful in basic scientific research, such as in validation
studies of atmospheric photochemistry and dispersion mechanisms.
There are two major types of photochemical models: trajectory
and grid. Trajectory models (also known as Lagrangian models)
estimate the concentration of ozone and other chemical species
within a specified parcel of air. These models "follow" a specific
hypothetical air parcel and estimate pollutant concentrations with-
in the selected parcel during subsequent hours. The concentrations
that are estimated result from an interaction of chemical reactions
1-4
-------
among precursor pollutants within the parcel and emissions encoun-
tered along the parcel's trajectory. Meteorological variables also
play key roles in determining concentrations of ozone and other
pollutants in the selected parcel. The trajectory of the parcel
(and hence the emissions encountered), the vertical and horizontal
exchange with surrounding air, and some key chemical reaction rates
are all determined by prevailing meteorological conditions.
Grid models (also called Eulerian models) calculate pollutant
concentrations at fixed locations in space at specified times. The
concentrations estimated at each location result from interaction
among emissions, chemical reactions, and transport and dilution
introduced by prevailing meteorological conditions. Pollutant con-
centrations are calculated for each cubicle of a three-dimensional
framework in the entire region of interest. A cubicle might have
horizontal dimensions of 1 to 10 kilometers on a side and be 50 to
500 meters deep. Some Eulerian and Lagrangian models are designed
to provide vertical (as well as horizontal) resolution of pollutant
concentrations by using a vertical "stack" of cubicles.
In order for photochemical simulation models to produce the
temporal and spatial resolution of which they are capable, the
inventories that supply emissions data to these models must be con-
siderably more detailed than inventories based on the procedures
prescribed in Volume I. Basically, the emission inventory used in
a photochemical modeling application must provide more detail in
three ways. First, instead of providing a single emission estimate
for each precursor pollutant for the entire area of concern (e.g.,
one county), emission estimates must be provided for each individ-
ual cell of a grid system within the area. Second, instead of
annual or seasonally adjusted anissions, typical hour-by-hour
1-5
-------
emission estimates must be provided. Third, instead of total reac-
tive VOC, specific information must be provided on quantities of
several classes of VOC. Emission information for NO and N0£ is
also required. In addition, some models require spatially and tem-
porally resolved emission estimates of CO, S02» and particulates
and, if the model provides for vertical resolution of pollutants,
knowledge of stack and exhaust gas parameters for each large point
source may be necessary.
The basic emission inventory requirements for trajectory pho-
tochemical models are similar to those for their grid model counter-
parts. The main difference is that for a single trajectory, emis-
sions are needed only along the trajectory. However, if a large
number of trajectories must be run to study control strategies over
an entire region, the emission inventory requirements for a trajec-
tory model become essentially identical to those for a grid model.
1.3 CONTENTS OF VOLUME II
This document offers technical assistance to those engaged in
planning and development of detailed emission inventories for use in
photochemical air quality simulation models.
Chapter 1 discusses the purpose of Volume II and its relation-
ship to Volume I; it also includes an introductory description of
the two types of photochemical air quality simulation models and
their emission inventory requirements. Chapter 2 describes various
technical considerations that aid in the planning and design of the
detailed emission inventory, and presents a brief overview of the
entire emission inventory process. Chapter 2 is intended to pro-
vide an overall perspective of the detailed inventory requirements
for those who will actually be utilizing the remainder of the docu-
ment, while at the same time serving as an executive summary to
1-6
-------
those concerned only at a general level with the inventory process.
Chapters 3 through 8 provide detailed "how to" procedures for
supplying the detail that is necessary to compile an inventory for
use in a photochemical model.
In addition, Appendix A briefly describes emission input
requirements for available versions of photochemical models, while
Appendix B contains a glossary of technical terms used in this
document.
1-7
-------
References for Chapter 1.0
1. Procedures for the Preparation of Emission Inventories for
Volatile Organic Compounds. Volume I, EPA-450/2-77-028, U.S.
Environmental Protection Agency (OAWM, OAQPS, MDAD), Research
Triangle Park, NC, December 1977.
2. Uses, Limitations and Technical Basis of Procedures for
Quantifying Relationships Between Photochemical Qxidants and
Precursors, EPA-450/2-77-021a, U.S. Environmental Protection
Agency (OAWM, OAQPS, MDAD), Research Triangle Park, NC, November
1977.
3. 40 CFR 50.9, "National Primary and Secondary Ambient Air Quality
Standards for Ozone", 1978.
1-8
-------
2.0 INVENTORY PLANNING AND DESIGN CONSIDERATIONS
The same basic steps (viz., planning; data collection, hand-
ling, and analysis) are necessary to compile both a detailed emis-
sion inventory for photochemical modeling and a less detailed
inventory for EKMA or rollback. Because the same basic steps are
involved, many of the planning and design considerations discussed
in Chapter 2 of Volume I are generally applicable. However, since
the requirements of photochemical modeling inventories are more
rigorous than those of the simpler source/receptor relationships, a
number of additional planning considerations must be taken into
account. The primary purpose of this chapter is to highlight the
additional inventory requirements of photochemical models and to
discuss the added responsibilities imposed on the agency developing
an emission inventory for use in a photochemical modeling effort.
2.1 DETERMINATION OF THE GRID SYSTEM
A primary planning consideration is the selection of a grid
system compatible with the proposed modeling system. The grid sys-
tem is defined by two main elements: (1) the grid boundary, outlin-
ing the area to be modeled, and (2) the collection of subareas
(called "cells") within which emissions are to be determined for use
in the model. For most models, the grid boundary must be rectangu-
lar and the cells must be squares, all of equal size. (The "grid
spacing" is the length of the side of one square grid cell.)
Thus, at the outset of the inventory effort, the agency must
define both the overall size and shape of the grid to be modeled and
the size and number of grid cells that compose the grid. It is
extremely important to develop an appropriate grid system at the
start so that emission estimates for various types of sources,
including area sources, are referenced to a single grid system. A
2-1
-------
great deal of wasted effort results if a common grid system is not
adopted or if the grid system is redefined during the course of the
compilation effort. In addition, since a number of factors not re-
lated to the inventory (e.g., meteorology) must be taken into ac-
count when defining the system, those responsible for the inventory
should consult the appropriate modeling specialists, planners, and
meteorologists before collecting data, to ensure that the grid sys-
tem meets the general objectives of the photochemical modeling
program.
The area covered by the inventory grid should be as large as
practical, for several reasons. It is generally desirable to in-
clude all major precursor emission sources within the grid. In
cases where the exact area to be modeled is not certain, a large
emission grid ensures that a smaller area chosen within the emission
grid can be modeled adequately. In this sense, the emission grid
can be larger than the actual grid used for modeling.
The grid should be large enough to encompass as many ozone and
precursor pollutant monitoring stations as possible, since an impor-
tant step in the application of the photochemical dispersion model
involves "validating" the model, i.e., determining whether the
observed ozone and precursor pollutant concentrations at each moni-
toring station agree with values predicted by the model. Ambient
air measurements are also necessary to define boundary pollutant
concentrations in the validation process.
The grid should also be large enough to encompass areas of cur-
rent limited land use activity that are expected to develop as a
result of projected growth. Since an important application of pho-
tochemical models is to evaluate ozone concentrations in future
years, land use projections should be consulted to determine what
types of growth are expected and in what areas.
2-2
-------
2.0 INVENTORY PLANNING AND DESIGN CONSIDERATIONS
The same basic steps (viz., planning; data collection, hand-
ling, and analysis) are necessary to compile both a detailed emis-
sion inventory for photochemical modeling and a less detailed
inventory for EKMA or rollback. Because the same basic steps are
involved, many of the planning and design considerations discussed
in Chapter 2 of Volume I are generally applicable. However, since
the requirements of photochemical modeling inventories are more
rigorous than those of the simpler source/receptor relationships, a
number of additional planning considerations must be taken into
account. The primary purpose of this chapter is to highlight the
additional inventory requirements of photochemical models and to
discuss the added responsibilities imposed on the agency developing
an emission inventory for use in a photochemical modeling effort.
2.1 DETERMINATION OF THE GRID SYSTEM
A primary planning consideration is the selection of a grid
system compatible with the proposed modeling system. The grid sys-
tem is defined by two main elements: (1) the grid boundary, outlin-
ing the area to be modeled, and (2) the collection of subareas
(called "cells") within which emissions are to be determined for use
in the model. For most models, the grid boundary must be rectangu-
lar and the cells must be squares, all of equal size. (The "grid
spacing" is the length of the side of one square grid cell.)
Thus, at the outset of the inventory effort, the agency must
define both the overall size and shape of the grid to be modeled and
the size and number of grid cells that compose the grid. It is
extremely important to develop an appropriate grid system at the
start so that emission estimates for various types of sources,
including area sources, are referenced to a single grid system. A
2-1
-------
great deal of wasted effort results if a common grid system is not
adopted or if the grid system is redefined during the course of the
compilation effort. In addition, since a number of factors not re-
lated to the inventory (e.g., meteorology) must be taken into ac-
count when defining the system, those responsible for the inventory
should consult the appropriate modeling specialists, planners, and
meteorologists before collecting data, to ensure that the grid sys-
tem meets the general objectives of the photochemical modeling
program.
The area covered by the inventory grid should be as large as
practical, for several reasons. It is generally desirable to in-
clude all major precursor emission sources within the grid. In
cases where the exact area to be modeled is not certain, a large
emission grid ensures that a smaller area chosen within the emission
grid can be modeled adequately. In this sense, the emission grid
can be larger than the actual grid used for modeling.
The grid should be large enough to encompass as many ozone and
precursor pollutant monitoring stations as possible, since an impor-
tant step in the application of the photochemical dispersion model
involves "validating" the model, i.e., determining whether the
observed ozone and precursor pollutant concentrations at each moni-
toring station agree with values predicted by the model. Ambient
air measurements are also necessary to define boundary pollutant
concentrations in the validation process.
The grid should also be large enough to encompass areas of cur-
rent limited land use activity that are expected to develop as a
result of projected growth. Since an important application of pho-
tochemical models is to evaluate ozone concentrations in future
years, land use projections should be consulted to determine what
types of growth are expected and in what areas.
2-2
-------
Finally, the grid should be large enough to encompass the ef-
fects of meteorology in the model area. Since peak oxidant levels
often occur downwind of the urban center, it is desirable to in-
clude as much "downwind area" as possible in the grid so that the
model can predict where and when these peak levels will occur. A
large grid area also minimizes the possibility that receptor loca-
tions will be impacted by air parcels that have left the grid and
then re-entered it due to a shift in wind during the time interval
over which the model is run.
Too large a grid size, however, may also cause certain prob-
lems in gathering and manipulating data. For instance, a grid may
extend into areas for which detailed spatially and temporally re-
solved emission estimates cannot be made due to lack of adequate
information. This might be the case where one part of a grid lies
within the jurisdiction of a metropolitan planning organization
(MPO) and another part lies within an outlying, undeveloped juris-
diction. Detailed records and projections will probably be avail-
able for the metropolitan areas, but may not exist for the outlying
area. Technical problems may also be encountered if various juris-
dictions within a grid maintain information in different formats.
For example, one area may maintain records for townships and use
EPA's Emission Inventory System (EIS/P&R), whereas another area may
maintain records for census tracts and use a locally developed data
handling system that is incompatible with EIS/P&R.
While the grid area should be large in order to maximize cover-
age, individual grid cells should be small in order to optimize the
spatial resolution of emissions. If the grid cells are too large,
the model may lose precision in estimating ozone and precursor pol-
lutant levels at receptor locations of interest. In addition,
smaller grid cells provide better resolution of the impact of
individual sources.
2-3
-------
Of course, practical considerations limit how small each grid
cell can be, just as they limit how large the overall grid can be.
Resources are an important factor to consider. As the size of the
grid or the number of grid cells increases, additional effort is
required to collect the necessary emission data and to develop
appropriate factors for allocating area source emissions. Also,
computer costs increase significantly because of the increased core
storage requirements associated with an increased number of grid
cells.
In most model applications, a compromise is reached concerning
the amount of area covered by the grid and the number of grid cells
defined within the grid. A typical grid size in a number of urban
areas where photochemical models have been run is about 50 to 150
km on each side, with grid spacing typically being about 2 to 5 km.
Details concerning the development of the grid system are presented
in Chapter 3.
2.2 DATA COLLECTION
When the grid system has been defined, the next important step
in the compilation of a photochemical modeling emission inventory is
the collection of appropriate emission data. Data are normally
acquired separately for point sources, highway motor vehicles, and
other area sources. It is assumed that a conventional annual,
countywide emission inventory, as described in Volume I, already
exists, and that additional data must be collected as a basis for
assigning emissions to grid cells, for determining temporal distri-
butions, and for estimating the proportions of VOC and NOX to be
assigned to species or classes required in the model.
2.2.1 EXISTING EMISSION INVENTORIES
Because many of the data requirements of the detailed emission
inventory are quite resource-intensive, the inventorying agency
2-4
-------
should make use of whatever data and systems are already available.
Thus, the agency should review the existing inventories, data hand-
ling systems, and planning models maintained by local agencies in
order to determine what framework has already been established that
may possibly be utilized in compiling the detailed emission
inventory.
The existing emission inventory should first be reviewed to
determine what source and emissions data are already available.
Most urban locations have VOC and NOX inventories at the level of
detail of EPA's National Emission Data System (NEDS) or Emission
Inventory System (EIS/P&R). If accurate, comprehensive and current,
such inventories can provide much of the basic data needed for the
detailed inventory. If not, the basic inventory should be updated
prior to or during the initial stages of compiling the detailed
inventory to be used for photochemical modeling.
The agency should also determine whether the existing inven-
tory represents average annual emissions or seasonal emissions dur-
ing the oxidant season. The seasonally adjusted inventory, if it
exists, is more easily used as a starting point for the detailed
modeling inventory. (Seasonal adjustment of the annual inventory is
discussed in Volume I.)
The modeling region normally encompasses a number of counties;
in rare cases, the entire modeling area is contained in one large
county. Because most counties do not have rectilinear boundaries,
portions of a county may extend beyond the boundaries of the model-
ing region. In this case, the existing countywide emission inven-
tories must be adjusted to include only those emissions in the
model ing region.
Local emission inventories usually incorporate point source
records that contain most of the data (including stack parameters
and operating information) needed for photochemical modeling. The
2-5
-------
only necessary point source data that are not generally included in
such inventories are those dealing with VOC (and NOX) species or
classes and detailed hour-by-hour emission information. Likewise,
most of the area-source countywide activity levels in local inven-
tories can be used as the basis for apportioning area-source emis-
sions to the grid cell level. Both temporal and spatial allocation
factors are needed to apportion countywide area source emissions to
the grid cell level.
The only important source category for which the basic county-
wide inventory does not ordinarily represent a good starting point
is highway motor vehicles. In annual countywide inventories, high-
way motor vehicle emissions are often based either on countywide
gasoline sales or on total vehicle miles traveled (VMT). These
gross techniques do not provide the necessary spatial resolution
for photochemical modeling, so it is desirable to obtain link-by-
link traffic data for the area from a transportation planning
agency.
2.2.2 PLANNING THE DATA COLLECTION EFFORT
The existing inventories should contain much of the required
information on total emissions for the area of interest. The docu-
mentation provided in support of the countywide point source inven-
tory (e.g., NEDS data) usually contains additional information on
stack parameters. The data collection effort in support of the
modeling inventory should be directed toward providing additional
information to (1) define the spatial distribution and temporal
variations in emissions from each source category, and (2) assign
VOC and NOX emissions to appropriate species classes. Ideally,
emissions data (including VOC species) would be available for each
source for each hour of any day selected for modeling. However,
this degree of detail is, in practice, neither necessary nor
2-6
-------
practical for all sources, since it can require an inordinate amount
of effort to secure such data, and since for many sources it would
have little effect on the oxidant levels predicted by the
photochemical model.
An important part of the data collection plan consists of
deciding which apportioning information should be collected and
which is unimportant (i.e., setting priorities). In a typical
situation, highway motor vehicles, gasoline marketing and storage,
solvent consumption, and power plants may account for much of the
total VOC and NOX emissions. The remaining VOC and NOX
emissions probably arise from a number of smaller sources, any of
which individually has only a minor influence on predicted oxidant
levels, even if small errors are made in allocating the emissions to
the proper grid cells or in estimating their seasonal or diurnal
variations. The existing basic inventory can tell the agency which
sources are the most important emitters and, as a result, which
should be surveyed or visited directly to obtain better spatially
and temporally resolved emission data. Many sources are such minor
VOC and NOX emitters that little, if any, additional effort is
warranted in gathering temporally and spatially resolved data
regarding them.
Finally, it is essential that the inventory agency work closely
with the metropolitan planning organization (MPO) or other planning
agencies in the area to determine what transportation and land use
planning models are being employed and what data they can provide
that are directly useful to the inventory compilation effort. In
general, in urban areas where comprehensive transportation and land
use planning are performed, much of the information needed to deter-
mine highway motor vehicle emissions, to make projections, and to
apportion emissions to the grid cell level will already be available.
2-7
-------
2.2.3 INVENTORIES OF POLLUTANTS OTHER THAN VOC AND NOY
A
The detailed inventory effort should be directed primarily
toward obtaining accurate emission data for VOC and NOX, because
both of these precursor pollutants are extremely important in the
photochemical production of ozone in urban atmospheres. Some photo-
chemical models have the capacity to accept emissions data arid gen-
erate air quality estimates for carbon monoxide (CO), sulfur oxides
(SOX), and total suspended participates (TSP), as well. This
capability is provided primarily to allow the user to study the
effects of control strategies on ambient levels of these pollutants.
Carbon monoxide, in addition, can serve as a nonreactive tracer
material to test the reasonableness both of the mobile source inven-
tory and of the basic dispersion mechanics of the model. However,
emissions of CO, SOX, and TSP have little effect on the ambient
oxidant levels predicted by the photochemical model. Thus, although
sources of these pollutants should be included in the photochemical
modeling inventory, a great deal of expense is not warranted in
developing spatially and temporally resolved emission estimates and
projections for these sources if NOX and VOC are not also emitted
in significant quantities.
2.2.4 ELEVATED POINT SOURCE REQUIREMENTS
Some photochemical models assign emissions from point sources
to elevated grid cells if they are characterized by a significant
effective stack height. When this is the case, the emission inven-
tory must include stack information (e.g., physical stack height
and diameter, stack gas velocity, and temperature) for the major
point sources in the area. The agency must therefore k.iow whether
the model it plans to employ requires these data. Certain stack
information is usually included in the less detailed point-source
emission inventories available in most urban areas. This informa-
tion should, of course, be examined and utilized to the greatest
2-8
-------
extent possible in order to minimize additional costs to the detail-
ed inventory effort.
2.3 SPATIAL RESOLUTION OF EMISSIONS
In order for photochemical models to provide spatially resolv-
ed predictions of ozone and various other pollutants at the grid
cell level, they must be supplied with emission data that have the
same degree of spatial resolution (i.e., that are resolved to the
grid cell level.) The amount of effort required to resolve emis-
sions at the grid cell level will vary depending on the type of
source. Point source locations are typically known to within a
fraction of a kilometer in existing, annual inventories; hence, it
is easy to assign them to the appropriate grid cell. This assign-
ment can either be made manually by overlaying an outline of the
grid system onto a map showing point source locations, or a comput-
er routine can be written to do it.
Area sources, on the other hand, require substantially more
effort. There are two basic ways to determine area source emis-
sions at the grid cell level. The most accurate (and most resource
intensive) approach is to obtain area source activity level infor-
mation directly for each grid cell. Airports are an example of a
source category for which this may typically be done. The alterna-
tive approach, most commonly employed, is to apportion the county-
wide emissions from the annual inventory to the grid cell level by
means of representative apportioning factors. This latter approach
requires that surrogate indicators of emission levels (e.g.,
population, type of land use) be known or estimated for each grid
cell and applied to the countywide emissions to yield grid cell
emissions. The major assumption underlying the apportioning
approach is that emissions from each area source behave spatially in
the same manner as the surrogate indicator. For those area sources
whose emissions must be apportioned from countywide totals, emphasis
2-9
-------
should be placed on developing accurate factors for the largest
emitters. In most large urban areas, local planning agencies can
provide the air pollution agency with detailed land use, population,
or in some cases, employment statistics at the subcounty level.
These data are commonly used for apportioning most of the area
source emissions in the detailed photochemical modeling inventory.
Highway motor vehicle emissions, which are usually a large
fraction of total VOC and NOX emissions, are considered separately
from other area sources in inventories used for photochemical model-
ing. Instead of using countywide VMT or gasoline sales to estimate
highway vehicle emissions (as annual inventories typically do),
urban transportation planning models are employed to generate VMT on
an individual link basis. The emissions for each link are then
assigned to the appropriate grid cells.
Planning, land use, and transportation models are already in
use in many urban areas, and can provide the inventorying agency
with much of the data necessary to allocate area source emissions
and develop link-by-link emissions for highway motor vehicle
sources. Moreover, such models are generally capable of developing
forecasts for projection years. Thus, it is extremely important
that local agencies (particularly MPO's) be contacted during the
inventory planning process to determine what planning models are
being utilized and how the data available from these models can be
used in the emission inventory effort. Obviously, much redundant
effort would be required on the part of the inventorying agency if
it independently attempted to develop all the necessary information
that should be available from the MPO. Moreover, any subsequent
photochemical modeling results would likely be subject to challenge
because of alleged nonconformity with other projections available
to the public.
2-10
-------
2.4 TEMPORAL RESOLUTION OF EMISSIONS
In order for photochemical simulation models to predict hourly
concentrations of ozone and other pollutants, hour-by-hour esti-
mates of emissions at the grid-cell level are needed. Several
approaches can be used for providing the temporal detail needed in
the inventory used in a photochemical model. The most accurate and
exacting approach is to determine the emissions (or activity
levels) for specific sources for each hour of a typical day in the
time period being modeled. This approach is sometimes applicable
to point sources but is often impractical.
An alternative approach is to develop typical hourly patterns
of activity levels for each source category, and then apply these
to the annual or seasonally adjusted emissions to estimate hourly
emissions. This approach is commonly employed for area sources,
including highway motor vehicles, and is usually used for all but
the largest point sources.
Usually, the photochemical air quality model, and therefore
the emission inventory, is applied to the season of the year in
which weather is most conducive to oxidant formation; for most
locations, this means the summer and near-summer months (i.e.,
April through October).
Similarly, emissions are usually chosen to represent the day
of the week on which polluting activities are at a maximum, normal-
ly a weekday. In some cases it can be useful, especially for vali-
dation studies, to simulate weekend conditions when automotive and
industrial emission levels are reduced. For this purpose a weekend
inventory will be needed, and additional temporal pattern informa-
tion will have to be obtained pertaining to weekend days. Gener-
ally, however, it is not recommended that the agency compile a
weekend inventory unless (1) significant reductions or changes in
emission patterns are expected, and (2) the same inventorying
2-11
-------
procedures can be used as for weekdays, so that any resulting
changes in predicted ambient ozone levels can be attributed to ac-
tual changes in precursor pollutant levels and patterns rather than
simply to changes in methodologies. The latter will not be possi-
ble in many areas for highway vehicles as transportation models are
based on information (e.g., travel pattern surveys) that is applic-
able only to weekday situations. If the agency plans to utilize
the model to estimate ambient concentrations of various pollutants
for time periods other than the oxidant season, additional seasonal
information may be required.
2.5 VOC (AND NOX) SPECIES ALLOCATION
Because photochemical models are intended to simulate actual
photochemistry, they utilize different chemistry for various types
of VOCs and require specific information as to the proportions of
these various types in the VOC emissions. For this reason, VOC
emission totals must be disaggregated into subtotals for various
VOC classes. In addition, NOX emissions may have to be distri-
buted into NO and N02« (Some models do not require an NOX
breakdown because they assume all NOX emissions to be NO.) Liter-
ally hundreds of individual chemical compounds typically compose
the total VOC emissions in an urban area. No photochemical model
considers each organic compound individually; rather, classes of
VOC are defined that behave similarly in photochemical reactions.
One classification scheme distinguishes four classes of compounds:
paraffins, olefins, aromatics, and aldehydes. Another distin-
guishes six classes and is based on the presence of certain types
of carbon bonds in each VOC molecule. Other classification schemes
employ different numbers and types of VOC classes. Typically,
three to six classes of VOC are distinguished by various photochem-
ical models. (See Appendix A for details on individual models.)
2-12
-------
The standard procedure for allocating VOC to classes is to
assume that VOC emissions from each type of source contain a cer-
tain percentage of each class of compound. This procedure is the
easiest of several procedures that can be followed for allocating
VOC emissions as the same VOC distribution is assumed to apply to
each facility within a given source category.
In some instances, source-specific VOC species data are known
for certain individual facilities (e.g., through source tests or
material composition considerations), and the agency may prefer to
use these in the detailed inventory rather than an assumed VOC spe-
cies distribution. Generally, however, most industries cannot pro-
vide reliable VOC (and NOX) species data or accurately apportion
their emissions into appropriate classes, in which cases generalized
VOC and NOX distributions must be assumed for various source cate-
gories. Generalized VOC and NOX "split factors" are available in
the literature for this purpose (as is described in Chapter 7.)
A potential problem when using split factors to apportion VOC
and NOX into classes is that the source classification scheme
(i.e., source category breakdown) employed in the inventory will
probably not be directly compatible with the available split factor
classification scheme in all cases. For example, the inventory may
not distinguish between automotive exhaust and evaporative emis-
sions, whereas VOC split factors are typically available for each
of these automotive emission components (and may be significantly
different in some classification schemes.) As another example,
many VOC classification schemes do not distinguish between differ-
ent types of fuel combustion and boilers, yet most inventories do.
Hence, it is important in the planning stages that the agency
examine the split factors available for use and compare the classi-
fication thereof with the source classification scheme of the basic
inventory. If serious inconsistencies are apparent for the more
2-13
-------
important VOC and NOX source categories, the agency may need to
consider modifying the classification scheme to minimize any error
that may result. The agency should be cautioned that if it chooses
to alter the inventory source classification scheme, the resources
required may be significant and should be carefully assessed prior
to instituting such a change.
2.6 EMISSION PROJECTIONS
Regardless of what ozone source/receptor relationship or model
is employed, projection inventories are needed in order for the
agency to determine whether a given area will achieve or exceed the
ozone standard in future years. There are basically two types of
projections: baseline projections and control strategy projections.
Baseline projections are estimates of future year emissions that
take into account both expected growth in an area and air pollution
control regulations that are in effect at the time the projections
are made.* Control strategy projections, on the other hand, are
estimates of future year emissions that also include the expected
impact of changed or additional control regulations.
In many respects, the detailed projection emission inventory
used in a photochemical model will be the same as the baseline
projection inventory of annual, countywide emissions compiled for
use in EKMA or rollback. That is, the same source categories and
pollutants will be emphasized; and the same emission factors,
activity levels, and control device data will be utilized. Hence,
just as in the base year case, the annual, countywide projection
inventory may serve as a good starting point for the projection
inventory used in a photochemical model. However, as has been
* Certain provisions in existing control regulations may take effect
only at some future date. Baseline projections include the
effects of these expected changes.
2-14
-------
discussed in the preceding sections, additional considerations are
required to incorporate the spatial and temporal resolution needed
by photochemical models and to apportion VOC (or NOX) into
classes. These considerations are highlighted below as they speci-
fically relate to projection inventories:
• Attention should be given to spatial changes in emissions
from the base year to projection years. Changes in point
source emissions due to growth or control measures should be
associated with specific locations within the modeling area,
i.e., at either new or existing facilities. In this regard,
it is especially critical to pinpoint the location of any
large point sources of VOC or NOX that will be coming on
line. Apportioning factors for apportioning area sources
should reflect future land use patterns, employment, popula-
tion, etc. Highway vehicle emission inventories should re-
flect changes in highway networks and changing driving pat-
terns.
• Attention should be given to changes in temporal emission
patterns. Any anticipated changes in hourly, daily, or sea-
sonal operating patterns between the base year and projection
years should be reflected in the projection inventories.
• To the extent that VOC (and NOX) split factors are expected
to change from the base year to any projection year, such
changes should also be incorporated.
Generally, of these three considerations, information will be most
readily available concerning changes in spatial distribution. This
is because the air pollution agency should know where large, new
point sources will be located (at least in near-term projections)
and because highway vehicle and area source emission patterns will
directly reflect changes in land use, employment, and transportation
data supplied by local planning agencies. Typically, most of the
temporal patterns and VOC (and NOX) split factors will not be
changed from the base year inventory either because no changes are
expected or because no data will be available to support any
changes. These considerations are discussed in more detail in suc-
ceeding chapters.
2-15
-------
Of course, in some instances, baseline projection inventories
of annual, countywide emissions for the particular years of interest
will not be available for use as starting points for the detailed,
photochemical modeling inventories. Or, in other cases, the projec-
tion inventories that do exist may not reflect all of the growth and
control scenarios that the agency may wish to evaluate with the
photochemical model. In these situations, the agency will have to
devote resources to the development of projection year inventories.
Specific recommendations for making baseline projections are also
discussed in the following chapters of this volume. However, the
following general considerations should be kept in mind from the
outset of the inventory planning stages.
1. To a large extent, projection inventories will be based on
forecasts of industrial growth, population, land use, and
transportation. The air pollution agency should jiot attempt
to make these forecasts itself, but rather should rely on
the local MPO or other planning agencies to supply them.
This course has several profound advantages. First, it
would be extremely costly for the air pollution agency to
duplicate the forecasts made by other planning agencies.
Second, the air pollution agency needs to base its emission
projections on the same forecasts that other governmental
planning agencies are utilizing. This consistency is
necessary to foster the credibility of any proposed control
programs that are based on these emission projections.
2. It is desirable to know what control strategies are being
considered in the modeling area in order to design control
strategy projection inventories to reflect these strategies.
This consideration may influence the type of data collected
as well as the structure of the inventory itself. As an
example, if the agency wants to test the effect of applying
Stage I controls only to service stations above a particular
size cutoff,, it may be desirable to treat these particular
stations as point sources rather than lumping them in with a
general service-station area-source category.
3. It is very important that all projected emissions for future
years be based on the same methodologies and computation
principles as the base year emissions. For example, if a
traffic model is used for estimating travel demand for the
2-16
-------
base year, the same traffic model should be applied to esti-
mate travel demand for projection years. Use of the same
methodology assures consistency between base year and pro-
jection year emission estimates, and prevents the possibil-
ity of spurious inventory differences based solely on
changes in methodology.
Projection inventories will always be open for attack be-
cause of their somewhat speculative nature. The technical
credibility of emissions projections will be a function of
their reasonableness, the amount of research and documenta-
tion of assumptions, and the procedures or methodologies
used to make the projections. Some degree of uncertainty
will always accompany emission projections. This fact
should be acknowledged openly. The art of projecting emis-
sion inventories is not eliminating uncertainty, but learn-
ing how to minimize uncertainty. Internal and external
review of emission inventory projections will improve their
technical quality and enhance their credibility.
2.7 DATA HANDLING
Due to the large amount of data that must be gathered and
stored in the inventory effort and to the complexity of developing
spatially and temporally resolved emission projections, a computer-
ized data handling system is usually a necessity in a photochemical
model ing effort.
Ideally, it is desirable to computerize as many data handling
functions as possible, because the agency can then devote more man-
power resources to collecting and analyzing the inventory data. The
computerization of each function requires that the agency develop a
flow chart of the entire data handling operation, from the initial
gathering of inventory data to the final development of a data file
(or "modeler's tape") that is in the input format of the photochem-
ical model being used.
Many of the data handling functions that are necessary in both
basic and detailed emission inventories (e.g., data storage) are
described in Volume I of this document. Several additional data
2-17
-------
handling requirements arise during the development of the detailed
inventory due to the need for spatial and temporal resolution of the
emission data and the need to disaggregate VOC and NOX emissions.
Spatial resolution is provided by allocating area source emis-
sions to grid cells using appropriate apportioning factors. This
step is usually computerized because of the large numbers of factors
and calculations involved.* Likewise, computerized programs are
generally used to assign point source and highway motor vehicle
emissions to the appropriate grid cells.
Temporal resolution is often provided for area and line sources
by developing typical hourly activity patterns for each major source
category. For many point sources, hourly activity levels can be
reasonably inferred from the operating information that is typically
supplied in annual, countywide inventories. In either case, rela-
tively simple algorithms can be developed and computerized to pro-
vide the necessary temporal resolution for the detailed inventory.
VOC emissions are usually disaggregated into species classes
through the use of an appropriate species distribution for each
source category. NOX emissions are either assumed to be all NO or
are split into NO and N02« The allocation of VOC and NOX emis-
sions into classes involves straightforward calculations that should
likewise be computerized.
A major data handling function involved in compiling a detailed
inventory is the development of emission projections. Growth is
typically accounted for by adjusting base year emissions in accord-
ance either with projected changes in the emissions themselves or
with changes in appropriate surrogate indicators of growth (e.g.,
* The actual determination of these apportioning factors, however,
often requires a great deal of manual effort. Refer to Subsection
6.2, "General Methodology for Spatial Resolution," for details.
2-18
-------
earnings, population, land use, and employment). Control regulation
and strategies are reflected in an inventory by adjusting activity
levels, control device efficiencies, or emission factors, as appro-
priate. The data handling system should be designed to automate the
development of growth and control-strategy projections as much as
possible, thus minimizing the amount of manual effort needed each
time a different scenario is modeled.
The final inventorying activity is the generation of the
"modeler's tape" which serves as the interface between the emission
inventory and the photochemical model. Because each model requires
a special computerized format for the inventory data, utility pro-
grams will be necessary to convert the emission inventory file to
the format of the modeler's tape.
Of course, there may be compromises between a completely compu-
terized data handling system and manually carrying out all data
handling functions. Certain functions, such as determining area
source apportioning or growth factors, may be difficult to automate
without developing extremely large and complex sets of programs and
files. Thus, during the planning stages, the agency must carefully
review the data handling flow chart to determine which activities
can be done most efficiently by computer and which functions are
best performed manually. Data handling is further discussed in
Chapter 8.
2.8 MANPOWER REQUIREMENTS
The amount of manpower and types of expertise are the necessary
considerations in planning an emission inventory for a photochemical
model. Depending on the status of the existing inventory and the
amount of detail required, emission inventories for use in a photo-
chemical model can require from 500 to 5,000 manhours. The low
estimate is for a case in which a gridded or very complete
2-19
-------
countywide inventory already exists, and the region is dominated by
only a few major types of sources. The high estimate is for a case
in which little or obsolete inventory information is available, and
the region has many types of significant sources. This case would
require a detailed analysis of all significant sources and much
individual contact with managers of specific sources.
In addition to those usually engaged in compiling the inven-
tory, the agency should enlist the services of (1) a photochemical
modeling specialist familiar with the operation and the VOC species
classes of the particular photochemical model to be used, (2) a
computer programmer to plan the storage and manipulation of the
large amounts of emission data needed, and (3) an urban or regional
planner to analyze transportation and land use data from local
planning agencies and to assist the air pollution control agency in
making emission projections.
2.9 OVERVIEW OF EMISSION INVENTORY PROCEDURES
The remaining chapters of this document discuss the detailed
"how-to" procedures necessary for producing an emission inventory
for use in a photochemical air quality simulation model. Specific
procedures are described for (1) defining the grid system, (2) col-
lecting and compiling source and emissions data from point, highway
motor vehicle, and other area sources, (3) allocating VOC (and
NOX) emission data into species classes, and (4) data handling.
For a perspective of the overall emission inventory process, an
overview is presented in Figure 2-1 of the procedures necessary to
produce a photochemical modeling inventory. Each box in Figure 2-1
has a chapter and section indicated inside that refers the reader to
where each inventory activity is discussed in detail in this
document. (Procedural options for highway motor vehicle data
collection and analysis are reviewed more specifically in Section
5.3).
2-20
-------
1
CHAPTER 4
4
POINT SOURCE
DATA
COLLECTION
(Section 4.1)
1
POINT SOURCE
SPATIAL
RESOLUTION
(Section 4.2)
1
POINT SOURCE
TEMPORAL
RESOLUTION
(Section 4.3)
1
POINT
SOURCE
PROJECTIONS
(Section 4.4)
, ,
INVENTORY
PLANNING
AND DESIGN
(Chapter 2)
1
GRID
SYSTEM
DETERMINATION
(Chapter 3)
i r
I CHAPTER 5 I
4
TRANSPORTATION
PLANNING
DATA
(Section 5.4)
1
EMISSION
FACTOR
MODELS
(Section 5.5)
1
NETWORK
EMISSION
MODELS
(Section 5.6)
1
HIGHWAY VEHICLE
EMISSION GRIDDING
PROCEDURE
(Section 5.7)
1
TRANSPORTATION
CONTROL MEASURE
ANALYSIS
(Section 5.8)
1
I
VOC (and NOX)
SPECIES
ALLOCATION
(Chapter 7)
i
INTERFACING WITH
PHOTOCHEMICAL
MODEL
(Chapter 8)
i
1 CHAPTER 6
4
OTHER AREA
SOURCE DATA
COLLECTION
(Section 6.1)
1
OTHER AREA
SOURCE SPATIAL
RESOLUTION
(Section 6.2)
4
OTHER AREA
SOURCE TEMPORAL
RESOLUTION
(Section 6.3)
i
OTHER AREA
SOURCE
PROJECTIONS
(Section 6.4)
i t
Figure 2-1. Overview of Emission Inventory Procedures
2-21
-------
Prior to initiating the data collection phase of the emission
inventory effort, the agency should be able to answer "yes" to the
following questions:
• Has the size and orientation of the grid been defined? Has
the grid cell size been defined?
• Has the time period (i.e., month, day) been specified for
which the emission inventory must apply?
• Has the existing emission inventory been reviewed to deter-
mine what data therefrom can be utilized in the photochemical
modeling inventory, and what additional temporal and spatial
data must be gathered?
• Does the agency know what VOC and NOX classifications are
needed by the particular photochemical model that will be
run?
• Are the source categories required in the photochemical
modeling inventory (in order to reflect specific control
strategies and be consistent with available VOC split factor
information) compatible with those in the existing, annual
inventory?
• Have the appropriate State and local transportation and plan-
ning agencies been contacted to determine what baseline and
projection data on traffic, employment, population, etc., are
available for use in the detailed inventory?
• Will the detailed emission inventory be utilized for model-
ing pollutants other than ozone? For other seasons than the
summer? Are additional or better-resolved source and emis-
sions data needed for these other uses?
• Are the existing inventory files and data handling systems
capable of generating and storing the added temporal, spa-
tial, and VOC and NOX classification data required by the
photochemical model? Can the modeler's tape of gridded
emissions be readily generated from the resulting point,
line, and area source and emission files? Are sufficient
stack data available to distinguish elevated point sources
from ground level point sources?
• Are sufficient resources available to complete both the base
year and projection inventories, considering both growth and
control strategy options?
2-22
-------
3.0 DETERMINATION OF THE GRID SYSTEM
3.1 CHOOSING AN APPROPRIATE GRID SYSTEM
A primary decision, which will influence all subsequent phases
of the emission inventory process, is the determination of the grid
system. Photochemical models require that emissions be localized
in terms of a number of individual square grid cells of equal size
which are contained within a rectangular boundary, or grid. As an
example, Figure 3-1 shows a map of the vicinity of Tulsa, Oklahoma,
with an emission inventory grid system superimposed. The gridded
area covers a rectangle 50 km wide by 124 km long; it contains
1,550 individual square grid cells, with a grid spacing unit of 2
km. Since emissions must be determined for each grid cell, it is
extremely important to develop an appropriate grid system at the
start of the emission inventory effort.
3.1.1 AREA COVERED BY THE GRID SYSTEM
As shown in Figure 3-1, the gridded area is normally rectangu-
lar. Although some models may accept an irregularly shaped grid
boundary, it is recommended, for the sake of simplicity and ease in
locating grid cells, that a rectangle be used for the boundary of
the emissions area. The forcing of a rectangular boundary around
an irregularly shaped city means that many of the peripheral grid
cells may contain zero emissions. For example, cities on the coast
usually have portions of the ocean included within the rectangular
grid boundary.
The area covered by the emission grid should be large enough
to (1) include all major emission sources in the region, (2) encom-
pass areas of future industrial and residential growth, (3) include
as many pollutant monitoring stations as possible, and (4) incor-
3-1
-------
Inventory Area Boundary
x" J County Boundaries
DCS Grid Boundaries
TULSA INVENTORY AREA
FIGURE 3-1
3-2
-------
porate downwind receptor sites of interest. In this latter regard,
the agency should extend the grid to encompass areas downwind of
the urban area where peak ozone levels generally occur. Thus, the
selection of the grid area to be covered by the emission inventory
is not a simple one, but rather an interdisciplinary decision which
will involve discussions with planning agencies, air pollution con-
trol agencies, meteorologists, and photochemical modeling special-
ists. Modeling specialists know that some photochemical models may
be very sensitive to the levels of ambient concentrations which
must be assumed along the grid boundaries.-*- Thus, in order to
assume relatively clean "background" ambient air concentrations
along the boundaries of the modeling region, the total area for the
emission grid should extend into areas with little or no emissions.
Of course, in certain areas (e.g., the Northeast United States),
pollutant transport from nearby urban areas may preclude the possi-
bility of a clean background along any boundaries that may be
chosen.
In some cases, the exact area to which a photochemical model
will be applied may not be initially known, due to uncertainties
about future land use or the effect of meteorological conditions.
In these cases, an emission grid as large as possible is desirable,
so that a smaller area can be chosen for photochemical modeling
within the emission grid. Thus, the emission grid can be larger
than the actual grid used for modeling. Generally, however, it is
recommended, for most efficient use of time and resources, that the
emission grid and the model ing grid coincide. This optimum grid
area is a result of close cooperation between inventory personnel,
planning agencies, and modeling experts.
3-3
-------
3.1.2 GRID CELL SIZE
The degree of spatial resolution of the emissions is also a
very important factor and must be decided during the planning
stages of the inventory effort. The choice of an appropriate grid
spacing depends on the overall modeling objectives, the total area
of interest, the amount of manpower available, and the cost of run-
ning the photochemical model.
Ideally, it is desirable to have a very fine grid spacing
(e.g., 1 km) for accurately representing emissions from a variety
of sources in different locations. However, if a very large area
is to be covered, this level of resolution becomes unmanageable.
For example, if the region of interest is 100 km by 100 km, a grid
cell size of 1 by 1 km would result in a total of 10,000 individual
cells for which emissions would have to be calculated. For such a
large area, it is unlikely that such fine resolution would result
in appreciable improvement in predicted ozone levels over the en-
tire region relative to a larger grid spacing, such as 5 km. The
major advantage of the larger grid cell size is that the total num-
ber of grid cells would be considerably less (400 as opposed to
10,000), and both the computer and manpower costs would be signifi-
cantly reduced.
Compromise between the maximum amount of area covered and the
smallest possible grid cell size normally results in grid spacing
being from 2 to 5 km in urban-scale photochemical models. Since
oxidant formation occurs over an appreciable amount of time and
space, grid spacing smaller than 1 km is not recommended. Also,
grid cells smaller than 1 by 1 km would exceed the resolution of
some of the transportation modeling data typically generated by
transportation agencies as well as much of the area source appor-
tioning factor data. On the other hand, grid spacing larger than
10 km would probably mask the effect of individual sources, because
all photochemical models assume that emissions are averaged over
3-4
-------
the entire grid cell area (i.e., when using 10 km grid cells, all
emissions in any cell, including individual point sources, are
assumed to be uniformly emitted from the entire 100 km^ grid cell
area). This type of dilution may lead to inaccuracies in using the
photochemical model.
If a specific photochemical model and grid system have not
been chosen when the inventory is compiled, it is recommended that,
if possible, the smallest grid cell size under consideration be
used. Aggregation to larger grid cells is then a simple procedure
of combining an integral number of grid cells. Thus, the initial
emission grid spacing can be smaller than the actual grid spacing
subsequently used for modeling. However, resource considerations
may preclude such fine emission resolution if the inventory in-
volves a very large grid area. It is important to avoid compiling
the inventory for a grid spacing larger than is subsequently con-
sidered desirable for photochemical modeling. It is very difficult
to allocate emissions to a finer grid cell network once the inven-
tory is completed, but it is easy to aggregate smaller grid cells
to form larger ones. The optimum situation is for the emission
grid cells to coincide with the grid cell size used for photochemi-
cal modeling.
An important consideration in determining the grid system area
and the size of individual grid cells is to examine the overall
objectives for using the photochemical model. If control strate-
gies are to be evaluated over a fairly large area (e.g., an Air
Quality Control Region), then a fairly large grid area is necessary
and a fairly coarse emission resolution may be acceptable. If con-
trol strategies are to be evaluated for a fairly small area (e.g.,
an individual city), then a relatively fine emission resolution may
be warranted. For the modeling objective of evaluating the air
quality impact of a proposed new source, a relatively fine emission
3-5
-------
resolution would probably be necessary, since very large grid cells
may mask the effect of the individual new source. Thus, before a
final grid system is chosen, a photochemical modeling specialist
should be consulted regarding the effect of emission resolution on
modeling predictions.
3.2 MAP GRIPPING PROCEDURES
3.2.1 UTM COORDINATE SYSTEM
When a grid system has been decided upon, it will be necessary
to overlay this grid system on an appropriate map in order to
determine which sources lie within what grid cells, and to deter-
mine area source apportioning factors for each grid cell. The
recommended coordinate system is the Universal Transverse Mercator
(UTM) system,^ which is used in the NEDS and EIS/P&R emission
data systems to reference all point source locations. It is very
important to use the UTM coordinate system from the outset in the
development of the grid system, because changing from one coordi-
nate system to another is usually very difficult. For example,
changing from latitude and longitude coordinates to UTM coordinates
is a difficult procedure involving computerized calculations.
The most accurate maps normally available for gridding pur-
poses are those produced by the U.S. Geological Survey (USGS),
which provides topographic maps in different scales for all sec-
tions of the United States. The more recent USGS maps have a 10 km
UTM grid system superimposed on them; older USGS maps simply have
blue tick marks along the edges that represent the UTM coordinate
system.
3.2.2 ORIENTING THE GRID SYSTEM
It is important to base a master grid system on a USGS map,
since other maps (e.g., highway maps) may have considerable
3-6
-------
inaccuracies. The technique normally used for overlaying a grid
system on a map is to position a transparent plastic sheet over the
3 4
map and to draw the gridded area on the plastic sheet. ' In-
stead of positioning the transparent sheet over the map and then
drawing the grid cell boundaries, it is easier to first draw the
rectilinear grid system to scale on the plastic sheet. The gridded
plastic sheet can then be superimposed on the map and the grid sys-
tem can be conveniently oriented.
The grid system should be aligned so that the grid lines es-
sentially run north-south and east-west.* This is not necessary
for running the photochemical model, but is recommended for the
sake of convenience in defining locations on the grid and to en-
hance compatibility of the inventory with meteorological data.
Likewise, for the sake of convenience, the grid should be ori-
ented so that the grid cell boundaries coincide with the UTM kilo-
meter grid lines (i.e., the grid cells are defined by whole UTM
kilometer numbers.)** This simplifies locating a particular grid
cell and allocating point sources to the appropriate grid cells.
3.2.3 PROBLEMS IN GRIDDING
In addition to making a master grid on a USGS map, it is often
necessary to transfer the grid system to other maps such as detail-
ed street, land-use, or population-density maps. This is necessary
because many of the features that need to be used in the inventory
effort will not be identified on the USGS maps. A major problem is
that other maps may not be as dimensionally accurate or on the same
scale as the USGS map. Thus, while attempting to align the master
grid on a land-use map, one may notice that certain major features
are located in slightly different grid cells than on the USGS map.
If the scale of the auxiliary map is not quite accurate, it may be
* Within the region typically modeled in most urban areas, a grid
system based on UTM coordinates will largely meet this criterion.
** Obviously, if nonmetric grid cell dimensions (e.g., miles) are
used, this will not be possible.
3-7
-------
possible to extend or decrease the map grid line dimensions so that
most grid cells correspond to those on the USGS map. In many
cases, the best procedure is to align the main urban area as cor-
rectly as possible. Inaccuracies in the outer portions of the grid
are less important because fewer emissions normally occur in the
outlying grid cells.
Although USGS maps are dimensionally accurate, they are not
always detailed enough to locate particular sources. This is part-
ly because the most detailed scale of any USGS map is only
1:24,000, and partly because some of the available USGS maps are
old (e.g., circa 1950) and so do not show current street locations.
Thus, it is usually necessary to obtain more detailed street maps
covering the entire area of interest in order to accurately locate
specific sources. If possible, it is desirable to have all street
maps on the same scale, so that a number of them can be combined to
show a larger area. The task of overlaying the grid system on the
individual street maps is a difficult one because street maps rare-
ly have UTM coordinates; thus, care must be used in positioning the
grid system according to the major street arteries shown on the
USGS map.
3-8
-------
References for Chapter 3.0
1. K. L. Demerjian, Photochemical Air Quality Simulation Modeling:
Current Status and Future Prospects. EPA-600/377-001b. 777-794,
International Conference on Photochemical Oxidant Pollution and
its Control, January 1977.
2. Grids and Grid References. Technical Manual No. TM 5241-1,
U.S. Department of the Army, Washington, DC, June 1967.
3. Guidelines for Air Quality Maintenance Planning and Analysis,
Volume 8: Computer-Assisted Area Source Emissions Gridding
Procedure, EPA-450/4-74- 009, U.S. Environmental Protection
Agency (OAWM, OAQPS), Research Triangle Park, NC, September
1974.
4. Guidelines for Air Quality Maintenance Planning and Analysis,
Volume 13: Allocating Projected Emissions to Sub-County
Areas, EPA-450/4-74-014 (OAWM. OAQPS). Research Triangle Park
NC, November 1974.
3-9
-------
-------
4.0 POINT SOURCE EMISSIONS
4.1 DATA COLLECTION
As discussed briefly in Chapter 2.0, a good starting place for
point source inventory used in a photochemical modeling application
is the existing, basic point source inventory prepared by the State
or local agency. Procedures for the preparation of the basic point
source inventory for use in less detailed source/receptor relation-
ships are outlined in Volume I of this series.* State and local
point source inventories can be obtained in standardized formats
from the State, if the State uses EPA's Emission Inventory System
(EIS/P&R), and from the National Emission Data System (NEDS).
An example of a NEDS point source record is shown in Figure
4-1. Three types of information are coded for each facility.
First, identifying information is associated with each level of data
to allow the user of the system to arrange, select, and manipulate
the inventory data to his needs. This includes plant and point
identification numbers, UTM grid coordinates, plant names and
addresses, and key contacts. The second type of information is
process information, including equipment descriptors (e.g., source
classification or basic equipment codes), stack parameters, control
device information, operating rates, and fuel characteristics. The
third type of information is on emissions from each process within
the facility. The primary pollutants of concern from an oxidant
standpoint, of course, are VOC and NOX; annual estimates of these
should be available from the basic inventory.
In relation to point sources, the existing, basic inventory
resulting from the procedures in Volume I will fulfill most of the
4-1
-------
i
i
x
Ul
1-
v
>
in
<
h-
4
'o
'.
:
in
in
z
UJ
_*
«
z
o
-
<
z
1
1*
z
^
tfl
_l
k,
u
K
3
O
in
*-
z
0
ft.
STATEI2AM MISSOURI AQCRI09H)! METROPOLITAN KANSAS CITY CITY(23gn|I KANSAS CITY
COUNTYI2210II JACKSON CO PLANT 10! 0031 POINT ID! OJ
HAME-ADDRESS- XYZAB BRASS CO 123 NOWHERE RD. SICI]3q|)| SFCONDARV SMELTING AND REFINING OF NONFERROUS MFTALS
PERSONAL CONTACT! DOE, JOHN R. 9CC I M-OI-? »9-9» ) I POINT SC EVAP -CLEANING SOLVENT .OTHER/NOT CLASIFO -SPECIFY |N REPARK
MSSIONS • ALLOWABLE EMISSIONS F0» POINT 1
• GENERAL INFORMATION I UTM GRID COORDINATES ! HAND CALCULATED POINT E>
N
in
z
c.
£
f
«
0,
•• ••
1
o
UJ.
«
3
1
-..
tfl
kl
Z
O
f-g
t-
3
k.
O
kl
H-
<
0
X V >. >
in inim w»
z z z z
o o c o
•• ••
X >
o c
in z
U 0*
z u
....^..;,-._l^.--.-^-..f
LCD SCC EMISSIONS f
COMPUTER n
*•**>•>•,
^ +J^^
m y\ in tn m
Z Z Z Zl 0
O 0; OOL O
0 a
X X1
SI
•. —,
i
r^»
rr
roi
f**
.
>—
z
O
ls<
CC
O
X
o
1^
»»
LAST UPDATE
• ^
O O Ul
«• . *
z
! 3
u o
X U
w
m
ro
en
m
V
-J
«
u
K
«
u
>
o
-J
<
>
z
OWNERSHIP:
tt
.>
in
z
c
»-
oe
<
ft.
•
i
•
• k
• _1
1. 0)
K
inl
S
H
u u
.j _i
<£ C
• o u u
II . _l _1 -1
-^ • ft. ft. ft.
in
u
X
c
in
3
u
J
|PP PROCESS
• » « «
• ,
• »-| h- >-
• O 0 0
• z z z
•
1» »- X X
• a. o c
• «. in z
.ft.,
>- t-i 0
U. kjk. «
o a o o
0
i
i
•4 kl
elK •'
•• kl 3 UJ
•- «-
X UJ
a x
Wl —
X 0
M K
u u
« «
!•- ^«
in in
in
in
Ul
SOURCE! PRO(
! GAS TEMPERAT
ITIONS • GAS FLOW RAT
a
f
roNS/YR •
•
1
K
Q
h-
^
U.
in
z
UJ
«
t
z
k
o 3
u m
o u
£ ..
a. ft.
k.1 «
s! s1
c, c
a
u
X
f
!
FT • HCj NADB-A
. - ! CO;_NPT AP
ENTIFICATION
oin —
^ *-
z. o
— 0
« 0. Z
o
w
"
o
I
^
EQUIPMENT f PAR
f
J
*
!
^.
EQUIPMENT 1
EQUIPMENT 1 SO
f
EQUIPMENT f
„
0
•
o
°
EQUIPMENT I NO
I
.1
o
•
o
EQUIPMENT f
EQUIPMENT f H
?
EQUIPMENT !
O 0 O O 0 ( O o| O
>- '«-l-l 1-
Z IZ Z Z
O O O| O
X i U
u in! x o o
< t-l V Z Z
1- Zl U !
in ui. U; ••• ••
>
o
Z X
iH
«n
UJ
X Ul
3 X
-1 «
ft. in
rsi
'I;
• •<
• -x.
,_ NORMAL OPf
HOUR
>
kl
O
!
i
fv
w
UJ
2
in
>- v
o u
jt
»~
K
«
0.
£
«
X
Of
ft.
»•
3.
ft.
X
O
3
0
•EBI 20 I f
xi T
* ANNUAL
u
a
1
0 0
z z
XX
O 0
in in
JV -
DCO
« Z
£ O
— U
ft. in
f •
0 0
IM 3-
>- 19
«
1 1
< Z
X 3
-5
"
i
»
5
•«
z
L
«
O
C*
>
O
i- .t- t-' t-
Z Z Z| Z
o loo! o
0
z
..
X
i
0
z
o
u
in
M
O
•
O
O
• «
G.
UJ
in
X
u
u
<
ft.
in
w
""
0 0
zz
'RIMARY HC
ECOND. .HC
o
u.
z
COMPLIANCE
i
•RIMARY CO
kl
Z
-J
ft.
X
o
w
z
o
•
o
o
! NO CONTROL EQUIPMENT 1 C
I SECOND. CO
FE! /
SCHEDULED
COMPLNCE DA]
OPERATING RATf's f 1 L-L-I
ERISTICS
! FUEL CHARACT
k,
53
Z
<
J
0.
z
o
o
ANNUAL OPF'«AT|N6 RATF! 20 TONS SOLVENT USEO I
1
HOURLY MAX* OFSIRN RATT: .005 Tn^s SOIVF>'T HSFO f
•
BOILER DESIGN CAPACITY! !
• _ • *
rnM''F»'TS! inLvrMTS L^ST f
i System Point Source Listing.
CONTENT! O',00 *
CONTENT! 00.0 *
CONTENT!
ional Emissions Dat<
UPDATE! / / «
•
FMERGENCY CONTROL " ! FUEL SULFUR
tCTtON PLAN •
HAS BEEN SUBMJTTEO !
! FUFL MEAT
Figure 4-1 . Nat
1 ,
4-2
-------
needs of the photochemical modeler. One important deficiency, how-
ever, is the lack of hourly emission estimates. Provision for this
additional temporal resolution is described in the subsequent sec-
tion. Another shortfall is that the basic inventory will not dis-
aggregate VOC and NOX emissions into the classes needed by photo-
chemical models. Techniques for accomplishing this are described in
Chapter 7.0. The only other data required that may not be in the
basic inventory are the various stack parameters needed in order to
segregate elevated point sources from ground level sources. If the
photochemical model requires vertical resolution of emissions, the
point source records should include the following stack parameters
to allow the modeler to calculate effective stack heights:
• Physical stack height
• Inside diameter of stack
• Exit gas temperature
• Exit gas velocity or volumetric flow rate
For point sources, information in the local agency files will often
be more current than NEDS data, as errors may be corrected first in
local files and only somewhat later in NEDS.
4.2 SPATIAL RESOLUTION
Photochemical models require that all emissions be associated
with specific grid cells. Hence, prior to running a photochemical
model, the emissions from each point source must be assigned to the
grid cell in which the point is located. This assignment can be
done manually using maps or automatically by computer.
Point sources can be manually assigned to grids by their UTM
coordinates (or street addresses) on a map of the area which is
overlaid with the inventory grid system (refer to Chapter 3.0). Map
pins are sometimes convenient to mark source locations. It is ad-
visable to check the street address against the UTM coordinates to
4-3
-------
identify possible errors in UTM assignments. Usually each grid cell
is assigned a number according to some model-specific system, and
this grid number and the source-type category should be entered into
the data handling system for each point source.
An alternative to this manual process is to develop a computer
program that automatically compares the UTM coordinates of each
point source with the coordinates of each grid cell and assigns the
point source emissions to an appropriate cell. For some photochem-
ical models, preprocessor programs are available that may accomplish
this function. The model documentation or developer should be con-
sulted in order to determine whether this capability is available.
Generally, an automatic grid-coordinate-to-grid-cell approach
is preferable because it is faster than manually locating all point
sources on a map. This approach is especially attractive if (1)
preprocessing programs for the purpose are available as part of the
photochemical model, and (2) the grid assignment process may have to
be repeated numerous times, as would be the case if the grid orien-
tation or grid cell size would change. However, even if the grid
assignment is computerized, it will usually be useful to provide
grid system overlay over an accurate map of the area, in order to
assist in visualizing and checking the assignments produced.
4.3 TEMPORAL RESOLUTION
One of the main differences between the detailed inventory used
for photochemical modeling and the basic inventory is that the for-
mer should represent a typical weekday during the oxidant season.
4-4
-------
Moreover, for that typical day, emission estimates must be made for
each hour. In contrast, basic inventories used in rollback or EKMA
contain annual or, in some cases, seasonally adjusted emissions.
Three basic approaches can be used for determining hourly emissions
from point sources for a given day: (1) plant or local agency
contacts, (2) extrapolation from operating information contained in
the basic point source record, and (3) engineering judgment.
Ideally, the most accurate means of determining hourly emis-
sions would be to contact each plant in order to get operating
records for each hour for a typical day during the oxidant season.
Certain local agencies may also have this information. In practice,
since this approach is quite resource-intensive, it is practical
only for the largest emitters in the area. For example, power
plants, which are large emitters of NOX, generally keep detailed,
hourly records of fuel firing rates and power output for each day of
operation. Automotive assembly plants, refineries, and tank farms
would be examples of other sources that could be contacted directly
for operating information.
For many smaller point sources, reasonable temporal resolution
can be obtained from the operating data that are typically coded on
each basic point source record. For example, the NEDS listing shown
in Figure 4-1 refers to an operation with annual emissions of 20
tons of VOC, with 40 percent of annual throughput occurring in the
summer. This source normally operates 12 hours per day and seven
days each week* Assuming uniform hourly emissions over a 13-week
summer, the emissions rate is estimated to be (20 x 0.4)/(12 x 7 x 13)
or 0.0073 tons per hour. Applying the conversion factor, 907 kg/ton,
gives 6.7 kg/hr as the average emissions during summer operations.
In the absence of more specific data, these emissions might be
assigned to the period 0700 to 1900 each day.
4-5
-------
For many sources, daily operation will be confined to one or two
workshifts; thus, hourly operation during working hours would be
determined by dividing the daily operating rate by 8 or 16. If
hourly operating information is contained in local agency files, it
may supersede the less detailed information from NEDS or EIS/P&R.
There are always some sources that are both too minor to
warrant a plant contact, and for which no operating data will be
coded into the basic point source record. In these cases, engi-
neering judgment can often provide satisfactory estimates of hourly
emissions. For example, many commercial establishments will operate
all year, but only be open 8 to 10 hours a day and 5 to 6 days a
week. Hence, as a good approximation, the annual operating rate in
the basic point source record can be divided by 2,080 (i.e., 52 x 5
x 8) to estimate an hourly operating rate applicable during working
hours.
Fixed-roof petroleum product storage tanks present a unique
situation because breathing loss emissions are not a function of
operation. Rather, test data^ show that breathing loss emissions
are simply a function of time of day. These tanks begin expelling
vapors when heated by sunshine in the morning, and cease expelling
vapors in the mid-afternoon when the heating process ceases. As an
approximation, breathing loss emissions from fixed roof storage
tanks can be assumed to -occur uniformly from 8:00 a.m. to 3:00 p.m.
Daily emissions from storage tanks can be estimated by procedures
given in AP-42, Compilation of Air Pollutant Emission Factors.3
4.4 POINT SOURCE PROJECTIONS
As discussed in Chapter 2.0, emission projections must take
into account growth as well as any regulations under consideration
4-6
-------
to control ozone precursor pollutants. The baseline projection con-
siders the impact of expected growth in future years and the reduc-
tion in emissions that should occur as a result of existing control
regulations. Control strategy projections, on the other hand, con-
sider the reductions in emissions that would occur if alternative or
additional regulatory programs were adopted. Control strategy pro-
jections may also take into account other-than-expected growth pat-
terns which might result from the alternate control programs.
4.4.1 INDIVIDUAL FACILITY PROJECTIONS
The best approach for projecting emissions from major point
sources is to obtain information on each facility. This type of
information would ideally be determined by contacting plant manage-
ment directly, or could be solicited on questionnaires. Generally,
questionnaires would not be sent out solely to solicit projection
information; however, this additional information may be solicited
on questionnaires used to periodically update the baseline inventory.
Permit applications submitted to various Federal, State, and local
agencies should also be screened to get information on expected
expansion or new construction. In addition, the local Metropolitan
Planning Organization and other planning bodies should be contacted
for any information they may have on projected industrial expansion
as well as to comment on the reasonableness of any plans submitted
by pi ant personnel.
Once this type of projected plant growth information is obtain-
ed, the agency needs to determine what regulations will apply, in
order to estimate controlled emissions. In the baseline case,
existing, applicable regulations would be imposed and evaluated.
For instance, a fossil-fueled power plant now under construction and
expected to start operation in 2 years would be subject to Federal
new source performance standards for particulate, $03, and NOX.
4-7
-------
Hence, it would be reasonable to assume emission levels equal to the
standards unless plant personnel indicate more stringent controls
will be applied for some reason (e.g., to meet a more stringent
local standard). Similarly, in control strategy projections,
effects of any alternative standards would have to be evaluated.
When obtaining projection information from plant management, it
is important to inquire whether any increased activity will occur at
the existing facility or whether it will occur elsewhere, at another
existing plant or at a new plant. If occurring at an existing fa-
cility, the agency also needs to determine whether the growth will
be expansion to existing capacity or will require additional capaci-
ty. These considerations are important especially for major
sources, since emissions must be assigned to a specific grid cell.
They will also help the agency to determine what additional control
measures are likely to be required. The schedule for completion of
any expansion or new construction is also needed, in order to deter-
mine in what year the source must be included in the projection
inventory.
As an example of making point source projections for specific
sources, consider a facility employing a large open-top vapor de-
greasing operation that emitted 100 tons of solvent per year in 1977
(based on an annual production of 10,000 of a certain kind of metal
part.) Assume that no control measures were taken to reduce solvent
losses from the process. Now, suppose a plant contact was made and
it was learned that 5 percent more metal parts would be produced per
year until 1982 using the existing operation and that, in 1986, a
replacement facility would be brought on line at another location
that produced 20,000 parts per year. Moreover, suppose that the
source is located in an ozone attainment area where RACT is not
required on VOC sources. Then, in order to estimate VOC emissions
from this source for a 1982 projection inventory, one could assume
4-8
-------
that since no additional controls would be expected, the current
emission level could be multiplied by the ratio of cumulative growth
in metal parts production (i.e., 5 years at 5 percent/year =
[1.05]5 = 1.28, or 128 percent) to estimate VOC emissions in 1982.
In this manner, emissions in 1982 would be estimated at 128 percent
of 100, or 128 tons per year, and the point source record for this
projection year should be adjusted accordingly to take this growth
into account.
Again, suppose a control strategy projection is desired for
1987 to evaluate the effect of RACT as an alternate control strate-
gy. In this case, both growth and controls would have to be consid-
ered. As a first approximation, one could assume that since produc-
tion is twice as much in 1987 as 1977, uncontrolled emissions from
the replacement plant would be twice those of 1977, or 200 tons per
year, assuming that a similar open-top vapor degreasing operation
would be used in the new facility. Since the new plant would be
subject to RACT in this control scenario, VOC emissions would be
reduced by 45 to 60 percent from the uncontrolled case.^ Hence,
projected emissions in 1978 would be only 80 to 110 tons per year,
depending on which RACT measures were instituted. Note that since a
new facility was to be built between 1977 and 1987, a new point
source form should be coded in the 1987 projection inventory and the
old record deleted or assigned zero emissions.
As is obvious from this example, even when projection informa-
tion is available for specific facilities, certain assumptions will
have to be made in order to project emission levels for some future
year. For instance, in the 1982 baseline projection, it was assumed
that emissions would increase proportionately with production. This
may not necessarily be entirely accurate, depending on the nature of
the operation. This same assumption, along with an assumed emission
reduction due to RACT, was also utilized in making the 1987 control
4-9
-------
strategy projection. This underscores the point made in Section 2.6
that projections are always somewhat speculative in nature.
4.4.2 AGGREGATE POINT SOURCE PROJECTIONS
In many instances, projection information will not be available
for every facility in an area of interest. Some plant managements
will not be willing or able to provide forecasts of their corporate
plans, especially for more distant years. In addition, many plants
in certain source categories (e.g., small industrial boilers) will
be too small and too numerous to warrant the solicitation of projec-
tion information individually. In these situations, other proce-
dures need to be employed to make projections of future emissions.
Two possible approaches are discussed below.
In one case, projection information may be available on many
point sources within a given category, but for various reasons is
not obtainable for one or several facilities. For example, there
may be 10 paint manufacturing plants in the area of interest, and
successful contacts may have been made with only eight of these. In
this situation, a reasonable approach to projecting growth and emis-
sions at the remaining two plants would be to evaluate the growth
trends in the plants for which projections are known and apply them
to the plants for which no information was available. In the exam-
ple of the paint manufacturing plants, if production was expected to
expand by 6 percent per year, on average, for the eight plants, then
this rate could be applied to the two plants to estimate expected
growth. Then, knowing the increase in production, the appropriate
control measures would be taken into consideration in making a base-
line projection. In some cases, the emissions could be directly
estimated by applying the average growth rate to a base year emis-
sion for each plant. Good engineering judgment is needed in this
practice to screen out any unreasonable projections that may result.
4-10
-------
For minor point sources, such as cold cleaning operations,
where individual solicitation of projection information is unwar-
ranted, the rate of growth of activity may be assumed equivalent to
that of some growth indicator category for which projections have
been made by local MPO's or by OBERS.^ For example, it might be
assumed that cold cleaning operations would grow at the same rate as
industrial manufacturing in general. This rate can be readily
estimated from projections of employment in industrial manufacturing
categories.
Table 4-1 provides an example of local employment projections
for a set of 10 employment categories or "sectors," of which one is
"manufacturing." OBERS projections are provided in many more cate-
gories; some of the relevant OBERS categories are listed in Table
4-2. These projections were developed and are regularly updated by
the U.S. Departments of Commerce and Agriculture and may be used, in
the absence of local projections, as general indicators of growth.
Regardless of the indicators used for projections, the basic
mechanics of projecting are the same: the ratio of the value of the
surrogate indicator in the projection year to its value in the base
year is multiplied by the aggregated activity level for the point
source category in the base year. For example, if cold cleaning
operations are assumed to correlate with the manufacturing sector in
Table 4-1, then the level of this activity in 1985 would be 42/39,
or 1.08 times that in 1980. In many cases the projection years of
interest to the air pollution control agency will not directly cor-
respond to the years for which growth projections have been made,
thus requiring interpolation. Local input should be sought regard-
ing whether straight-line or other interpolation methods should be
employed.
4-11
-------
Table 4-1. PROJECTED EMPLOYMENT BY SECTOR IN HILLSBOROUGH COUNTY3
(in thousands)
Employment Sector
Agriculture
Mining
Construction
Manufacturing
Transportation, com-
munications, and
public utilities
Wholesale and retail
trade
Finance, insurance
and real estate
Services, miscell-
aneous , and mining
Government
Other
TOTAL
1980
7
2
19
39
21
72
16
47
43
28
293
1985
7
2
21
42
23
79
18
51
47
31
321
1990
8
2
22
45
25
85
19
55
50
33
344
1995
8
2
24
48
27
90
20
58
53
35
365
2000J
9
2
25
51
28
95
21
61
56
37
385
a Refer to Reference 5
4-12
-------
Table 4-2. SELECTED OBERS INDUSTRIAL CATEGORIES5
Chemicals and allied products
Food and kindred products
Primary metals
Petroleum refining
Lumber products and furniture
Paper and allied products
Fabricated metals and ordnance
Textile mill products
Other manufacturing
Total manufacturing
4-13
-------
When aggregate growth is determined for a source category in
the above manner, a major problem arises in attempting to distribute
the increased activity appropriately to the grid cell level. In
other words, difficult assumptions must be made regarding the proba-
ble location of the new activity. One way to apportion- growth is to
assume that it occurs only at existing plants in the same source
category. For instance, if the 10 paint manufacturing plants in the
previous example manufactured 10 million gallons of paint in 1977,
and 15 million gallons were projected in 1982, then the additional 5
million gallons could be assumed to be manufactured at these same
plants. The amount of production assigned to each plant would be in
proportion to the quantity currently manufactured there.
If the agency feels it is not reasonable to assume that all
growth within a particular source category will occur at existing
facilities, an alternative is to apportion the growth according to
the fraction of increase of industrially zoned land within each
grid. Using the above example again, if 5 percent of the projected
areawide increase in industrially zoned land occurred within a given
grid cell, then 250,000 gallons per year (or 5 percent of 5 million
gallons) of additional paint production would be assigned to that
cell. In this approach, since the growth is not assigned to exist-
ing facilities, new point source records may have to be coded.
Again, the applicable control regulations would have to be evaluated
in order to determine the resulting emissions.
4.4.3 POINT-SOURCE PROJECTION REVIEW AND DOCUMENTATION
Because the projection inventories are so important in control
strategy development, they should be reviewed internally by the air
pollution control agency and presented to as many other groups as
possible for comment before being finalized. In order for the pro-
jection inventories to undergo this careful scrutiny, it is essen-
4-14
-------
tial that all assumptions, procedures, and data sources be carefully
documented. This review and documentation process will help assure
that the projections are (1) consistent with other projections being
made by various groups in the area, (2) objective in the sense that
they are not biased in order to promote a particular policy, (3)
open, because all assumptions, etc., are clearly stated for public
review, and (4) defensible, because of all the above characteristics.
The key aspects of point-source projections that will invite
criticism are: (1) which indicators are used for projecting activi-
ty level growth; (2) when and where this growth will occur, and con-
comitantly, whether it will be accommodated by expansion of existing
facilities or new construction; and (3) what emissions will be asso-
ciated with this growth, either in the baseline case or as a result
of various candidate control strategies. When planning, compiling,
and reviewing the point source projection inventory, the agency
should focus particular attention on these issues.
4-15
-------
References for Chapter 4.0
1. Procedures for the Preparation of Emission Inventories for
Volatile Organic Compounds. Volume I. EPA-450/2-77-028, U.S.
Environmental Protection Agency, (OAQPS, MDAD), Research
Triangle Park, NC, December 1977.
2. Breathing Loss Emissions From Fixed-Roof Petrochemical Storage
Tanks (Draft). EPA Contract No.68-02-2815, Work Assignment
No. 6, Engineering-Science, Inc., July 1978.
3. Compilation of Air Pollution Emission Factors, Third Edition,
AP-42, U.S. Environmental Protection Agency, Office of Air
Quality Planning and Standards, Research Triangle Park, NC,
August 1977.
4. Control of Volatile Organic Emissions From Solvent Cleaning,
EPA-450/2-77-022, U,S, Environmental Protection Agency,
Emissions Standards and Engineering Division, Research Triangle
Park, NC, November 1977.
5. Regional Economic Activity in the U.S., 1972 OBERS Projections,
U.S. Department of Commerce, Bureau of Economic Affairs, Office
of Business Economics, Analysis Division and U.S. Department of
Agriculture, Economic Research Services, 1974.
4-16
-------
5.0 HIGHWAY VEHICLES
5.1 INTRODUCTION
Highway vehicles are frequently the largest single source of
both VOC and NOx emissions in an urban area; hence, it is essential
for the photochemical modeling effort to have an accurate inventory
of highway vehicle emissions. Typically, the annual, countywide
emission inventory of highway vehicle emissions is determined by
some technique (e.g., countywide gasoline sales) that yields only
gross emission estimates for the area. This type of inventory is
not a good starting point for the detailed photochemical modeling
inventory for several reasons. First, such an inventory is not
spatially or temporally resolved to the degree needed by photo-
chemical models. Second, many highway vehicle control strategies
cannot readily be evaluated using this type of inventory. Third,
and most important, transportation planning data that can often be
utilized to compile detailed base year and projection inventories
are available in most urban areas (>50,000 population). Hence,
during the planning stages of the inventory effort, the air
pollution control agency should contact the local MPO and proceed
with the development of the detailed photochemical modeling
inventory based on detailed transportation planning data.
5.2 NATURE OF AUTOMOTIVE EMISSIONS
Because the various models available for compiling detailed
highway vehicle inventories treat the exhaust and evaporative com-
ponents of automotive emissions differently, it is important that
the reader have a basic understanding of the nature of automotive
emissions prior to selecting a particular approach. The three
major components of automotive emissions are briefly discussed
below.
5-1
-------
Travel related emissions - Travel related emissions are also
referred to as exhaust emissions, or hot stable operating mode
emissions. These emissions are a function of VMT and occur during
travel when the vehicle is warmed up past the cold start (or hot
transient) mode.
Trip-end related emissions - These consist of startup and hot
soak emissions. Startup emissions are those exhaust emissions that
occur near the origin of the trip when the vehicle is warming up.
Startup emissions include both cold start and hot transient emis-
sions and are a function of ambient temperature, the number of
trips, and the number of vehicles started by engine operating tem-
perature mode. Cold start emissions are those that occur roughly
within the first 505 seconds of a start after the engine has been
turned off for four hours or more, in the case of noncatalyst-
equipped vehicles, and for one hour or more for catalyst-equipped
vehicles. Hot transient emissions are those that occur roughly
within the first 505 seconds of a start after the engine has been
turned off for less than one hour. Hot transient emissions are
only associated with catalyst-equipped cars, and are considerably
less than cold start emissions. Hot soak emissions are evaporative
losses that occur from the carburetor after the engine has been
shut down. These emissions occur at the end of a trip, primarily
within the first hour after engine off.
Vehicle Related Emissions - Also known as diurnal emissions,
these are evaporative emissions from the fuel tank, caused by
diurnal changes in ambient temperature. An increase in temperature
results in the expansion of the air-fuel mixture in a partially
filled fuel tank, and as a result, gasoline vapor is expelled into
the air. These emissions are a function of the number of vehicles.
5-2
-------
5.3 OVERVIEW OF HIGHWAY VEHICLE INVENTORY PROCEDURES
The basic process for compiling the detailed highway vehicle
emission inventory necessary for photochemical modeling is shown in
Figure 5-1. Four distinct tasks are involved, viz., development of
the required transportation input, use of an emission factor model,
use of a network emission model, and conversion of line source
emissions to total grid emissions.
Two fundamentally differing approaches can be used for carry-
ing out the general methodology outlined in Figure 5-1; (1) a
link-based approach and (2) a "hybrid" approach.1 in the link-
based approach, all three types of automotive emissions are calcu-
lated for each link in the highway network based on VMT on that
link. A single emission factor is developed for each link in terms
of g/VMT, representing total automotive emissions. Trip-end and
vehicle related emissions are included in this single emission fac-
tor and are assumed to be "smeared out" evenly over the entire link.
With the link-based approach the procedure works basically as
follows. The core of the analysis is the network emission model.
This model takes each traffic link, multiplies the link length by
the traffic volume to determine VMT for the link, and multiplies
the VMT by an emission factor. This procedure yields emissions for
each link and these link emissions are then summed using various
methods during the conversion to gridded emissions function.
In the hybrid approach, separate estimates are made of travel,
trip-end, and vehicle related emissions, and assigned to those
zones in which they actually occur. The hybrid approach is concep-
tually more accurate than the link-based approach, and allows
certain control strategies to be simulated more easily. However,
this approach is generally more data intensive.
5-3
-------
TRANSPORTATION
MODELS
EMISSION
FACTOR MODEL
NETWORK EMISSION
MODELS
CONVERSION TO
GRIDDED EMISSIONS
Figure 5-1. Overview of Processes for Compiling Detailed
Highway Emission Inventories
5-4
-------
Because the hybrid approach is so much more data intensive,
and because, at this time, no generally available models using the
hybrid approach have been developed, the emphasis in Volume II will
be on the link-based approach. Some discussion of the hybrid
approach will be included, however, for those agencies that wish to
incorporate this added level of detail to their highway vehicle
emission inventory.
The choice of whether to use the link-based or hybrid approach
must be made before the highway vehicle emission inventory is
begun. This choice affects the types of data that must be col-
lected as well as the calculation procedures.
5.4 TRANSPORTATION PLANNING DATA
As shown in Figure 5-1, the starting point for the detailed
highway vehicle inventory is the transportation planning process.
All urban areas with populations greater than 50,000 are required
by Federal law to have continuous, comprehensive and cooperative
(3-C) transportation planning processes to enable local governments
to better plan their major street systems. Generally, metropolitan
planning organizations (MPO's) are the designated agencies to do
transportation planning.
As part of the transportation planning process, batteries of
computer programs (i.e., transportation models) are utilized to
estimate traffic flow on transportation networks for current and
projection years. In most urban areas, either FHWA's PLANPAC/
BACKPAC battery2 or UMTA's UTPS battery3 is used for this pur-
pose. A few states use different program batteries but these are
usually quite similar to the FHWA or UMTA systems.
5-5
-------
The basic input to transportation models consists of current
and projected population and economic data (disaggregated to the
traffic zone level) as well as descriptions of the current and pro-
jected traffic networks in the area of concern. The primary output
from transportation models that is of concern to the air pollution
agency is the historical record. This file contains link-by-link
information on traffic volumes as well as various other link-
related data needed to calculate emissions at the link level.
Table 5-1 indicates the type of data that can be associated with
each link in the PLANPAC/BACKPAC series. Each link is spatially
defined by an a-node and b-node and x and y coordinates. The
associated information constitutes the historical record. It
should be noted that not all data items are gathered for each
transportation study; the basic battery of programs can operate
with only columns 2-21 specified.
With the link-based approach, the desired product from the
transportation model is link-by-link traffic volume, by hour, for
both the base year period and the future or target year period.
With the hybrid approach additional data requirements are data on
trip origin and destination, by purpose, by hour. These data are
potentially available from the transportation modeling process but
several limiting factors should be considered.
First, the transportation modeling process used today is fun-
damentally similar to the process that began 20 to 30 years ago.
While this modeling process is effective for transportation
purposes, it was not originally intended for use in air quality
work. Several assumptions and procedures are included that have
little effect on transportation planning but do affect air quality
planning. The most significant of these is the "total VMT" concept
used in air quality work. Total VMT is seldom used in transporta-
tion modeling and is therefore difficult to derive. Total travel
5-6
-------
Table 5-1. LINK DATA FOR PLANPAC/BACKPAC URBAN TRANSPORTATION
PLANNING BATTERY
Columns
Contents
1
2-6
7
8-12
13
14-17
18
19-21
22-24
25-28
29-31
32-36
37-38
39
40
41
42-44
45-47
48-51
52-54
55-59
60-61
62
63
64
65
66
67
68
69-70
71-74
75-78
79
80
Unused (perhaps identification)
a-node number
a-node leg number (0-3)
b-node number
b-node leg number (0-3)
Distance (XX.XX)
T or S for time or speed (a-b)
Time or speed (a-b) (X.XX/XX.X)
Turn penalty codes at node b
Hourly capacity (a-b)
Conversion factor (VPH/ADT)
Directional count (a-b)
Street width (a-b)
Parking (a-b)
Unused (a-b) -
T or S for time or speed (b-a)
Time or speed (b-a) (X.XX/XX.X)
Turn penalty codes at node a
Hourly capacity (b-a)
Conversion factor (VPH/ADT)
Directional count (b-a)
Street width (b-a)
Parking (b-a)
Unused (b-a)
Administrative classification
Functional classification
Type facility
Surface type
Type area
Predominant land use
Link location
Route number
Condition
Unused
a-b
(Note 1)
b-a
(Note 2)
Notes:
1. If an "X" is coded in Column 41, the a-b values in Columns 18-40
are halved and moved to Columns 41-63 of the card.
2. If only an "all or nothing" assignment is desired, only Columns
1-21 need to be coded.
Source: Computer Programs for Urban Transportation Planning, PLANPAC/
BACKPAC General Information Manual, U.S. Department of Transportation,
Federal Highway Administration, Washington, D.C., April 1977.
5-7
-------
in the transportation planning process consists of VMT on: (1)
links; (2) centroid links, representing interzonal travel on un-
coded streets; and (3) intrazonal travel, representing short trips
within the traffic zone of origin that are not assigned to links
because they do not enter another traffic zone. All three catego-
ries must be included to properly characterize total VMT. To include
intrazonal travel with the PLANPAC/BACKPAC system, the TRPTAB pro-
gram must be run, which generates the number of zone-to-zone traf-
fic interchanges. With the DTPS system, program UMFTR is used to
obtain the number of intrazonal trips. The number of intrazonal
trips for each zone is indicated by interchanges with itself,, i.e.,
zone 1 to zone 1, 2 to 2, 3 to 3, etc. Once the total number of
intrazonal trips is known, it is necessary to develop an average
intrazonal trip distance for each zone to develop a VMT figure.
The centroid connector length can be used if properly located. The
use of this information in calculating emissions from intrazonal
VMT is discussed in Section 5.6 on network emission models.
A second limitation deals with the accuracy of the data input
to the transportation modeling process. The accuracy of traffic
volumes is no better than the population, employment, and network
data used to project these volumes. If the transportation modeling
process is several years old, the projections of population, em-
ployment, and new roads or transit systems (network description)
may no longer be accurate. This will cause errors in the projected
traffic volumes. The air pollution agency should work very closely
with local land use and transportation planners to obtain accurate
projections, and the projections should be reviewed by qualified
personnel before the traffic volumes are used for air ouality work.
If they are found to be significantly inaccurate, it may be neces-
t
sary to update the transportation study. This is an extremely
5-8
-------
costly procedure and may involve several months of effort by land
use and transportation planners.
A third limitation involves the availability of traffic as-
signments for the desired years of analysis. Unless short-term
traffic projects have been made for transportation system manage-
ment work or specifically for air quality planning work, traffic
assignments will not be available for 1982 or 1987. A base year
period and a target year period, such as 1995 or 2000, are the pro-
jections most likely to be available.
Two methods to obtain link volumes for interim periods without
a major transportation study update are available, although neither
results in a high degree of accuracy. The first method uses manual
calculations based on land use trip generation rates, an estimate
of through travel, and professional judgement. This method should
only be used by an experienced traffic analyst and as a last
resort. The second method relies on linear interpolation of
traffic assignment from two known periods. The future year traffic
assignment assumes that all projected land use growth has occurred
and that all projected transportation improvements have been built.
Actually, in interim periods, this is rarely the case. Land use
growth does not usually occur evenly across the city at a steady
rate; rather, it more often occurs in spurts in specific areas.
Similarly, the traffic diversion resulting from construction of a
major transportation facility is often extremely difficult to cal-
culate. Thus, in high growth areas, this latter method is espe-
cially difficult to use. Hence, because of the large effort in-
volved in photochemical modeling, it is recommended that special
traffic assignments be prepared unless the existing assignments are
within a very few years of the projection years being used in the
photochemical modeling effort.
5-9
-------
Another limitation stems from the inherent lack of temporal
resolution in the transportation planning data. First, since the
underlying travel survey information used in the modeling process
is only available for weekdays, the resulting link and trip genera-
tion data will only apply to weekday conditions. Hence, without
extensive modifications, transportation planning data cannot be
utilized to develop inventories representative of weekend condi-
tions. Second, traffic volume data in the historical record gener-
ally represent average daily traffic (ADT) or peak hourly traffic.
Hence, traffic count data must generally be obtained to generate
the diurnal traffic volume information required in the detailed
highway vehicle inventory, or else such data must be borrowed from
another urban area. (Temporal resolution is discussed in more de-
tail in Section 5.6.) Third, transportation planning data may not
reflect any strong seasonal component that may exist in a given
area. The local transportation planning agency should be consulted
to determine if the transportation data are representative of the
oxidant season and if not, what adjustments can be made.
Another limitation is that many of the data items required to
accurately develop highway vehicle emission factors are not readily
available from transportation planning data. EPA's MOBILE1, for
example, applies corrections based on vehicle speed, cold/hot
operating mode, vehicle mix and age, temperature, humidity, load,
trailer towing, and air conditioning. With the exception of speed
and cold/hot operating mode, little information on the other cor-
rection parameters is available from the transportation planning
process. The specification of those correction parameters from
transportation data is discussed in the next section.
Finally, differences may exist between the areas covered by
the photochemical modeling grid and the transportation planning
boundary. In most cases, the air quality analysis area is likely
5-10
-------
to be larger than the area used for transportation modeling. Where
this is the case, no traffic volumes will be available from the
transportation study for the outer areas. Fortunately, however,
because these outer areas are likely to be rural and to have very
little VMT, they probably contribute little to the overall highway
emission total. The best way to include them is to have transpor-
tation modelers code the outer area streets into the transportation
network and to add the new links to the historical record. The new
links cannot be used in the normal traffic assignment process but
traffic volumes can be manually added. Existing volumes can be
obtained from traffic counts, or if these are not available, can be
estimated by a transportation planner. For future year periods,
existing volumes can be factored. In rural areas, an annual growth
factor of 1 to 3 percent is probably appropriate unless significant
growth is expected during the forecast period. Local transpor-
tation planners should verify the factor used. In special cases
where major freeways run through the outer area, carrying significant
traffic volumes, traffic counts will likely be available from the
State Highway Department or the State Department of Transportation.
For future years, projected traffic volumes can be obtained by
using the growth factor of each freeway when in the transportation
area; where this would be inaccurate because of adjacent growth,
the State can usually provide an estimate based on historical
growth from traffic counts or from a statewide model.
5.5 EMISSION FACTOR MODELS
As shown in Figure 5-1, the emission factor model is a neces-
sary component in the detailed photochemical modeling inventory
process. Emission factors are multiplied by VMT in the case of
link-based models, as well as trip ends and number of vehicles in
hybrid models, to estimate highway vehicle emissions. Two EPA
highway vehicle emission factor models are currently available for
5-11
-------
use: (1) MOBILE1 and (2) Modal Model. The latter is not generally
recommended for use in conjunction with the detailed photochemical
modeling inventory; however, it is discussed briefly for those
agencies that desire the added accuracy that may result in certain
instances.
5.5.1 MOBILE1
In almost all cases, MOBILE1 is the most appropriate emission
factor model. While available as a separate program, MOBILE! is
also built into several available network emission calculation
models such as HWYEMIS1 and APRAC-II. (These network models are
discussed in the next section.) The basis and use of MOBILE1 are
documented in detail in Reference 4 and 5. The following inputs
are needed to run MOB ILEI:
• Speed
• Percent VMT of non-catalyst-equipped LDV, accumulated in
cold start mode
• Percent VMT of catalyst-equipped LDV, accumulated in cold
start mode
• Percent VMT of catalyst-equipped LDV, accumulated in hot
transient mode
• Ambient Temperature
• Vehicle Mix
• Mileage accrual by vehicle age
• Air conditioners in use
• Fraction of vehicles with an additional 500-1b loading
• Percent trailer towing
• Absolute humidity
• Additional heavy duty vehicle (HDV) correction •'"actors
Given these user supplied inputs, MOBILE1 computes composite
emission factors for three regions:
5-12
-------
• Low Altitude Regions
• California
• High Altitude Regions (>4,000 feet)
The six vehicle types for which composite emission factors are
calculated are:
• Light-Duty Vehicles (LDV)
• Light-Duty Gasoline-Powered Trucks, 0-6,000 Ibs GVW
(LDT1)
• Light-Duty Gasoline-Powered Trucks, 6,000-8,500 Ibs
GVW (LDT2)
• Heavy-Duty Gasoline-Powered Trucks (HDG)
• Heavy-Duty Diesel-Powered Vehicles (HDD)
• Motorcycles (MC)
The three pollutants for which MOBILE1 computes composite
emission factors are:
• Hydrocarbons (HC)
• Carbon Monoxide (CO)
• Oxides of Nitrogen (NOX)
MOBILE1 calculates composite emission factors for calendar
years 1970 through 1999. These factors account for current
deterioration rates as well as anticipated emission reductions due
to the Federal Motor Vehicle Control Program.
Several options are allowed in MOBILE1 to make it more
flexible. For instance, MOBILE1 includes options to allow for
various rates of emission control device deterioration in addition
to those assumed in the base case factors. Different timetables
for the implementation of new vehicle emission standards can be
reflected.
MOBILE1 allows the user to optionally apply inspection/main-
tenance (I/M) credits to emissions estimates, taking into account
5-13
-------
these factors. Emission reduction credit attributable to an annual
I/M program will vary according to the type of program in effect,
depending on (1) the stringency factor for determining the emission
level for passing or failing tested vehicles; (2) the calendar
year of interest and calendar year when I/M was first implemented;
(3) the presence of an adequate program of mechanic training; and
(4) subsequent years of I/M program operation.
MOBILE1 optionally outputs composite crankcase and evaporative
HC emission factors for each vehicle type, in addition to the total
HC emission factor (which includes the exhaust HC emissions). It
is important in the detailed photochemical modeling inventory to
maintain the distinction between exhaust and evaporative emissions
as the species composition of each of these emission components is
significantly different. MOBILE1 also has an option which allows
the user to generate either total or non-methane hydrocarbon
composite emission factors. The choice here in the photochemical
modeling inventory will depend on whether the "split factors"
available for apportioning VOC into classes have been adjusted to
apply to nonmethane or total VOC emissions. (This is discussed
further in Chapter 7.)
The most difficult part of using MOBILE1 in the compilation of
the detailed highway vehicle inventory is the specification of ap-
propriate values for the user inputs. The more important input
variables to specify accurately are speed, cold/hot start operating
percentages and temperature. Table 5-2 demonstrates the sensitiv-
ity of MOBILE 1 composite emission factors to changes in certain
input variables. Similar investigations regarding the importance
of cold start percentages indicate that this variable is as impor-
tant as the speed variable in affecting emission rates. The
approaches described in the following sections are recommended for
specifying values for the MOBILE1 input variables.
5-14
-------
Table 5-2. EFFECT OF SOME MOBILE! VARIABLES ON 1978 EMISSION FACTORS
Variable
Speed, mph
Temperature
Relative humidity
Barometric pressure
% LDV's using A/C
% LDV's towing trailers
% LDV's with excess load
% LDTl's with excess load
% LDTZ's with excess load
2
Heavy-duty truck load
Heavy-duty truck engine
Travel weight by vehicle
age
Variable Change Tested
Value A
15
30° F
60%
29.5 in.
of Hg
0%
0%
0%
0%
Q%
All Empty
All Small
MOBILE!
default
Value B
55
80° F
80%
30.5 in.
of Hg
48%
1%
15%
40%
75%
All loaded
All large
10% shift
to newer
vehicles
Percent Change in
Composite Emission
Factors3
NMHC1
-64
-22
NA4
NA
3
O5
1
0
0
5
6
-9
N0x
45
-27
-10
1
4
0
0
0
0
9
1
-2
Notes:
1. NMHC excludes crankcase blow-by.
2. Heavy-duty truck characteristics are extremes of range allowed by MOBILE!
3. Percent change = 100* (EF(B) - EF(A))/EF(A).
4. NA = not applicable
5. Percent change of zero indicates change less than 0.5%
Source: Reference 1
5-15
-------
5.5.1.1 Speed
Two sets of vehicle speeds are commonly included for each link
in the historical record. They are referred to most often as input
speeds and output speeds. Input speeds are the speeds coded
directly into the link description (see Table 5-1, columns 19-21,
42-44). These are often posted speed limits and do not necessarily
reflect actual driving speed. In other cases, the coded speed may
be based on field surveys of actual driving time. Field surveys,
when complete, reflect current driving speeds. They do not,
however, reflect future year driving speeds because of the effect
of increased traffic volumes on driving speed.
Output speeds are those speeds resulting from a capacity re-
straint traffic assignment or manual adjustment. The capacity re-
straint program compares vehicular capacity of a link to assigned
volumes. Depending on the volume to capacity ratio, the speed is
lowered proportionately on the overloaded link to discourage its
use as a trip path between certain zonal pairs. With this process,
output speeds are always lower than input speeds if there is any
congestion in the network. They may, in fact, be too low on cer-
tain links and may show illogical route continuity. This is be-
cause the speed factor is actually an impedance factor that is
mechanically manipulated and does not necessarily correspond to
reality. On certain links, after continued application of the ca-
pacity restraint process, speeds are often below 5 mph with no
apparent justification. In other cases, speeds on individual links
are manually adjusted by traffic analysts during the network cali-
bration process. Again, these adjustments are made to mechanically
change the link impedance, and speeds may not correspond to driving
conditions. For these reasons and contrary to frequent practice,
neither the input nor the output speeds should be used as input to
MOBILE1.
5-16
-------
The best available method for determining speeds at this level
of analysis is to use program SPEED. This program will soon be
added to the BACKPAC battery.* The purpose of SPEED is to estimate
individual link speeds from link characteristics (facility type,
area type), degree of congestion (volume/practical capacity rela-
tionship), and volume capacity versus speed curves (user supplied
or program specified). SPEED reads, updates, and copies the tradi-
tional historical record file with the insertion of an additional
link speed. The link speed can later be used by either of the
recommended network emission models. A similar program is avail-
able with the UTPS system within UROAD.
Even after using SPEED or a similar program, the calculated
speeds may differ by as much as 25 percent or more from actual
speeds because of the difficulty in quantifying the volume/cap-
acity/speed relationship and the difficulty of calculating capa-
city. Still, these speeds are the best available and, on a region-
wide basis, probably cause no errors of an unacceptable magnitude.
5.5.1.2 Cold/Hot Mode Vehicle Operation
MOBILE1 requires that the user define the three cold/hot start
percentages shown below:
Percent of non-catalyst-equipped LDV VMT accumulated in cold
start mode (PCCO).
Percent of catalyst-equipped LDV VMT accumulated in cold start
mode (PCCC).
Percent of catalyst-equipped LDV VMT accumulated in hot start
mode (PCHS).
*Program SPEED is available from: Chief, Urban Division, Federal
Highway Administration, U.S. Department of Transportation,
Washington, D.C. 20590.
5-17
-------
Several options are available to develop these percentages.
The simplest course is to assume single, fixed percentages for all
speeds, locations, times, etc. Alternately, if the link-based
network emission models HWYEMIS1 or APRAC-2 (these are described in
the next section) are used, a subroutine automatically calculates
these percentages based on assumed values (Table 5-3) expressed as
a function of locale and time during the day. Area types and land
use classifications coded into the historical record are combined
into the locales shown in Table 5-3. Unfortunately, not all cities
will have coded these fields on the historical record, or if they
have been coded, they may not correspond to those accepted by
HWYEMIS1 and APRAC-2. If the area and land use type by link are
not in the historical record or do not correspond to those accepted
by the models, the only way to derive them is through a cumbersome
manual procedure whereby the proper classification is added to each
record in the historical record. This latter alternative will be
resource intensive and should be carefully evaluated prior to
implementation.
Locale-specific cold/hot start percentages can be derived
using a procedure developed for FHWA.^ A detailed description of
this procedure is too lengthy to include here. Basically, origin/
destination survey data, or other suitable transportation planning
sources, are used to determine the cold and hot start fractions as
well as the fraction of travel represented, both by trip purpose
and by hour. Then, using trip distribution data with terminal
times included, the transient mode by purpose is determined.
Finally, the contributions of each trip purpose to the cold/hot
mode fractions are calculated for both catalyst and non-catalyst
equipped cars. Theoretically, cold/hot percentages can be calcu-
5-18
-------
Table 5-3. ASSUMED PERCENTAGES OF COLD AND HOT TRANSIENT
VEHICLE OPERATION IN APRAC-2 AND HWYEMIS1
Time
Interval
Local Time
1800 - 0700
0700 - 0900
1000 - 1600
1600 - 1800
Locale
Downtown
(%)
5 C
10 HT
0 C
0 HT
0 C
30 HT
60 C
20 HT
Suburban
Commercial
(%)
0 C
10 HT
0 C
0 HT
0 C
10 HT
0 C
0 HT
Residential
and Other
(%)
50 C
0 HT
80 C
0 HT
10 C
70 HT
10 C
10 HT
Source: Reference 6
5-19
-------
lated for each zone and each hour by this procedure. In practice,
however, it is recommended that percentages be calculated for zonal
aggregates and for longer time intervals to minimize the amount of
work involved. The resulting values can be substituted for the
percentages incorporated in HWYEMIS1 and APRAC-2.
5.5.1.3 Ambient Temperature
When MOBILE1 is run, an ambient temperature must be input for
each composite emission factor calculated. When HWYEMIS1 is used,
incorporating MOBILE1, a single temperature is input for the period
of interest. When APRAC-2 is employed, hourly meteorological data
are input that include ambient temperature. In either case, appro-
priate temperatures should be chosen that are representative of
conditions on the typical day being modeled.
5.5.1.4 Vehicle Mix
Vehicle mix refers to the fraction of each type of highway
vehicle in the entire vehicle fleet. A default option exists to
the user of MOBILE1 to use the following fractions for the six ve-
hicle types (see Section 5.5.1) if locale-specific data are not
available: 0.803, 0.058, 0.058, 0.045, 0.031, and 0.005.
Data on vehicle type are available from two sources in the
transportation study process. The first source of data is the on-
going traffic count program. Vehicle classification counts are
usually taken at a few locations each year. These data are
sometimes of limited benefit for at least three reasons:
1. An adequate breakdown of truck types is not usually avail-
able from these counts.
2. The percentage of vehicle types on a street at a given
location does not indicate the relative fraction of total
VMT contributed by that vehicle type, unless the overly
5-20
-------
simplified assumption is made that all vehicle types
travel the same distance each day.
3. Counts are highly variable by location; no adequate
measure is available to determine what the areawide
fractions should be.
A second source of vehicle mix data is available from trans-
poration studies. In larger cities, truck surveys were made as
part of the original inventory phase. In these cities, a truck
trip table is developed (sometimes taxis are added) for the base
year in much the same manner as the purpose trip tables. It is
added to the purpose trip tables when the purpose trip tables are
merged after trip distribution. The truck trip table can be as-
signed separately before the trip tables are merged. The assign-
ment process is done with the standard assignment programs, i.e.,
LOADVN with the PLANPAC/BACKPAC series and UROAD with the UTPS sys-
tem. When the truck assignment is compared to an assignment with
all trip purposes, the percentage of trucks by subarea is readily
apparent. This process is of sufficient difficulty to make it in-
advisable to pursue in most cases. Even after the process, the
percentage of trucks must still be broken down into the four
MOBILE1 truck categories.
The third and most usable approach is to obtain data on ve-
hicle registration from the state. These data are usually avail-
able on a county level and can be manipulated into the six vehicle
type categories. Again, the assumption required is that all
vehicle types travel the same distance each day.
5.5.1.5 Mileage Accrual By Vehicle Age
MOBILE1 takes into account vehicle age and mileage accumulated
by each model year when calculating composite emission factors. A
5-21
-------
set of default values are built into MOBILE1 for the six vehicle
types described previously and for model years 1970-1999. Alter-
nately, the user can supply vehicle registration data or mileage
accrual by vehicle age, in which case MOBILE1 uses this information
to calculate emission factors. Generally, many states will have
registration data for use in MOBILE1 but fewer have data on mileage
accrual by vehicle age.
5.5.1.6 Air Conditioning, Heavy Load, Trailer Towing
Sensitivity studies summarized in Table 5-2 indicate these
values affect emission rates very little. Local data on these
factors are almost never available; however, since the impact is
small, reasonable values can be estimated for the summer season
without incurring significant error.
5.5.1.7 Absolute Humidity
Humidity is important in determining the NOx emission rate.
The MOBILE1 default value is 75 grains per pound of dry air. Since
NOX is sensitive to humidity, local meteorological records should
be consulted to obtain a more accurate figure for the summer season.
5.5.1.8 Additional HDV Correction Factors
Parameters for additional HDV correction factors are:
• Average gross vehicle weight (GVW) for heavy-duty gasoline
(HDG) vehicles
• Average GVW for heavy-duty diesel (HDD) vehicles
• Average cubic inch displacement for HDG vehicles
• Average cubic inch displacement for HDD vehicles
These parameters have little effect on emission factors and
data on them are almost impossible to obtain. Default values are
recommended.
5-22
-------
5.5.2 MODAL MODEL
Evidence suggests that EPA's Modal Model may, in certain
instances, yield more accurate estimates of highway vehicle emis-
sions than MOBILE1.8 However, the usefulness of the Modal Model
for a regional analysis is limited for several reasons. First, it
is designed to estimate only light duty vehicle (LDV) exhaust emis-
sions. The planner must compute truck emissions and evaporative
emissions by other means. Second, the model accurately predicts
the emissions of LDV's operating in a stabilized mode, but not of
those in a startup mode. Third, emissions for calendar years after
1977 are not yet available. Finally, and most importantly, the
model requires vehicle speed profile data at one-second intervals.
Such detailed speed information aids in developing a more accurate
emission factor, since it accounts for acceleration, deceleration,
idling, and free-flow operations. Such detailed speed information,
however, is rarely available even for a particular link, much less
for all links within an urban region or subarea. To employ this
model in practice requires that each link be characterized by one
or more speed curves. The difficulty in assigning speed curves to
each link, together with the above- mentionedlimitations, generally
precludes the use of this model in compiling photochemical modeling
inventories.
5.5.3 COMPUTING S02 AND PARTICULATE EMISSIONS
Neither MOBILE1 nor the Modal Model calculates emission
factors for S02 or particulate. Average emission factors for
these pollutants, available from Supplement 5 of AP-42^, can be
used to complete the highway vehicle emission inventory.
5-23
-------
5.6 NETWORK EMISSION MODELS
The hub of the highway vehicle emission calculation process is
the network emission model. As shown in Figure 5-1, the network
emission model combines the transportation data and emission fac-
tors to compute highway vehicle emissions at the link level.
As mentioned in Section 5.3, two types of network models can
be employed for developing highway vehicle emission estimates: (1)
link based and (2) hybrid models. The use of each of these two
types of network model is discussed below. Emphasis is placed on
link based models as no standard hybrid models are currently avail-
able for general use.
5.6.1 LINK BASED NETWORK EMISSIONS MODELS
The two network emission models recommended for use are
HWYEMIS110 and APRAC-2.6 With the HWYEMIS1 program, link
speeds and link data are taken from the historical record and, with
MOBILE1 input parameters, emission factors are produced on a link-
by-link basis. The program produces a file to interface with
SAPLSM, an FHWA program that sums link emissions by grid cell. The
file can also be interfaced with other emission gridding programs,
as discussed in the next section.
APRAC-2 was originally written containing the AP-42 Supplement
5 mobile source emission factors; the program has recently been
modified to include MOBILE1 emission factors. This model has two
major components: (1) an emission module called EMOD and (2) a
diffusion module call DIFMOD. Only the EMOD module is of use in
this context. The EMOD module works much the same as HWYEMIS1,
i.e., historical record data and MOBILE1 data are combined to pro-
duce link-by-link emission factors. Link emissions are summarized
by user-specified grids for as many as 625 grid cells.
5-24
-------
Of the two network emission models recommended, HWYEMIS1 is
less data intensive than APRAC-2, is easier to use, and interfaces
easily with other DOT programs. At the same time, it has certain
disadvantages when compared to APRAC-2. These disadvantages are
chiefly related to its handling of traffic data and cold start per-
centages. APRAC-2 excels in these two areas because of (1) the
ability to vary traffic by hour, by locale, by direction; (2)
because cold start percentages can be input directly and a new set
of emission factors are created for each hour; and (3) because the
ability to change traffic and emission factors (including the tem-
perature) by hour allows a more realistic simulation of actual con-
ditions over a multihour period.
SAPOLLUT is another program that has been used as a
network emission model. HWYEMIS1 is an update of this model
incorporating MOBILE1 rather than the obsolete Supplement 5
emission factors. Although HWYEMIS1 has lost some of the
flexibility of SAPOLLUT, on balance, it is more desirable to use
HWYEMIS1 than obtain the SAPOLLUT program and insert MOBILE1 or
otherwise modify the program. Specific details on HWYEMIS1 and
APRAC-2 are given in the following section.
5.6.1.1 HWYEMIS1
Programs and documentation for HWYEMIS1 are available from:
U.S. Department of Transportation
Federal Highway Administration (HHP-23)
Urban Planning Division
Washington, DC 20590
It can be run with either the PLANPAC/BACKPAC or UTPS program
batteries. The necessary input data to HWYEMIS1 are:
5-25
-------
• Historical record data containing:
network description
speed by 1 ink
traffic volume by link
area type
land use type
facility type
• Time interval of the analysis
• Percentage of ADT occuring during time interval of analysis
• MOBILE1 inputs
The network description coded on the historical record is de-
scribed in Section 5.4 on transportation planning data.
HWYEMIS1 reads ADT (i.e., 24-hour traffic volume) from the
historical record and factors it into four time intervals:
1800-0700, 0700-0900, 0900-1600, and 1600-1800. These time inter-
vals are used in connection with the cold start percentages (see
Table 5-3). Optionally, these can be manipulated in combination
with the percentage of ADT to produce any time period.
Temporal factors for distributing ADT from this historical re-
cord into hourly or other time increments can be derived from
several sources. Generally, local traffic count data yield the
most acceptable traffic distributions. To use this kind of data,
counts from across the area must be combined in some manner to re-
present an average hourly distribution for the entire analysis
area, a specific link type, or a subarea, depending on model re-
quirements. Various methods are available. The method to be used
depends on the variations in distributions and whether the data are
on a computer in an accessible form. If computer access is avail-
able, the most accurate method is to develop a weighted average
distribution based on all counts in a given category. For example,
all adjusted north-bound CBD expressway links can be stratified and
a weighted average developed. In other cases where this capability
5-26
-------
is not available, manual techniques can be used. One technique is
to draw north-south and east-west screenlines across the community
or the subarea desired and to use the closest count for all streets
bisected by the screenline. Another method, less exact, is to
select counts at random. Still another is for an experienced
traffic analyst to select typical counts. Regardless of the pro-
cedure used, all distributions should be examined by an experienced
traffic professional for reasonableness. While city-specific data
sources as described above are preferable, in some cases it may be
necesssary to use published data on hourly distribution of traffic.
Examples appear in Tables 5-4 and 5-5.
In HWYEMIS1, link speeds are an output of program SPEED or its
equivalent in UROAD, which were discussed in Section 5.5. Area and
land use types are used in connection with determining cold/hot
start percentages, as discussed in Section 5.5.1.2. Facility type
is used only to output a summary.
MOBILE1 inputs were treated in a previous section. The
HWYEMIS1 program documentation gives instructions on how to input
the necessary parameters when MOBILE1 is not run as an internal
program within HWYEMIS1.
It should be noted that HWYEMIS1 does not compute intrazonal
VMT related emissions since only VMT for interzonal links and cen-
troids are generally coded on the historical record. Travel emis-
sions for intrazonal trips may be included in the network emission
model by inserting dummy links into each zone, or the number of in-
trazonal trips per zone may be multiplied by an average trip dis-
tance to develop a VMT figure (as discussed in Section 5.4) which
is then multiplied by an emission factor. An emission factor is
developed with MOBILE1. The only input modification from the
MOB I LEI parameters used in the network emission model is speed. A
speed commonly used, although undocumented, is 13 mph.
5-27
-------
Table 5-4. HOURLY DISTRIBUTIONS OF AVERAGE DAILY TRAFFIC FOR
LARGE CITIES AS A PERCENTAGE OF THE TOTAL
(Percent ADT for cities with >500,000 population)
Hour
0
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
Freeways
CBD
1.5
1.0
0.5
0.5
1.0
2.0
5.5
8.5
7.0
4.5
4.0
4.5
4.5
4.5
5.5
7.5
9.5
8.0
5.0
4.0
3.5
3.0
2.5
2.0
Central
City
1.5
1.0
0.5
0.5
0.5
1.5
4.5
7.5
6.5
5.0
4.5
4.5
4.5
5.0
6.0
7.5
9.0
8.5
5.5
4.0
3.5
3.5
3.0
2.0
Suburban
1.5
1.0
0.5
0.5
1.0
1.5
4.5
8.5
6.5
5.0
5.0
4.5
4.5
4.5
5.5
7.0
8.5
8.5
5.5
4.5
3.5
3.0
2.5
2.5
Arterial s
CBD
1.5
1.0
1.0
0.5
0.5
1.5
3.5
6.5
6.5
5.0
5.5
5.5
5.5
6.0
6.5
7.0
8.5
7.5
4.5
4.0
3.5
3.0
3.0
2.5
Central
City
1.5
1.0
0.5
0.5
0.5
1.0
4.5
8.0
6.0
4.5
4.5
5.0
5.0
5.0
5.5
7.0
9.0
8.0
5.5
5.0
4.0
3.5
3.0
2.0
Suburban
1.5
0.5
0.5
0.5
0.5
1.0
4.0
7.5
5.5
4.5
4.5
5.0
5.0
5.0
6.0
7.0
8.5
8.5
6.0
5.5
4.5
3.5
3.0
2.0
Source: Reference 11
5-28
-------
Table 5-5. HOURLY DISTRIBUTIONS OF DIRECTIONAL SPLIT FOR LARGE
CITIES AS A PERCENTAGE OF HOURLY VOLUME
(Directional split, inbound, for cities with >500,000 population)
Hour
0
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
Freeways
CBD
44
44
50
52
58
66
66
60
58
54
48
48
48
50
52
44
38
40
44
46
50
52
52
50
Central
city
38
40
40
46
56
64
70
70
68
62
58
52
52
52
50
46
38
38
46
52
46
42
42
40
Suburban
44
46
48
54
60
68
68
64
58
54
52
50
50
52
52
48
42
40
44
48
48
44
46
44
Arterial s
CBD
48
48
46
48
54
64
62
62
62
58
54
54
52
52
52
48
44
40
46
50
50
48
48
50
Central
city
44
46
44
48
54
62
66
68
64
56
54
52
50
50
50
46
40
38
46
52
48
46
46
46
Suburban
40
42
44
48
58
66
72
68
60
56
54
50
50
50
50
46
40
38
46
50
46
44
44
44
Source: Reference 11
5-29
-------
5.6.1.2 APRAC-2
The APRAC-2 program and documentation are available from:
U.S. Environmental Protection Agency
Region IX
215 Fremont Street
San Francisco, California 94108
APRAC-2 continues to be revised by various organizations, so
the user should be sure to obtain the latest version. Early ver-
sions contained Supplement 5 emission factors and several program
errors. APRAC-2 is considerably more data intensive than HWYEMIS1.
Data required are:
o Historical record data containing:
network description
speed by link (as an option, this can be computed with
the volume/capacity method by the program)
traffic volume
land use type
area type
facility type
capacity, if speed is to be calculated by the program
o Secondary traffic data
o Ten sets of hourly traffic distributions for weekday
traffic
o Seasonal traffic adjustment
o MOBILE1 input
Historical record data needed for APRAC-2 are similar to those
needed for HWYEMIS1, with two exceptions. APRAC-2 can either use
speeds from the historical record (program SPEED or the UROAD
equivalent) or compute link speeds within the program using a
volume/ capacity relationship. If the program option to compute
link speed is used, capacity and free-flow speeds by facility type
and locale are required. The facility types needed in APRAC-2 do
not correspond to common FHWA indicators.
5-30
-------
APRAC-2 incorporates the concept of secondary traffic, which
is synonymous with intrazonal travel. If the user cannot calculate
intrazonal VMT by the procedure outlined in the previous section on
HWYEMIS1, APRAC-2 allows specification of the amount of secondary
traffic in either of two ways. The first method allows specifica-
tion of secondary traffic as a percentage of primary traffic
(traffic on links). The second procedure allows specification of
secondary traffic by locale. Both methods required an assumed
regionwide ratio of primary to secondary traffic. The user should
consult the APRAC-2 documentation for more detail on its handling
of secondary traffic.
In APRAC-2, extensive data on hourly distributions, by direc-
tion and by locale, are required. Specifically, directional dis-
tributions for each locale are required. General procedures for
developing hourly traffic distributions were included in the pre-
vious section on HWYEMIS1. Professional judgment of a traffic
analyst is needed to develop these inputs.
A seasonal traffic adjustment figure is also an essential in-
put to APRAC-2. Again, a traffic analyst should develop this re-
gionwide figure. The adjustment figure should be based on monthly
counts and weighted toward the area with highest traffic volumes.
This value rarely exceeds 10 to 15 percent.
The MOBILE1 input parameters are the same-as discussed for
HWYEMIS1.
5.6.2 HYBRID MODELS
As mentioned previously, the hybrid approach, wherein trip-
end and vehicle related emissions are calculated separately from
travel related emissions, is potentially more precise than the link-
based approach. To date, however, no standardized hybrid models
are available for use in calculating network emissions. Some
groups are developing their own hybrid models whereas others have
5-31
-------
modified SAPOLLUT to this end. EPA and DOT are jointly developing
a hybrid model that will directly interface with the UTPS and
PLANPAC/BACKPAC transportation batteries. Because no standard hy-
brid models are currently available, the concepts involved will be
discussed only briefly here.
If the hybrid approach is used, only the travel related (i.e.,
hot stabilized) emissions from interzonal links and centroid links
are considered in HWYEMIS1 or APRAC-2. If dummy links or the sec-
ondary traffic procedure in APRAC-2 are used, stabilized emissions
from intrazonals may also be included. The remaining emissions to
be included in the highway vehicle inventory are:
• Trip-end emissions - cold/hot transient starts
• Trip-end emissions - hot soaks, for LDV's only
• Vehicle related emissions
• Intrazonal trips (if not added earlier) - travel, trip, and
vehicle related emissions
The basic process is simplistically described in the following
paragraphs. For cold/hot transient starts, auto trip origins are
determined by zone and by purpose for the desired time increments.
Then, using tables in the FHWA report? (discussed in Section
5.5.1.2), the number of cold/hot transient starts are calculated by
trip purpose. Finally, appropriate emission factors, representing
excess emissions over hot stabilized operation, are multiplied by
the number of cold/hot transient starts to estimate startup emis-
sion by zone. Examples of such factors, expressed as excess start-
up emissions in units of grams per trip, are presented in Reference
12.
Emissions from hot soaks are calculated for autos only. As
with the startup component, truck trip-end related emissions are
incorporated in the truck travel related emission rates. The re-
quired data input is the number of trip destinations in each zone.
The number of destinations by zone is available from the origin/
5-32
-------
destination trip table used in the cold start analysis (Section
5.5.1.2). Because hot soaks are usually thought to occur within
one hour of engine shutdown, all destinations are usually consid-
ered to entail a full hot soak. Each hot soak is multiplied by the
appropriate emission factor to estimate hot soak emissions per
zone. Examples of hot soak factors, in units of grams per trip,
are also shown in Reference 12.
Vehicle related (diurnal) emissions are simply a function of
the number of vehicles in the study area. The number of vehicles
is best obtained from state registration data. Once the total num-
ber of vehicles for the analysis area is known, emission factors
such as those presented in Reference 12 (in terms of grams per day
per vehicle) are used to produce total vehicle related (diurnal)
emissions.
If travel related emissions for intrazonal trips are not in-
cluded in the network emission model by inserting dummy links into
each zone, they must be added in this step. The total number of
intrazonal trips in each zone is multiplied by an average trip dis-
tance, which can be zone specific if the computational capacity is
available. An emission factor is developed using MOBILE1.
To calculate emissions from intrazonal starts, the number of
trip origins in each zone are needed along with cold/hot transient
start percentages and emission factors representing excess emis-
sions per cold/hot start. The number of trip origins is known from
the previous step. Cold/hot transient start percentages must be
developed by the analyst. No good sources of this data are avail-
able because the intrazonals are, by definition, short trips. The
percentage would be expected to be higher than those specified in
Table 5-3.
5-33
-------
Intrazonal hot soak emissions are determined, in the hybrid
approach, by multiplying appropriate hot soak emission factors by
the number of trip destinations. By definition, for intrazorial
trips, the number of trip destinations in a zone equals the number
of trip origins; hence, the number of intrazonal trip origins may
be used, as determined above.
Vehicle related (diurnal) emissions from intrazonal trips need
not be accounted for separately as these emissions will be included
in the calculation based on areawide vehicle registrations.
5.7 HIGHWAY VEHICLE EMISSION GRIPPING PROCEDURE
Referring again to Figure 5-1, the last step necessary prior
to completing the detailed highway vehicle emission inventory is to
convert all link emissions to the grid system to be used in the
photochemical model.
Because link-based and hybrid network models compute highway
vehicle emissions differently, some procedural differences will
exist in this link/zone-to-grid-cell conversion step. Each type of
model is discussed separately in the following sections.
5.7.1 GRIDDING PROCEDURES FOR LINK-BASED MODELS
Currently, neither the PLANPAC/BACKPAC nor UTPS program bat-
teries contains adequate programs to convert link emissions to
gridded emissions. SAPLSM (PLANPAC/BACKPAC) and UROAD (UTPS) are
the available gridding programs but, in both cases, no more than 50
grid cells can be managed. APRAC-2 can output emissions for 625
grid cells. This may be adequate spatial resolution in some appli-
cations, but a greater number of cells are required in most cases.
Thus, the agency responsible for compiling the inventory will have
to develop this programming capability. Several MPO's have written
5-34
-------
programs that provide this function. All programs follow the same
basic principles. Link emissions are apportioned to grid squares
by assuming that the emissions are uniformly distributed along a
straight line. The fraction of the line segment that falls within
a given grid square determines the fraction of the total emissions
for the link that fall in that square. The contributions of all
links are summed for each square.
There are two basic requirements for the conversion of link
emissions to gridded emissions. The first is a grid system with
coordinates; the second is a compatible coordinate system for the
links. In most areas, air quality agencies compile emission
inventory data with UTM coordinates. Traffic studies, while often
having geographical coordinates coded onto the historical record,
seldom use UTM coordinates. Therefore, in most cases, the two
coordinate systems are incompatible. A specially written program
is required for conversion to UTM coordinates. As with the grid-
ding program, several agencies have developed their own programs
which can be modified to individual system requirements. With
compatible coordinate systems, a preprocessor program is required
that determines the percent of each link included in each grid
cell.
Intrazonal emissions, if not determined by coding dummy links
on the historical record, will be calculated at the zonal level, as
described in previous sections. In this instance, a zone-to-grid-
cell conversion program is necessary. Again, as with the link-to-
grid-cell conversion, no standard programs are available that are
compatible with the UTPS and PLANPAC/BACKPAC batteries. Hence, the
agency will have to develop its own programs. Several local agen-
cies have done this.
5-35
-------
5.7.2 GRIDDING PROCEDURES FOR HYBRID MODELS
The travel related (hot stabilized) emissions calculated by
hybrid models for links are assigned to grid cells in the same man-
ner as in link-based models. Likewise, the travel related emis-
sions associated with intrazonal VMT are also handled similarly.
The major difference is that trip-end emissions (in Section 5.6.2)
yield emissions for each zone. Hence, the zone-to-grid-cell
conversion procedures described above can be used for this
conversion.
The procedure outlined in Section 5.6.2 for computing vehicle
related (diurnal) emissions results in total emissions for the
study area. The problem is then one of allocation. Theoretically,
vehicle emissions occur throughout the time period during which the
temperature is rising, usually considered to be 6:00 a.m. to 3:00
p.m. The emissions occur regardless of whether the vehicle is *
parked or running. Therefore, these emissions should be spatially
allocated to zone, according to the vehicle minutes spent in each
traffic zone. Ordinarily, however, available data are not
sufficient to determine this spatial distribution. Parking data
may be used, but they may be of questionable accuracy and rele-
vance. Lacking better information, a practical approach is to use
either VMT or travel related emissions as a surrogate indicator for
gridding.
5.8 TRANSPORTATION CONTROL MEASURE ANALYSIS
An important part of the photochemical modeling effort is the
evaluation of alternate control strategies. For this purpose, pro-
jection inventories must be compiled for each proposed set of con-
trol measures. Table 5-6 shows examples of various transportation
control measures that may be considered for reducing highway ve-
5-36
-------
Table 5-6. EXAMPLE TRANSPORTATION CONTROL MEASURES
Vehicle inspection and maintenance
Vapor recovery from fuel transfer
and storage and from solvent
operations
Improved public transit
Exclusive bus and carpool lanes
Areawide carpool programs
Restrictions on private automobile
use
Long-range transit improvements
On-street parking controls
Park-and-ride and fringe parking
lots
Pedestrian malls
Employee programs to encourage car-
and vanpooling, mass transit use,
bicycling, and walking
Bicycle lanes and storage facilities
Staggered work hours
Road pricing to discourage single
occupancy automobile use
Controls on extended vehicle idling
Traffic flow improvements
Fleet vehicle controls including
alternative fuels and engine use
Vehicle retrofits for other than
light duty vehicles
Extreme cold start emission reduction
programs
Source: Workshop on Requirements for Nonattainment Area Plans, Compilation
of Presentations, Revised Edition, U.S. Environmental Protection Agency,
Research Triangle Park, North Carolina, April 1978.
5-37
-------
hide emissions. There are two basic types of control strategies
involving highway vehicles: (1) those that can be simulated by
making emission factor changes (e.g., in MOBILE1) and (2) those
that result in traffic pattern changes, i.e., changes in the mag-
nitude of VMT, spatial and temporal VMT patterns, etc.
Measures related to emission factor changes are easily accom-
modated by the processes described in this chapter. For example,
MOBILE1 contains a provision for considering the I/M control pro-
gram. The user specifies a year, the stringency of I/M mechanics'
training, and the model years to which I/M applies. When running
the emissions model, which in this case is contained within the
network emission model (HWYEMIS1 or APRAC-2), MOBILE1 parameters
that include I/M are specified and the procedure continues as de-
scribed. As another example, the composite emission factors pro-
jected by MOBILE1 automatically incorporate the emission reductions
mandated by the Federal Motor Vehicle Control Program; however,
alternate rates of implementation of these standards can be
considered.
Measures resulting in traffic pattern changes are much more
difficult to simulate. The basic problem is one of changing link
volumes from the transportation models to reflect results of the
control measures, yielding new link volumes which are then input to
the network emission model. Methods to estimate the changes in
link volumes vary widely in both approach and complexity. Some
measures can be simulated within the PLANPAC/BACKPAC and DTPS
program batteries. These measures include such things as traffic
flow improvements and pedestrian malls (auto-free zones). The
basic approach is to change the link description. For example, in
the case of traffic flow improvement that consists of street widen-
ing, the capacity of the link on the roadway is increased. In the
traffic assignment process, the volume/capacity relationship is
5-38
-------
changed, increased volume is assigned to the link, and the speed is
recalculated. When input to the network emission model, a new link
volume and a new link speed are incorporated; the change in link
speed results in a revised emission factor.
Other measures resulting in traffic pattern changes cannot be
simulated with the PLANPAC/BACKPAC and UTPS batteries. These are
the measures that require a modal choice such as transit use, car-
pooling, bicycling, and walking. While the UTPS system does con-
tain some transit choice capability, policy-sensitive disaggregate
demand models ' ' are required in most cases. Demand
levels between alternate choices are estimated by the disaggregate
models, and volumes and speeds input to the network emission model
are changed accordingly. These disaggregate models represent the
state-of-the-art method for analyzing such measures as CBD parking
surcharges, carpool programs, park-and-ride facilities, etc. For
more information on the simulation of specific transportation
control measures, the reader should consult References 12 through
14, and numerous other publications on the subject.
5-39
-------
References for Chapter 5.0
1. Ossi, J. A., and J. L. Horowitz. Data Requirements for
Estimation of Highway Emissions. Draft. U.S. Environmental
Protection Agency, Washington, D.C. and Massachusetts
Institute of Technology, Cambridge, Massachussetts. 1979.
2. Computer Programs for Urban Transportation Planning, PLANPAC/-
BACKPAC, General Information. U.S. Department of Transporta-
tion, Federal Highway Administration, Washington, D.C. April
1977.
3. Urban Transportation Planning System, Introduction. U.S.
Department of Transportation, Urban Mass Transportation Admin-
istration and Federal Highway Administration, Washington, D.C.
June 1977.
4. Users Guide to MOBILE1: Mobile Source Emissions Model. U.S.
Environmental Protection Agency, Washington, D.C. Publication
Number EPA-400/9-78-007. August 1978.
5. Mobile Source Emission Factors. U.S. Environmental Protection
Agency, Washington, D.C. Publication Number EPA 400/9-78-005.
March 1978.
6. Users Manual for the APRAC-2 Emissions and Diffusion Model.
Stanford Research Institute, Menlo Park, California. June
1977. Pp. 7, 51-53.
7. Ellis, G. W., et al. Determination of Vehicular Cold and Hot
Operating Fractions for Estimating Highway Emissions"U.S.
Department of Transportation, Federal Highway Administration,
Washington, D.C. September 1978.
8. Richard Pollack, Terrence W. Austin. Comparison of Two
Methods for Predicting Corridor-Based Automotive Emissions.
Proceedings, Emission Factors and Inventories, West Coast
Section, APCA. November 1978.
9. Compilation of Air Pollution Emission Factors, Supplement 5,
AP-42. Third Edition. U.S. Environmental Protection Agency,
Office of Air Quality Planning and Standards, Research Triangle
Park, N.C.
5-40
-------
10. HWYEMIS1 Program Documentation, U.S. Department of Transporta-
tion, Federal Highway Administration, Urban Planning Division,
Washington, D.C. February 1979.
11. Final Manual. Special Area Analysis. U.S. Department of
Transportation, Washington, D.C. August 1973. Pp. 85-87.
12. How to Prepare the Transportation Portion of Your State Air
Quality Implementation Plan. U.S. Department of Transporta-
tion, Federal Highway Administration, Washington, D.C. 1979.
13. Richard H. Pratt, et al. Traveler Response to Transportation
System Changes: A Handbook for Transportation Planners.
Prepared for the U.S. Department of Transportation, Federal
Highway Administration, Washington, D.C. February 1977.
14. Alan M. Voorhes, Inc. Transportation System Management: An
Assessment of Impacts. Prepared for U.S. Department of
Transportation, Office of Policy and Program Development,
Urban Mass Transportation Administration, Washington, D.C.
November 1978.
5-41
-------
6.0 AREA SOURCES
6.1 GENERAL
Except for highway vehicle emissions, the basic annual county-
wide inventory is a good starting point for developing the detailed
photochemical modeling inventory for most other area sources. Gen-
erally, the basic inventory should contain activity level and emis-
sion estimates for many of the following area sources of VOC or NOX:
• Ai reraft
• Railroads
• Vessels
• Other Off-Highway Vehicles (farm tractors, construction
equipment, lawn mowers, etc.)
• Gasoline Handling
• Drycleaning
• Degreasing
• Nonindustrial Surface Coating
• Cutback Asphalt
• Pesticides Applications
• Miscellaneous Solvent Use
• Oil and Gas Wells
• Onsite Incineration
• Open Burning
• Residential Fuel Use
• Commercial/Institutional Fuel Use
• Forest Fires
• Slash and Agricultural Field Burning
• Structural Fires
Of course some of the above source categories, as was discussed
in Volume I, may be treated as point sources in the basic inventory
and some may be handled as both point and area sources depending on
6-1
-------
the cutoff level used for making this distinction. Likewise, a num-
ber of other source categories that traditionally are considered to
be point sources may, at least in part, be treated as area sources.
Industrial fuel use, industrial surface coating, and gasoline bulk
tanks are examples of this type of source category. The inventory-
ing agency should be aware of all such distinctions and may need to
institute certain changes prior to completing the detailed photo-
chemical modeling effort. For example, if the agency needs to make
a detailed analysis of the effect of controlling drycleaning opera-
tions, it may prefer to treat each plant as a point source rather
than collectively as part of an area source, as this would allow
distinct control measures to be evaluated for each plant. Conversely,
if drycleaning plants were treated as point sources in the existing
basic inventory, but no specific information was available on growth
at specific locations, the agency might wish to treat drycleaning as
an area source in the detailed inventory. It is generally recom-
mended, however, that the point/area source distinctions made in the
basic inventory be utilized in the detailed inventory to minimize
additional resources needed.
In the basic inventory, emissions a~e generally determined over
a broad area, such as each county within an urban area. A primary
reason for this is that estimates of the activity levels used in
calculating area source emissions are generally available at the
county level. Moreover, these estimates typically represent annual,
or, in some cases, seasonal emissions, and do not distinguish between
different reactive classes of VOC and NOX. These area source
estimates can be used in the photochemical modeling inventory; how-
ever, as with point sources, several additional steps need to be
taken by the inventorying agency in order to provide the spatial,
temporal, and species resolution required of the detailed inventory.
First, the countywide emission estimates for area sources must be
allocated to the grid cell level. Second, hour-by-hour emission
estimates must be developed for a typical weekday during the oxidant
6-2
-------
season. Third, VOC emissions for each source will have to be appor-
tioned into classes; in some models, NOX emissions will have to be
distinguished as NO and N02« Techniques for providing the neces-
sary spatial and temporal resolution in the photochemical modeling
inventory are prescribed in general in Sections 6.2 and 6.3, respec-
tively. Specific recommendations are given in Sections 6.5 through
6.16 for several of the more important area source categories.
Chapter 7 presents procedures for apportioning area source emissions
into classes.
6.2 GENERAL METHODOLOGY FOR SPATIAL RESOLUTION
As mentioned in Chapter 2, photochemical models require that
all emissions be resolved to the grid cell level. Either of two
approaches can be employed to accomplish this task. In certain
cases, the agency may be able to determine the activity levels and
emissions of some area sources directly for each grid cell. More
commonly, it will be necessary to apportion the countywide emissions
to the grid cell level by assuming that the distribution of such
area source activity behaves similarly to some surrogate indicator.
Both approaches are discussed herein.
6.2.1 DIRECT GRID CELL LEVEL DETERMINATION OF EMISSIONS
In limited instances, sufficient information may be available
to the air pollution agency to calculate area source activity levels
and emissions directly at the grid cell level. This would primarily
be the case where sufficient information is available on individual
facilities such that they could be considered as minor point sources
in the annual inventory; however, for various reasons, the agency
has chosen to consider them collectively as an area source. One
example of this situation is where a local gas company has informa-
tion on the quantity of natural gas fired in every household or
commercial establishment. This situation may also arise when survey
results are available on a particular type of commercial or
6-3
-------
industrial establishment. For instance, a survey may have been con-
ducted resulting in information on the sales and location of each
gasoline service station in the modeling area. Hence, in the photo-
chemical modeling inventory, instead of aggregating gasoline sales
and calculating emissions at the county level, as may be done in the
annual inventory, gasoline sales can be aggregated and emissions
calculated directly for each grid cell. When emissions are calcu-
lated directly at the grid cell level, the resulting emission totals
are potentially more accurate than if calculated at the county (or
equivalent) level.
Of course, another option available to the agency if this type
of survey information is available is to "reassign" each facility as
a point source in the detailed, photochemical modeling inventory.
This latter course might be advantageous if the agency is consider-
ing implementing a control measure that can best be evaluated if
each facility within the source category is treated as a point
source. An obvious disadvantage in doing this is that many addi-
tional point source records would have to be generated and main-
tained both in the base year and projection inventories. In this
latter regard, it is generally much easier to handle numerous, small
establishments as area sources in projection inventories than as
point sources - especially if information is not available on the
location of each facility in the projection years.
6.2.2 SURROGATE INDICATOR APPROACH
When area source activity levels and emissions cannot be deter-
mined directly at the grid cell level, surrogate indicators must be
specified whose distribution is known at some subcounty level and
which behave similarly to the activity levels of interest. These
surrogate indicators are then applied to the basic, countywide
inventory to apportion activity levels or emissions to each grid
6-4
-------
cell. The variables most commonly used as surrogate indicators of
area source activity are land use parameters, employment in various
industrial and commercial sectors, population, and dwelling units.
Different surrogate indicators will be used for various area source
categories, of course, depending on which one best describes the
spatial distribution of the emissions and on what information is
available for the modeling area. Engineering judgment is most often
used to select the appropriate indicators for apportioning area
source emission totals. Table 6-6 suggests appropriate surrogate
indicators for various area source categories, as utilized in
various urban areas.
Two basic sources of information on surrogate indicators are
(1) land use and other maps, and (2) Census Bureau demographic sta-
tistics by traffic zone. Both sources may be used in the photochem-
ical modeling inventory to develop apportioning factors, as dis-
cussed herein.
6.2.2.1 Developing Apportioning Factors From Land Use Maps
Most urban areas large enough to apply photochemical models
will have land use maps available for the present and several pro-
jection years. These maps can be used by the air pollution control
agency to develop apportioning factors for those area sources whose
emissions will be distributed based on various land use classi-
fications.
This procedure is best illustrated by use of an example. As-
sume that the existing inventory contains an estimate of total emis-
sions from drycleaning for the entire area, and no specific survey
or other information is available for individual drycleaning estab-
lishments. A surrogate indicator must be chosen that will permit
the distribution of these emissions among the individual grid cells
in the study area. For making distributions at a subcounty level, a
land use map supplied by a local planning agency is often the best
available tool. The land use map cartographically characterizes
6-5
-------
each part of the study area in terms of what kinds of activities are
predominant in that area. Figure 6-1, for example, shows a land use
map of a part of the Tampa Bay, Florida region. The various areas
are identified in great detail by numbers; Table 6-1 exhibits the
coding system used in this application. Other land use maps may use
colors or shading techniques to differentiate areas.
Since drycleaning is a typical commercial activity, it is rea-
sonable to assume that drycleaning area source emissions emanate
uniformly from the commercial areas as shown on the land use map.
Thus, the surrogate indicator will be the area devoted to commercial
land use (represented in Figure 6-1 by codes 12 and 15). In this
approach, it is necessary to estimate the area within each grid cell
which is designated as a commercial area on the land use map. For
this purpose, the grid system network must be superimposed on the
land use map, as shown in Figure 6-1. The estimates of land use
area in a grid cell can be fairly rough (e.g., to the nearest tenth
of a grid cell). As an example, consider the grid cell designated
(15,15) in Figure 6-1. For this grid cell, about 20 percent of the
area is indicated as commercial (code number 12), while the remain-
ing 80 percent of the grid cell is designated as single-family resi-
dential (code number 11). If a grid cell contains an area desig-
nated by code number 15 (industrial and commercial combined), such
an area may be weighted at 50 percent in this computation.
The emissions for each grid cell are then estimated as a simple
fraction of the total, as follows:
Ei = ET (
where
E denotes emissions
S indicates surrogate indicator,
i indicates the value in grid cell i
and
T indicates the total for the county or region
6-6
-------
2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
Figure 6-1. Segment of Land Use Map for Tampa Bay, Florida
6-7
-------
Table 6-1. LAND USE CATEGORIES FOR FIGURE 6-11
URBAN OR BUILT-UP LAND
10 Multi-family residential
11 Single family residential
12 Commercial and service
13 Industrial
14 Transportation, communi-
cation and utilities
15 Industrial and commercial
combined
16 Mixed urban or built-up
land
17 Other urban or built-up
land
2. AGRICULTURAL LAND
23 Confined feeding operation
24 Other agricultural land
25 Cropland
26 Improved pasture
27 Specialty farms
28 Orchards, groves, nurseries
and ornamental horticul-
tural groves
29 Citrus groves
3. RANGELAND
31 Herbaceous rangeland
32 Shrub and brush rangeland
33 Mixed rangeland
4. FOREST LAND
41 Deciduous forest land
42 Evergreen forest land
43 Mixed forest land
5. WATER
51 Streams and canals
52 Lakes
53 Reservoirs
54 Bays and estuaries
6. WETLAND
63 Freshwater forested
wet!and
64 Freshwater marsh
65 Saltwater forested
wetland
66 Saltwater marsh
7. BARREN LAND
72 Beaches
73 Sandy areas other
than beaches
75 Extractive
76 Transitional areas
6-8
-------
The units for the surrogate indicator can be arbitrary (e.g., per-
cent of grid cell, square kilometer, square mile). For example,
assume that the total commercial area in Figure 6-1 covers an area
the size of 26.3 grid cells. Then the fraction of the total commer-
cial area (S^/Sy) for grid cell (15,15) will be 0.2/26.3, or
0.0076. (This fraction is known as an "apportioning factor.")
Thus, the emissions for drycleaning attributed to grid cell (15,15)
will be 0.0076 times the total drycleaning emissions from the entire
region. Mathematically, this can also be expressed by the equation
n
f. . = s../ y s. ,
i ,k i ,k ^ T ,k
i = 1
where
f. . is the apportioning factor for grid cell i with
1a respect to source category k, and
n is the total number of grid cells.
Other maps, if reasonably current, can be used to develop appor-
tioning factors for various area sources. United States Geological
Survey (USGS) maps, for example, show the location of oil and gas
wells. By counting the number of wells per grid cell, total oil and
gas well emissions can be apportioned by multiplying them by the
fraction of wells lying in each grid cell. (In this case, the num-
ber of wells serves as the surrogate indicator of production activ-
ity.) Similarly, USGS maps show railroad track mileage by grid
cell, which may be used to develop apportioning factors for railroad
emissions. A disadvantage of developing apportioning factors from
maps other than land use maps is that corresponding projection in-
formation will often not be available for allocating future year
emissions. In these cases, the air pollution control agency will
either have to (1) assume that projection year emission patterns of
these sources will not change, or (2) locate additional information
6-9
-------
that shows what changes are expected in these surrogate indicator
distributions.
The foregoing discussion has only dealt with the allocation of
area source emissions based on a single surrogate indicator. There
may be cases where the agency feels that no one parameter accurately
describes the subcounty distribution of emissions from a particular
area source category. In these cases, apportioning can be based on
two or more surrogate indicators. For example, since miscellaneous
solvent use can be associated with both consumer (residential) and
commercial applications, the agency may wish to distinguish between
the possible different rates of use in these land use categories
(10, 11, and 12 of Table 6-1).
There are two principal ways to manage this type of apportion-
ment. First, the agency may estimate solvent emission subtotals for
three types of land use involved, then apportion each of these sub-
totals according to the corresponding land use areas. (In effect,
this creates three new emission subcategories where the countywide
inventory may have had only one.) For example, the agency might
choose to assign one third of the miscellaneous solvent emissions to
multifamily residences (land use 10), one third to single family
residences (land use 11), and one third to commercial and service
use (land use 12). Hence, if countywide emissions from miscellane-
ous solvent use are 12 tons per day, 4 tons per day would be
apportioned at the grid cell level for each of these subcategories,
based on the distribution of the corresponding surrogate indicator.
Alternatively, the agency might decide to estimate the relative
intensity of solvent use in the three types of areas. For example,
assuming single-family residential areas have the smallest emission
rate per unit area, the agency might estimate that the emission rate
in multiple family residential areas is three times as large, and in
commercial and service areas, five times as large. In this case,
6-10
-------
the apportioning factors would be calculated using an appropriate
weighting factor for each of the three types of land use. This
would be expressed, mathematically, by the equation
fik
3
= I WjkSij
j = 1
n 3
I I Wjksij
i = 1 j = 1
where
Wjk is the weighting factor selected for land use type j in
relation to source category k.
Sjj is the value of the surrogate indicator (i.e., the
area) of land use type j in cell i.
In this application, the summation term appearing in the numer-
ator is essentially a composite surrogate indicator for the entire
category. Thus, if solvent emissions are weighted according to the
previous suggestion (W^ = 1, ^2 ~ 3> W2 = 5) and tne respec-
tive areas in a given grid cell are 0.6, 0.2, and 0.2, then the val-
ue of the composite surrogate indicator for that cell is 0.6 x 1 +
0.2 x 3 + 0.2 x 5, or 2.2. The entire category is then apportioned
as usual, based on this composite surrogate indicator.
6.2.2.2 Developing Apportioning Factors From Demographic Statistics
at the Traffic Zone Level
As part of the transportation planning process routinely per-
formed in larger urban areas, employment and other demographic sta-
tistics are aggregated at the zonal level. These statistics can be
used instead of land use maps to obtain the information needed to
apportion area source emissions to the subcounty level. In theory,
the same data contained in these zonal statistics are available from
land use maps; thus, the only difference in using one approach or
the other is procedural. In practice, however, the land use maps
that are typically prepared in certain areas may not show all the
6-11
-------
detail available in these zonal statistics. For instance, zonal
statistics in a particular urban area may typically be compiled for
five or more commercial and industrial subcategories; however, the
corresponding land use maps may only identify generalized commercial
and industrial land uses. Also if land use maps are used for devel-
oping apportioning factors, the manual step of coding land use data
for each grid cell must be repeated for every growth projection. If
zonal statistics are used, however, this process can be largely
automated once a set of zone-to-grid-cell conversion factors has
been developed. These conversion factors are discussed later in
this section. The data handling aspects of utilizing zonal statis-
tics are discussed in more detail in Chapter 8.
The following example illustrates the use of detailed zonal
statistics for developing allocation factors as well as the use of
multiple surrogate indicators to apportion emissions from a number
of area source categories. In the San Francisco Bay area of Cali-
fornia, emissions from 58 area sources are apportioned using combi-
nations of the 19 demographic parameters shown in Table 6-2, all of
which are compiled at the subcounty level by the local MPO as part
of transportation planning studies. For some area source catego-
ries, a single parameter from Table 6-2 is used as a surrogate indi-
cator of the distribution of emissions. For instance, the source
category "farming operations" is linked with the single employment
category "AGRI" from Table 6-2, which includes agriculture produc-
tion and services. Similarly, the source category "printing" is
distributed with the variable "MFGI," which includes printing, pub-
lishing, and related industries. However, emissions from other area
source categories are not felt to be accurately distributed by a
single variable. In these instances, emissions are apportioned
based on two or more parameters. An excerpt from a cross-classifi-
cation table used in the Bay Area is presented in Table 6-3, which
6-12
-------
Table 6-2. DEMOGRAPHIC PARAMETERS USED IN SAN FRANCISCO BAY AREA
FOR MAKING ZONAL ALLOCATIONS OF AREA SOURCES2
Variable3
Name
SICb
Classification
Description
DWELL
AGRI
MIN
MFG1
MFG2
MFG3
MFG4
MFG5
MFG6
TRAN
WHOL
FIN
SERV 1
SERV 2
GOV
RET
BUS. SERV.
RET. SERV.
OTHER
SERV.
(not applicable)
1. 7-9
10, 13, 14
27
26, 28, 29, 32, 33
20
19, 36, 38
34, 35, 37
22-25, 31, 39
40, 42, 44-46
50, 52
62, 63, 67
73
82, 84, 89
91, 92
53-59
80, 81, 96
70, 72, 75-79
15-17, 41, 47-49, 60
61, 66, 93-95, 99
Dwelling units
Agriculture, forestry
Mining, quarry, oil and gas
extraction
Printing, publishing
Petrol., chem., paper, metal
industries
Food and kindred products
Electrical, optical,
machinery & instruments
Fabricated metal products
Textiles, apparel, wood,
leather
Transportation (non-auto),
pipelines
Wholesale trade, building
material
Financial, insurance
Business services
Educ. service, museums,
galleries
Government
General merchandise & food
stores
Health, legal, admin, services
Hotels, personal service,
repairs
Construction, transit, utili-
ties, banking, real estate,
other
aThe variable referred to is the employment, totalled in each
zone, for the SIC classifications listed in the next column.
(DWELL is an exception, as described in Column 3)
bStandard Industrial Classification Code
6-13
-------
>•
CQ
O
oo
»—I
O
•z. oo
ea; LU
oo i—i
Q
LU
UJ
UJ O
_i o:
en =>
< o
i— oo
•z. <
O UJ
i— i o:
r- <
eC
CJ U.
•-« o
oo o
OO i— i
O O
o
OO _l
oo _i
o <:
a:
o >-
CJ3 Z
et =3
DQ O
< O
co
s: =>
o oo
C£.
u_ o:
o
h- u.
0.
Q;
UJ
o
X
OO
I
a)
.a
> or
•«a- o: uj
CM LU 31
""S
ro LU o;
CM o: LU
CO
CO >
CM o ce
CM CQ LU
CO
~«
8|
CM
en S
' — LU
CO
CO OC
r- LU
CO
r- =
*~~ u_
i
ID O
LO «£
t—
LO
*r cj
1 — u_
nS
i — u_
CM S
" — U_
ro
CM
O t3
i— Lu
o»|
oo |
^S
CD
— 1
— 1
i — LU
3
O
c
O
4->
cp re
££
3 «+-
o ••-
CO l/l
in
re
o
O LO LO LO
O moo
O
5 °gg
O
o
LO
LO
in
O LO in LO
0 0
CO CO
o in oo
o
o
o
0 OO 00 0
en o LO f— en »—
o o
o
O
O LO O
-o
C 4-> U
re vt . o i
flj co re o */> • i-
+j i — o C • cno_ . m
(j OJ'CC11.OU4-)|4- (Q.^
£l_0> Q.-r-OirtZ- --M C
uc "o 4^v--*- E coco ret/)
Q- • •— CO C i— ro Q. Z t/> O •— t— -*
Zreo-«-> nret- 4->u or- — a> c
O (jcjco coi-ai '>>u^^ i— i — cncure
E LJ-*- •— euo-ores- co-i-rocDi—
QJ «^cniieuicO-'-i--oo ^-.u_r-re
J= .1- >.,_ t-O-OCQ .— S- •
<_} ^ 3 p^ p— re. (— z 01 CTILO ^. >—• d) ••— o _F
<; . ^-.^-i_i.csfc..-^-^4-> • z: ucoco:>
ou recua; OIUETJC-CJU t— -r-
•— « (/> Ct 4->CCoBci/)l-O'f-Ov) — IJZI 1 1
z -r- LU t~ •(- -r- re o ro o •<- o cu
LkJZ 3TZZZ ZZU-Li_Ci-3Z CO =»
CJ O CJ
LO CO CO O
LO OLO coroOCDO O
r- CM O O i—
cn ro
coco
ro CO
CO CO
CO CO
co ro
Lft LO CO CO
O O CO CO
0 0 OLO
CM r— •—
00 OO 0
CM t— LO cn \o
o o o o
CM • — i — > —
LO
CD O O O O
CM CM r- O Cn
0
CD cn ro
in co ro
VO 10
QJ CU
a.' a. re re c
< E CO CO CJ • C O
> o • a: > o •«-
LU CJ • t- 4-> LU r— •!— 4->
i > u re a. o -*-1 ro
. -i — +JO.JCLI coreu
o_»/> cnoro(_>t/)t/i i (_)-i—
§^£ i_co2 rero ••»— u
C-MOl 1 -CQCD 01 U i- -O
cjioc ••£ ioi-wjare
i — cu '4J-*-*o -t-s-cU'i-reu-
>urereo>cucuczL|-
cs IP »— i/i o o r— -M t/i re i u
o recozr co^cui — • — aj*-'
s- ... i_ u u -Q VP
o: o i i -o X) E i i cn >, >,.a re
LU+J CCO QJi-l-3'—
^CO ^-i^-.(_) QQOOiO-
fe
LO
o
LO
LO
LO
o
LO
8
LO
LO
o
o.
ro
UJ
Cn S_
c o
c u
a-z
6-14
-------
shows the percentage of each area source emission total that is
apportioned by each demographic parameter.
To continue this example, assume that area source degreasing
emissions in a given county are 42 tons/day of VOC. According to
Table 6-3, 10 percent of this total should be apportioned according
to the distribution of employees engaged in the manufacture of
electrical and optical machinery and instruments (MFG4), 60 percent
based on fabricated metal product employment (MFG5), 20 percent
based on retail service employment (RET SERV.), and 10 percent based
on other services, including local transit and transportation ser-
vices (OTHER SERV.). Thus, 4.2 ton/day (42 x 0.10) are apportioned
according to the fraction, in each grid cell, of the total number of
employees in the "MFG4" category; 25.2 ton (42 x 0.60) are appor-
tioned according to the fraction, in each grid cell, of the total
number of employees in the "MFG5" category; 8.4 ton (42 x 0.20)
apportioned according to the fraction of employees in each grid cell
in the "RET. SERV," category; and 4.2 ton (42 x 0.10) apportioned
according to the fraction of "OTHER SERV." employees in each grid
cell. For instance if the ith grid cell contains 0.1 percent of
the total areawide "MFG5" employees, 0.05 percent of the "RET.
SERV." employees, 1 percent of the "MFG4" employees, and no "OTHER
SERV." employees, then the degreasing emissions would be apportioned
to that grid cell as follows:
ith Grid Cell Emissions = 25.2 x 0.001 + 8.4(0.0005)
+ 4.2(0.01) + 4.2(0)
= 0.0714 ton/day
The degreasing emissions for the other grid cells would be appor-
tioned accordingly, as would the emissions for the other area
sources. An equivalent formulation of this procedure is simply to
6-15
-------
subdivide the area source degreasing category into four subcatego-
ries, namely, (1) degreasing, MFG4; (2) degreasing, MFG5; (3) de-
greasing, RET SERV.; and (4) degreasing, OTHER SERV. Then, if the
agency has estimated the total countywide degreasing emissions for
these subcategories as 4.2, 25.2, 8.4, and 4.2 ton per day, respec-
tively, these amounts will be allocated in the appropriate subcate-
gories, using the corresponding demographic parameter as the surro-
gate indicator in each case.
In the preceding apportioning calculation, it is assumed that
apportioning factors are compiled at the grid cell level. In actu-
ality, as was pointed out at the outset of Section 6.2.2.2, the sur-
rogate indicators (such as the demographic parameters shown in Table
6-2) used for apportioning are initially compiled at the zonal level
for transportation and other planning purposes, and not at the grid
cell level. For example, in the San Francisco Bay area, the local
MPO develops its population, land use, and employment data for 440
zones, each of which comprises one to seven census tracts. By con-
trast, there are some 5,000 grid cells to which area source emis-
sions are apportioned for photochemical modeling purposes.^
Thus, if zonal statistics are used for apportioning area source
emissions, the air pollution control agency is confronted with the
task of making a zone-to-grid-cell conversion prior to completing
the apportioning steps. This step will not be required if appor-
tioning factors are developed from land use maps because when this
is done, the grid system is overlaid onto the land use map and the
values of each surrogate indicator are directly determined for each
grid cell by visual means.
6-16
-------
The first steps in this zone-to-grid-cell conversion process
are (1) overlaying a map outlining the grid system over a map show-
ing the zone boundaries, and then (2) determining or estimating
fractions of zonal area lying within specific grid cells.
To determine the fraction of the area of a given zone which
lies within a given grid cell, it is useful to develop a zone-to-
grid-cell correspondence table, such as that shown in Table 6-4.
For each zone, the area falling in each grid cell is estimated in
terms of the fraction (A) of the grid cell covered by that zone;
then, the total of these fractional areas for all of the affected
grid cells (2A) is the total area of the zone. (An exception must
be noted in case a part of the zone lies outside the emission grid.)
For each grid cell, the appropriate fraction (g) of the given zone
is obtained by dividing the fractional area of intersection by the
total area of the zone. The contribution of the zonal emissions to
the grid square can be obtained by multiplying the zonal emissions
(in any or all categories) by this fraction.
The next step in this zone-to-grid-cell conversion process is
to multiply the latter fractions (jj) by the known zonal values of
each demographic parameter in order to reaggregate the data at the
grid cell level. Mathematically, this process may be expressed as
follows:
aik = I gijbjk'
where
a., is the value of the L. demographic parameter, re-
aggregated to grid ceil j_;
b .. is the value of the k., demographic parameter, as
compiled for zone j_;
g. . is the fraction of area of zone ^ in cell i_.
. .
6-17
-------
Table 6-4. ILLUSTRATIVE EXCERPTS FROM ZONE-TO-GRID-CELL
CORRESPONDENCE TABLE FOR DETERMINING APPORTIONING FACTORS
Zone
1
2
17
18
23
GC
A
g
GC
A
g
GC
A
g
GC
A
g
GC
A
g
0101
1.0
.19
0103
0.3
.08
0514
0.6
1.0
0514
0.1
0.33
0624
0.3
0.33
0102
1.0
.19
0104
1.0
.26
0614
0.2
0.67
0625
0.6
0.67
0103
0.7
.13
0202
0.5
.13
0201
1.0
.19
0203
1.0
.26
0202
0.5
.10
0204
0.6
.15
0301
0.7
.13
0301
0.2
.05
0401
0.3
.06
0303
0.2
.05
0304
0.1
.03
Total , SA
5.2
3.9
0.6
0.3
0.9
Legend: GC = grid cell number
A. . = area of intersection of zone with grid cell
g1.^ = apportioning factor from zonal level to grid cell
1J (all activities assumed uniform throughout zone)
6-18
-------
Further, g.. is determined by
g. . - A. ./£ A. .
i
where
A-• is the fractional area of intersection of zone j with
cell i, in terms of the fraction of cell i covered by zone j.
To calculate apportioning factors from the county level to the grid
cell level, the grid cell parameters must be normalized to the total
for the county, i.e.,
fi,k = aik/I aik
The same normalizing factor can, of course, be obtained by totaling
the zonal values; that is, \ a.. = ^b., , except for
i J
necessary corrections for any zone which falls partly outside the
county. These apportioning factors (f. , ) are then applied in
i ,K
an identical manner as the (S^/Sj) factors described on page 6-6
that were determined from land use or other maps.
The most difficult part of this zone-to-grid-cell conversion
process is the determination of the g. . fractions. This step
' >J
must generally be performed manually because the irregular nature of
zonal boundaries in most urban areas complicates the computerization
of this assignment. (However, the rest of the above calculations
are readily automated.) One important assumption that should be
kept in mind when estimating these g. . fractions in this zone-
i »J
to-grid-cell conversion step is that the distribution of each demo-
graphic parameter is uniform within each zone. In situations where
zones are much larger than the coincident grid cells, this assump-
tion can lead to erroneous distributions if most activity within
6-19
-------
particular zones actually takes place on one side or the other.
Hence, prior to use, some review of the resulting correspondence
table values should be made by local planners or others knowledge-
able with the land use patterns in the urban area. In select cases,
more activity might be distributed to one or several grid cells
rather than being distributed to all grid cells simply on the basis
of area. Generally, however, since zones are defined as areas of
similar activity, this should not be a major consideration to the
inventorying agency in most instances.
6.3 GENERAL METHODOLOGY FOR TEMPORAL RESOLUTION
Since the basic area source inventory will generally only con-
tain estimates of annual (or perhaps seasonally adjusted) emissions,
additional effort is needed in order to estimate hour-by-hour emis-
sion rates. Several approaches can be employed to develop hourly
emissions resolution; all involve the use of assumed diurnal pat-
terns of activity. Moreover, in addition to hourly patterns, esti-
mates will be needed of the fraction of annual (or seasonal) opera-
tions for each area source that occurs on typical weekdays during
the oxidant season.
The first step in this process, if only annual emission esti-
mates are available from the countywide inventory, is to estimate
the seasonal component of activity for each area source. For many
sources, activity is fairly constant from season to season. For
example, degreasing operations would not normally have a strong
seasonal component, and hence, degreasing activity in the July-
September quarter could be assumed to be 25 percent of the annual
total. For some sources, a strong seasonal component will be evi-
dent. For instance, off-highway vehicles such as farm tractors,
lawn mowers, etc., will be most active in the warmer months of the
year, as will cutback asphalt paving operations. Conversely,
6-20
-------
residential fuel use will occur primarily in the colder winter
months. In this latter case, it may be reasonable to assume no such
fuel use during the oxidant season. Seasonal adjustment is dis-
cussed in detail in Chapter 7 of Volume I.
Once the seasonal component is known, the daily component
should be determined. Again, some area source activities are fairly
constant from day to day, making it a simple matter to estimate
daily activities. For example, gasoline storage losses and natural
gas leaks would be expected to be uniform over the week. Many work-
related area sources, on the other hand, would be more active on
weekdays. For instance, drycleaning plants and degreasing opera-
tions will concentrate their activities during Monday through Friday
(or Saturday, in some cases). Hence, in these instances, the sea-
sonal activity should be distributed to only those days on which the
source is active. If, hypothetically, drycleaning emissions for the
entire area are 312 tons of solvent over the 92-day period from July
to September, and most plants are typically open 6 days a week (for
a total of 78 operating days), then daily emissions from drycleaning
would be 4 tons (312 *• 78). This daily emission rate would not, of
course, be applicable to a Sunday. As explained in Chapter 2,
photochemical models are usually run for weekday conditions.
After the daily activity level has been determined for each
area source, the next step is to estimate hourly emissions. This is
generally accomplished by applying a 24-hour operating pattern to
the daily activity level. An example of this is shown in Table 6-5
for gasoline service stations. As seen in this table, more gasoline
is handled in the Tampa Bay area in the afternoon than other times
in the day. For instance, 13 percent of the daily operation in
large stations occurs from 4 to 5 o'clock; hence, 13 percent of the
daily emissions from large service stations would be assigned to
6-21
-------
Table 6-5. DIURNAL PATTERNS FOR GASOLINE STATIONS IN TAMPA BAY,
IN PERCENT OF DAILY OPERATION1
Type of Gasoline Station9
Hour
6-7 a.m.
7-8
8-9
9-10
10-11
11-12
12-1 p.m.
1-2
2-3
3-4
4-5
5-6
6-7
7-8
8-9
9-10
10-11
11-12
Small
5
6
6
5
6
6
5
5
7
7
9
9
6
6
5
5
1
1
Medium
4
4
6
5
7
7
7
7
6
7
8
8
8
7
3
3
2
1
Large
8
8
8
7
2
2
8
9
5
6
13
13
4
4
2
1
a Separate diurnal distributions were analyzed for three classes
of gasoline stations: (1) small, below 200,000 gal/yr
throughput, (2) medium, between 200,000 and 500,000 gal/yr
throughput, and (3) large, above 500,000 gal/yr throughput.
Data are based on 1,133 gasoline stations in the Tanpa Bay
area.
6-22
-------
that particular hour in the detailed photochemical modeling
inventory.
The hourly operating information in Table 6-5 is an example of
the case where a special survey has been made to determine diurnal
operating patterns. This approach is preferable, where resources
allow, for the more important area source emitters. For many espe-
cially smaller sources, however, engineering judgment can provide
sufficiently accurate temporal factors. For example, for off-high-
way fuel usage by farm and construction equipment, it is probably
adequate to assume that the bulk of all emissions on weekdays occurs
uniformly from 7 a.m. to 7 p.m.
The development of hourly area source emission estimates from
annual emissions requires a great deal of repetitive data handling,
and should generally be computerized. Area source data handling is
discussed in more detail in Chapter 8.
6.4 AREA SOURCE PROJECTION PROCEDURES
As in apportioning area source emissions, two approaches can be
used for making growth projections of area source emissions. The
more accurate approach involves projecting the activity levels them-
selves. The more common approach, however, involves the use of
growth indicators to approximate the increase or decrease of each
activity level.
The first of the above-mentioned approaches is generally em-
ployed when a local survey has been made or local estimates are
available projecting growth in specific areas. For example, if a
survey of drycleaners has been performed and the average estimated
growth in the modeling area is 5 percent per year, then in 5 years,
drycleaning activity would be projected to increase by 28 percent*.
As another example, a local asphalt trade association may be able to
*(1.05)5 = 1.28, or a 28 percent increase
6-23
-------
project cutback asphalt usage. When considering such estimates, the
inventorying agency must recognize the possibility of deliberate or
inadvertent biases due to wishful thinking or self-serving motives,
and should strive to obtain opinions which are as objective as pos-
sible. Moreover, the agency should be careful to determine whether
or not such estimates of future activity levels reflect the effects
of anticipated control measures. This is important, as some such
estimates may be more appropriately used in control strategy projec-
tions than in the baseline inventory. Most importantly, any such
projections should be consistent with projections made by other
planning agencies.
A common alternative to directly projecting activity levels is
to use indicators of growth. A surrogate indicator concept was dis-
cussed in Section 6.2 with respect to spatially allocating area
source emissions. In the context of projections, a surrogate growth
indicator is one whose growth in the future is fairly certain and is
assumed to behave similarly to the activity level of interest. The
most commonly used surrogate growth indicators are those parameters
typically projected by local MPO's, such as population, housing,
land use, and employment. As one example, the quantity of
miscellaneous solvent use in a projection year might be assumed to
grow proportionally with population. Hence, if population increased
in an area by 10 percent from the base year to the projection year,
miscellaneous solvent usage would be assumed to increase by 10
percent, as well. Regardless of what variables are used as growth
surrogates, the basic calculation is the same: the ratio of the
value of the growth indicator in the projection year to its value in
the base year is multiplied by the area source activity level in the
base year to yield the projection year activity level.
A major difference between making area source projections for
the basic, countywide inventory and for the detailed, photochemical
6-24
-------
inventory is that, in the latter, emission estimates must be re-
solved at the grid cell level. This adds a dimension of complexity
to the projection effort as changing growth patterns may require
that different apportioning factors be determined for the projection
years. Fortunately, in most large urban areas where photochemical
models are employed, the local MPO will be able to provide land use
maps as well as detailed zonal projections of employment, popula-
tion, etc., for future years. Hence, these projections can be used
directly, as described above, to determine changes in spatial emis-
sion patterns.
If the surrogate indicators used for apportioning certain area
source emissions are not projected at a subcounty level, engineering
judgment must be used to decide whether spatial distributions of
various activities will change enough to warrant the effort of
identifying new patterns. Changes may be warranted in rapidly
growing areas for the more important area source emitters. For
regions where little growth is expected, and especially for minor
area sources, the same apportioning factors can be used in baseline
and projection inventories without incurring appreciable error.
In special cases, temporal factors and VOC split factors may
also change in the projection years. Temporal factors may change as
lifestyles and working patterns change, or as a result of govern-
mental policy. For example, if a 4-day workweek is expected in a
projection year, daily emission patterns from sources such as small
degreasing operations may be altered. Or if gasoline sales are pro-
hibited on certain days or during certain hours, daily emission pat-
terns may change. Generally, however, temporal patterns for most
area sources would not be expected to change, and hence, the same
daily and hourly apportioning factors would be used in the baseline
and projection years. VOC split factors are discussed in Chapter 7.
When making area source emission projections, control measures
will have to be considered for certain source categories. The
6-25
-------
effect of controls on area sources can generally be indicated by
changes either in emission factors or in activity levels, depending
on the source and the nature of the control measure being consid-
ered. As an example, RACT for gasoline service stations could be
accounted for by using an emission factor lower than the uncontrol-
3 4
led factor given in AP-42. ' As another example, RACT for cut-
back asphalt paving could be evaluated simply by reducing the activ-
ity level in proportion to the fraction of cutback asphalt that
would be replaced with emulsified asphalt.5
As with point sources, the area source projections should be
carefully reviewed by the inventorying agency in light of all of the
points (e.g., objectivity, openness, etc.) discussed in Section 2.6.
It is especially important with respect to area sources that consis-
tent methodologies be utilized for the base year and the projection
years to estimate and apportion emissions for each source. For
example, if emissions from gasoline evaporation at service stations
in a base year are estimated and distributed as a result of a spe-
cial study (e.g., questionnaires to individual service stations), it
would be inconsistent to estimate such emissions for a future year
based on projected VMT and to apportion these emissions based on the
number of miles of road within each grid. Such inconsistencies will
likely lead to changes in emissions that are not due to growth or
control measures but, rather, to changes in the inventory procedures
themselves.
A test to determine whether different base and projection year
methodologies are mutually consistent is to apply the projection
year methodology to the base year case and see if the results are
identical. If important discrepancies are found, then one methodo-
logy should be chosen to apply to both years. Generally, any metho-
dology which applies growth factors to base year estimates to esti-
mate projection year emissions (or activity levels) will meet this
consistency criterion.
6-26
-------
6.5 SPECIFIC AREA SOURCE DATA
As discussed in preceding sections of this chapter, the detail-
ed photochemical modeling inventory differs from the basic emissions
inventory in the increased spatial and temporal resolution required.
These additional considerations apply in the base year as well as in
projection years.
Table 6-6 summarizes some approaches that have been employed
for incorporating spatial and temporal resolution for area source
categories in the detailed emissions inventory and for making pro-
jections of countywide emissions from area sources in future years.
Surrogate indicators of area source activity are suggested for
resolving emissions to the grid cell level. Temporal resolution
procedures are illustrated by examples based on engineering judgment
and emissions inventory experience*. Projection methodology is
indicated in terms of suggested surrogate indicators of growth
(which are sometimes the same as the suggested surrogate indicators
of area source activity).
The information presented in Table 6-6 is not exhaustive, nor
are the approaches suggested automatically applicable to all cases.
Local information, if sound, is generally to be preferred. In vari-
ous urban areas, the MPO's will have developed entirely different
sets of surrogate indicators, both for spatial disaggregation and
for projections; in some cases, substantially different methodolo-
gies may have been developed. For temporal resolution, local work-
ing hours and seasonal activity patterns may differ from those
suggested here; some simpler suggestions are provided in the later
sections of this chapter. The most general default option, in the
absence of knowledge about temporal resolution, is the assumption of
complete temporal uniformity. However, it is usually easy to deter-
mine whether any important emitting activity takes place mainly in
the summer (as opposed to the winter), on weekdays (as opposed to
6-27
-------
Table 6-6. EXAMPLE METHODOLOGIES FOR SPATIAL AND TEMPORAL
RESOLUTION AND PROJECTION OF COUNTYWIDE EMISSION TOTALS FOR AREA
SOURCE CATEGORIES
(For interpretation of land-use codes see Table 6-1)
1. Aircraft, Commercial
Spatial Resolution
Surrogate indicator: airport area by airport (code
14)
Information source(s): land use map, local airport
authority, and References 6, 7
Temporal Resolution
Seasonal: Since the temporal profiles of commercial
airports vary widely, the respective airport
Daily: managers should be contacted. The airport
managers usually have very detailed temporal
Hourly: information.
Projections
Growth indicator(s): projections should be done on a
case-by-case basis. Projected
land use maps may be useful.
Information source(s): local airport authority, local
MPO, state aviation system
plan
Aircraft, General
Spatial Resolution
Surrogate indicator: airport area by airport (code
14)
Information source(s): land use map, local airport
authority, and Reference 8
Temporal Resolution
Seasonal: uniform through the year
., 40 percent of operations occur on weekends
31 ^' and 60 percent on weekdays.
Hourly: uniform from 0700 to 2100, otherwise zero
Projections
Growth indicator(s): projections should be done on a
case-by-case basis; projected
land use maps may be useful.
6-28
-------
Table 6-6. EXAMPLE METHODOLOGIES FOR SPATIAL AND TEMPORAL
RESOLUTION AND PROJECTION OF COUNTYWIDE EMISSION TOTALS FOR AREA
SOURCE CATEGORIES (Continued)
Information source(s): local airport authority, local
MPO, state aviation system plan
Comments: the local airport managers should be contacted
for this information.
3. Aircraft, Military
Spatial Resolution
Surrogate indicator:
Information source(s):
Temporal Resolution
Seasonal:
Daily:
Hourly:
Projections
Growth indicator(s):
Information source(s)
airport area by airport (code
14)
land use map, local airport
authority, and Reference 9
estimate on individual basis -
contact local airport
authorities and appropriate
military agencies
4. Agricultural Equipment
Spatial Resolution
Surrogate indicator: agricultural land (codes 21, 22,
24, 25, 28, and 29)
Information source(s): land use map and Reference 10
Temporal Resolution
Seasonal: uniform through the agricultural season
Daily: uniform through the week
Hourly: uniform from 0700 to 1900, otherwise zero
Projections3
Growth indicator(s): agricultural land use;
agricultural employment
Information source(s): local MPO
aDisparate trends for gasoline-fueled and diesel equipment suggest
use of separate subcategories in these projections. See text.
6-29
-------
Table 6-6. EXAMPLE METHODOLOGIES FOR SPATIAL AND TEMPORAL
RESOLUTION AND PROJECTION OF COUNTYWIDE EMISSION TOTALS FOR AREA
SOURCE CATEGORIES (Continued)
5. Construction Equipment
Spatial Resolution
Surrogate indicator:
Information source(s):
Temporal Resolution
Seasonal3:
Daily:
Hourly:
Projections*3
built-up land and transitional
areas (codes 16, 17, and 76)
land use map and Reference 11
20 percent Dec-Feb, 25 percent
Mar-May, 30 percent Jun-Aug, 25
percent Sep-Nov
Monday through Saturday
uniform from 0700 to 1900 Monday
through Friday and uniform from
0700 to 1200 on Saturday
Growth indicator(s): heavy construction employment
(SIC code 16)
Information source(s): local MPO
aNational average. Refer to Reference 11 for further details.
^Disparate trends for gasoline-fueled and diesel equipment suggest
use of separate subcategories in these projections. See text.
6. Industrial Equipment
Spatial Resolution
Surrogate indicator: industrial areas (codes 13 and
15)
Information source(s): land use map and Reference 12
Temporal Resolution
Seasonal3: 20 percent Dec-Feb, 25 percent Mar-May,
30 percent Jun-Aug, 25 percent Sep-Nov
Daily: uniform Monday through Saturday
Hourly: 80 percent from 0700 to 1900 anJ 20 percent
from 1900 to 2400
Projections
Growth indicator(s):
industrial employment (SIC codes
10-14, 20-39, and 50-51) or
industrial land use area
6-30
-------
Table 6-6. EXAMPLE METHODOLOGIES FOR SPATIAL AND TEMPORAL
RESOLUTION AND PROJECTION OF COUNTYWIDE EMISSION TOTALS FOR AREA
SOURCE CATEGORIES (Continued)
Information Source(s): local MPO
National average. Refer to Reference 12 for further details.
7. Lawn and Garden Equipment
Spatial Resolution
Surrogate indicator: residential areas (codes 10, 11,
16, and 17)
Information source(s): land use map and Reference 12
Temporal Resolution
Seasonal: uniform during months which have
an average temperature of 38°F
or higher
Daily: 50 percent Monday through Friday;
50 percent Saturday and Sunday
Hourly: uniform from 0900 to 1900,
otherwise zero
Projections
Growth indicator(s) : single-unit housing or
population a
Information source(s): local MPO
Population may be taken as a growth indicator if it is assumed
that average housing densities will remain constant.
8. Off-Highway Motorcycles
Spatial Resolution
Surrogate indicator3: open areas (codes 24, 31, 32,
33, and 73)
Information source(s) : land use map from local MPO;
References 13, 14
Temporal Resolution
Seasonal: uniform during months which have an average
temperature of 38°F or higher
Daily: 30 percent Monday through Friday, 70 percent
Saturday and Sunday
Hourly: uniform from 0700 to 1800, otherwise zero
6-31
-------
Table 6-6. EXAMPLE METHODOLOGIES FOR SPATIAL AND TEMPORAL
RESOLUTION AND PROJECTION OF COUNTYWIDE EMISSION TOTALS FOR AREA
SOURCE CATEGORIES (Continued)
Projections
Growth indicator(s) : population
Information source(s): local MPO
aTake into account local restrictions on off-highway motorcycles.
9. Snowmobiles
Spatial Resolution
Surrogate indicator: areas where local restrictions
allow their usage
Information source(s): land use map from local MPO
Temporal Resolution
Seasonal3: 63 percent Dec-Feb, 18 percent
Mar-May, 0 percent Jun-Aug, 19
percent Sep-Nov
Daily: 30 percent Monday through Friday;
70 percent Saturday and Sunday
Hourly: uniform from 0700 to 1800,
otherwise zero
Projections
Surrogate indicator(s): projected county population
Information source(s): local MPO
aNational average. Refer to Reference 12 for further details.
10. Railroads
Spatial Resolution
Surrogate indicator: track mileage3
Information source(s): USGS transportation maps,
Reference 15
Temporal Resolution
Seasonal: uniform through the year
Daily: uniform through the week
Hourly: 70 percent from 0700 to 1800 and 30 percent
from 1800 to 0700
Projections
Growth indicator(s): revenue ton-miles
6-32
-------
Table 6-6. EXAMPLE METHODOLOGIES FOR SPATIAL AND TEMPORAL
RESOLUTION AND PROJECTION OF COUNTYWIDE EMISSION TOTALS FOR AREA
SOURCE CATEGORIES (Continued)
Information source(s): References 16, 17
aAssume emissions originate uniformly over all track miles, except
that track mileage should be doubled in counties having significant
rail yard operations.
11. Vessels, Ocean-Going and River Cargo Vessels
Spatial Resolution3
Surrogate indicator: Ports for hotel ing, shipping
lanes for underway
Information source(s): Waterborne Commerce of the U.S.
and available maps, local port
authorities
Temporal Resolution
Seasonal: uniform through the year
Daily: uniform through the week
Hourly: 75 percent from 0700 to 1900 and
25 percent from 1900 to 0700
Projections
Growth indicator(s): cargo tonnage
Information source(s): local port authorities, U.S.
Maritime Administration, or the
U.S. Army Corps of Engineers
aPorts and shipping lanes are identified in the Waterborne Com-
merce of the U.S. (Reference 18). Generally, ocean-going emissions
are computed by port and by shipping lane. Separate subcategories
may be used for ports versus shipping lanes.
12. Vessels, Small Pleasure Craft
Spatial Resolution
Surrogate indicator3: assumed usage areas
Information source(s): available maps and Reference 18
Temporal Resolution
Seasonal: uniform during the months which have an
average temperature of 45°F or higher
6-33
-------
Table 6-6. EXAMPLE METHODOLOGIES FOR SPATIAL AND TEMPORAL
RESOLUTION AND PROJECTION OF COUNTYWIDE EMISSION TOTALS FOR AREA
SOURCE CATEGORIES (Continued)
Daily: 30 percent Monday through Friday, 70 percent
Saturday and Sunday
Hourly: uniform from 0700 to 1800, otherwise zero
Projections
Growth indicator(s) : population
Information source(s): local MPO
aAssume that small craft utilize inland water areas (taking into
account local restrictions) and waters 1 mile out from ocean-
exposed shores uniformly
13. Gasoline Handling
Spatial Resolution
Surrogate indicator : Industrial and commercial-insti-
tutional land use areas
Information source(s): land use map and Reference 19
Temporal Resolution
Seasonal: uniform through the year A
Daily: Monday through Saturday
Hourly: uniform from 0600 to 2000, otherwise zero
Projections
Growth indicator(s): gasoline demand, vehicle use
(VMT), or population
Information source(s): U.S. Department of Transporta-
tion, state transportation
agency, state tax agency, or
local MPO, and Reference 20
14. Drycleaning
Spatial Resolution
Surrogate indicator: commercial areas (codes 12 and
15)
Retail service employment
Information source(s): land use map
Temporal Resolution
Seasonal: uniform
Daily: uniform Monday through Saturday
Hourly: uniform 0700-1900
6-34
-------
Table 6-6. EXAMPLE METHODOLOGIES FOR SPATIAL AND TEMPORAL
RESOLUTION AND PROJECTION OF COUNTYWIDE EMISSION TOTALS FOR AREA
SOURCE CATEGORIES (Continued)
Projections
Growth indicator: population, retail service
employment
Information sources: solvent suppliers, trade
associations
15. Degreasing
Spatial Resolution
Surrogate indicator: industrial area (codes 13 and
15)
Information source: land use map and Reference 21
Temporal Resolution
Seasonal: uniform
Daily: uniform Monday through Saturday
Hourly: 80 percent from 0700 to 1900, 20 percent from
1900 to 2400
Projections
Growth indicators: industrial employment
Information source: trade associations
16. Nonrlndustrial Surface Coatings
Spatial Resolution
Surrogate indicator: urban or built-up land (codes 10
through 17)
Information source(s): land use map
Temporal Resolution
Seasonal: uniform through the year
Daily: uniform through the week
Hourly: uniform 0700 to 1900, otherwise zero
Projections
Growth indicator(s): population or residential
dwelling units
Information source(s): local MPO
17. Cutback Asphalt
Spatial Resolution
Surrogate indicator: urban or built-up land,
(codes 10 through 17)
Information source(s): land use map
6-35
-------
Table 6-6. EXAMPLE METHODOLOGIES FOR SPATIAL AND TEMPORAL
RESOLUTION AND PROJECTION OF COUNTYWIDE EMISSION TOTALS FOR AREA
SOURCE CATEGORIES (Continued)
Temporal Resolution
Seasonal: uniform spring through fall
Daily: Monday through Friday
Hourly: uniform from 0700 to 1900, otherwise zero
Projections
Growth indicator(s): consult industry
Information source(s): consult industry
18. Pesticides Applications
Spatial Resolution
Surrogate indicator: agricultural land (codes 23-29)
Information source(s): land use map
Temporal Resolution
Seasonal: coincides with growing season
Daily: uniform through the week
Hourly: daylight hours, 0700 to 1900
Projections
Growth indicator(s): agricultural operations
Information source(s): State Department of Agriculture,
MPO
19. Fuel Combustion, Residential
Spatial Resolution
Surrogate indicator: residential areas (codes 10, 11,
16, and 17); dwelling units
Information source(s): land use map
Temporal Resolution
Seasonal: 10 percent uniform through the year and 90
percent uniform during months which have an
average temperature of 68°F or less
Daily: uniform through the week
Hourly: uniform through the day
Projections
Growth indicator(s): residential housing units or
population
Information source(s): local MPO and Reference 22
6-36
-------
Table 6-6. EXAMPLE METHODOLOGIES FOR SPATIAL AND TEMPORAL
RESOLUTION AND PROJECTION OF COUNTYWIDE EMISSION TOTALS FOR AREA
SOURCE CATEGORIES (Continued)
20. Fuel Combustion, Commercial/Institutional
Spatial Resolution
Surrogate indicator: commercial areas (codes 12 and
15
Information source(s): land use map and Reference 23
Temporal Resolution
Seasonal: 25 percent uniform through the year and 75
percent uniform during months which have an
average temperature of 68°F or less
Daily: 95 percent Monday through Saturday and 5
percent Sunday
Hourly: 90 percent from 0600 to 2400 and 10 percent
from 2400 to 0600
Projections
Growth indicator(s): commercial/institutional employ-
ment, population, or commercial/
institutional land use area
Information source(s): local MPO, land use map projec-
tions
21. Fuel Combustion, Industrial
Spatial Resolution
Surrogate indicator: industrial areas (codes 13 and
15)
Information source(s): land use map
Temporal Resolution
Seasonal: uniform through the year
Daily: uniform Monday through Saturday
Hourly: 80 percent from 0700 to 1800 and 20 percent
from 1800 to 2400, otherwise zero
Projections
Growth indicator(s): industrial employment (SIC codes
10-14, 20-39, and 50-51) or
industrial land use area
Information source(s): local MPO, land use projections
6-37
-------
Table 6-6. EXAMPLE METHODOLOGIES FOR SPATIAL AND TEMPORAL
RESOLUTION AND PROJECTION OF COUNTYWIDE EMISSION TOTALS FOR AREA
SOURCE CATEGORIES (Continued)
22. Solid Waste Disposal, On-Site Incineration and Open
Burning
Spatial Resolution
Surrogate indicator: residential areas (codes 10,11,
16 and 17) commercial/institu-
tional areas (codes 12 and 15)
industrial areas (codes 13 and
15)
Information source(s): land use map
Temporal Resolution
Seasonal: base on information gathered from local
Daily: regulatory agencies
Hourly:
Projections
Growth indicator(s): base on information gathered
Information source(s): from local regulatory agencies
23. Fires: Managed Burning, Agricultural Field Burning, and
Frost Control (Orchard Heaters)
Spatia1NResolution
Surrogate indicator: areas where these activities
occur
Information source(s): U.S. Forest Service, state
forestry departments, state
agricultural department,
extension services, citrus grove
operators, and land use map
Temporal Resolution
Seasonal: base on local regulations and on data
Daily: collected from information sources
Hourly:
Projections
Growth indicator(s): base on anticipated local
regulations as indicated by
Information source(s): information sources.
24. Fires: Forest Wildfires and Structural Fires
Emissions from these two categories are not customarily
input to models. Refer to the text for discussion.
6-38
-------
Table 6-6. EXAMPLE METHODOLOGIES FOR SPATIAL AND TEMPORAL
RESOLUTION AND PROJECTION OF COUNTYWIDE EMISSION TOTALS FOR AREA
SOURCE CATEGORIES (Concluded)
25. Miscellaneous Solvent Use
Spatial Resolution
Surrogate indicator: residential, commercial/institu-
tional areas (codes 12, 13, 15)
Information source(s): land use map
Temporal Resolution
Seasonal: uniform through the year
Daily: uniform
Hourly: 80 percent from 0700 to 1900, 20 percent from
1900 to 2400
Projections
Growth indicator(s): population
Information source(s): local MPO
6-39
-------
weekends), or in the daytime (rather than at night). When such in-
formation is available, it should be utilized, especially if impor-
tant emission categories are involved.
The approaches listed in Table 6-6 have been tested in prac-
tice, however, and are believed to be relatively plausible and fea-
sible. The surrogate indicators are coded with reference to Tampa
Bay area land use categories, shown in Table 6-1 and Figure 6-1.
Similar maps are available for most urban areas, although not always
with the same degree of detail.
It should, of course, be remembered that the only emissions to
be disaggregated by these procedures are area source totals. If
some of the sources in any category have been listed and treated as
either major or minor point sources, the emissions from those
sources must be subtracted from the countywide category total before
the disaggregation procedure is applied. If local information is
available that allows a major fraction of the emissions to be treat-
ed as minor point sources for both the base year and the projection
years, such handling may be advantageous. However, if the number of
such sources is minor and their aggregate emissions are inconsequen-
tial, or if the necessary projection information is inadequate, this
subdivision of the area source category may not be worth the addi-
tional effort required.
6-40
-------
The following sections in this chapter provide additional dis-
cussion of the material summarized in Table 6-6 for several area
source categories.
6.6 AIRCRAFT
6.6.1 DATA COLLECTION
If aircraft emissions are a relatively large proportion of
total VOC, it is advisable in the detailed inventory to obtain LTO
data specific to the major airports of interest. The procedures
that can be used to estimate VOC emissions from aircraft are summa-
rized in Section 5.1.2 of Volume I of this report.
AP-42 gives emission factors for each mode of aircraft opera-
tion and representative time that the aircraft spends in each
mode.l However, the times spent in the preflight and postflight
taxi-idle modes are highly variable at most airports and may be con-
siderably less than the times indicated in AP-42, which represent
large, congested airports. More accurate values for the taxi-idle
times should be obtained from local airport officials.
Also, the AP-42 times assumed in the approach and climbout
modes are based on the times required to descend from and ascend to
3,500 ft (1,067 m), which may not be a representative height for the
airport being considered. For example, if the emission data are
being prepared for a model simulation run of a particular day for
which there was a low inversion base of 1,000 ft (305 m), the esti-
mate based on a 3,500 ft (1,067 m) height would overestimate the
actual emissions that would impact the mixing layer, because emis-
sions occurring in the stable layer above the inversion base would
become trapped aloft and so would have no impact on the concentra-
tions in the mixing layer or at ground level. Climbouts and ap-
proaches are made at fairly uniform rates, so the times given in
6-41
-------
AP-42 can be adjusted accordingly if the 3,500 ft (1,067 m) value
has to be revised. A photochemical modeling specialist should be
consulted prior to making any such adjustment to determine what con-
ditions should be assumed in the inventory.
6.6.2 SPATIAL RESOLUTION
When the modeling grid system has been drawn on a map which
shows the location of the airports, it will be apparent how to
assign the airport emissions to the individual grid cells: taxiing
and idling occur between the runways and the terminal buildings or
hangars; takeoffs and landings take place along the runways; and
climbouts and approaches should be assigned to the areas that extend
beyond the ends of the runways. Many airports are large enough to
require several grid cells for emission distribution. If only total
airport emissions are known, they can be distributed by the fraction
of airport area in each grid cell.
6.6.3 TEMPORAL RESOLUTION
Most airports keep detailed records of daily flight activity.
The diurnal activity patterns should be obtained by specific inquiry
for the individual airports in the modeling area.
6.6.4 PROJECTIONS
Estimates of aircraft emissions in future years require as-
sumptions regarding the following factors:
1. Future airport locations
2. Distribution of operations by type of aircraft
3. Rates of emission by type of aircraft during landing and
takeoff operations
6-42
-------
Information about possible plans affecting these factors can usual-
ly be obtained from local airport directors, airport authorities,
State Departments of Transportation, and the Federal Aviation Admin-
istration. For military airports and air bases, officials of each
relevant service should be consulted.
6.7 OTHER OFF-HIGHwVW FUEL USAGE
6.7.1 DATA COLLECTION
Other off-highway fuel usage includes motorcycles, farm equip-
ment, construction equipment, industrial equipment, and lawn and
garden equipment. Emission estimation techniques for these cate-
gories are described in Section 5.1.5 of Volume I.
6.7.2 SPATIAL RESOLUTION
Since these VOC sources are relatively minor, the grid cell
allocation techniques are not critical. If possible, off-highway
motorcycle emissions should be allocated to park or recreational
areas, construction and industrial equipment emissions to industrial
areas, farm equipment emissions to agricultural areas, and lawn and
garden equipment emissions to residential areas.
6.7.3 TEMPORAL RESOLUTION
For all off-highway sources, it is probably adequate to assume
that 75 percent of the emissions occur uniformly from 7 a.m. to 7
p.m., with 25 percent occurring uniformly over the remaining 12
hours. Some more detailed suggestions, however, are shown in Table
6-6.
Seasonal variations estimated by Southwest Research Institute
(References 11 and 12) indicate that off-highway fuel usage is about
20 percent greater in the summer than on the annual average.
6-43
-------
6.7.4 PROJECTIONS
When locale-specific growth data are not available, the follow-
ing trend indicators can be used:(Reference 1)
1. For off-highway motorcycles, population
2. For lawn and garden equipment, single-family housing
3. For agricultural equipment, farm acreage
4. For industrial equipment, industrial employment
5. For heavy construction equipment, construction employment
6.8 RAILROADS
6.8.1 SPATIAL RESOLUTION
The best method of apportioning railroad emissions to individ-
ual grid cells is to obtain suitable maps of railroad track mileage
(including detailed maps of switchyards) from railroad companies or
state and local planning agencies, and to determine manually the
track mileage in each grid cell. The total county or regional emis-
sions can then be apportioned by the fraction of total track mileage
in each grid cell.
6.8.2 TEMPORAL RESOLUTION
Seasonal and diurnal activity curves can be obtained by speci-
fic inquiry of the individual railroad companies operating in the
modeling area.
6.8.3 PROJECTIONS
The Association of American Railroads (Reference 17) has re-
cently constructed three scenarios for national railroad industry
6-44
-------
growth, which may be consulted as a point of reference for estimat-
ing trends in railroad emissions. Locale-specific growth data
should be utilized first, of course, if available.
6.9 VESSELS
6.9.1 SPATIAL RESOLUTION
In order to determine to which grid cells the emissions from
vessels should be assigned, it is necessary to obtain detailed maps
of the harbors, shipping lanes, and recreational areas in the study
area. These shipping and boating areas can then be used as a sur-
rogate indicator for allocating the countywide vessel emissions to
the appropriate grid cells.
6.9.2 TEMPORAL RESOLUTION
Vessels contribute relatively minor amounts of emissions to
the urban total; diurnal patterns can be obtained by specific in-
quiry for any major sources within this group. For combined smaller
sources, it is probably adequate to assume that 75 percent of the
emissions occur uniformly between 6 a.m. and 6 p.m., with 25 percent
occurring uniformly over the remaining 12 hours.
6.9.3 PROJECTIONS
Estimates of emissions from vessels in future years require
assumptions regarding the following factors:
1. Future development or expansion of harbor facilities
2. Trends in volume of shipping into harbors
3. Changes in distribution of types and drafts of vessels
entering harbors and using waterways
4. Trends in recreational boat usage
6-45
-------
Sources of information for projection scenarios include various
port authorities and Federal agencies, including the U.S. Coast
Guard, Interstate Commerce Commission, and the U.S. Department of
the Army Corps of Engineers. Where Naval activity appears to be an
important factor, the appropriate U.S. Naval District should be
consulted.
6.10 GASOLINE SERVICE STATIONS
6.10.1 SPATIAL RESOLUTION
If a local survey has been made which yields the amount of gas-
oline handled by each station, emissions can be directly computed
for each station and assigned to the appropriate grid cell. If only
countywide service station emissions are known, emissions may be
apportioned based on the amount of residential and commercial land
use in each grid cell.
6.10.2 TEMPORAL RESOLUTION
Gasoline station evaporation losses may, with little error, be
taken to be proportional to throughput. The diurnal variation of
throughput from gasoline stations in Tampa Bay was studied in some
detail and tables were derived for stations in three categories,
based on rate of throughput. Table 6-5 shows the results.^ These
data suggest that little error would be involved in assuming a uni-
form distribution of gasoline sales from 6 a.m. to 8 p.m.
6.10.3 PROJECTIONS
Since gasoline handling is, in many areas, an important area
source of VOC emissions, it is desirable to apply a detailed method-
ology in forecasting gasoline usage and projecting emissions. Fed-
eral, state, and local transportation agencies may be able to
6-46
-------
provide official forecasts of VMT, and state and Federal energy
agencies are likely to have information relating to gasoline supply
and average miles-per-gallon fuel use estimates. If this informa-
tion is not available locally, national gasoline demand may be the
best growth indicator; however, if the population of the area is
expected to grow at a rate significantly different from the growth
rate in national population, an adjustment should be made to take
this into account. This may be done for example, by the following
formula:
- G0 x (1 +
Rr
100
where
GJ is gasoline demand in the projection year,
j is the number of years from the base year to the
projection year,
G0 is the gasoline demand in the base year,
Rn is the average annual rate of increase of national
gasoline demand, in percent,
rc is the average annual rate of increase of the county
(or regional) population, and
rn is the average annual rate of increase of the national
population.
A simple projection of recent countywide trends may give mis-
leading results if chronic gasoline supply problems develop in fu-
ture years. Even in the absence of such problems, the effects of
rising prices on gasoline demand may cause substantial deviations
from recent trends in both gasoline consumption and in vehicle miles
traveled (VMT). Another factor which can complicate the projection
problem is the increasing mileage rating of light-duty vehicles; if
VMT is to be used as a surrogate indicator of growth in gasoline
consumption, an adjustment which reflects this increase will be
needed.
6-47
-------
Projected emissions for gasoline handling must also incorporate
factors for the reduction of emissions due to imposed control mea-
sures. A reduction of 95 percent in emissions generated during
delivery of gasoline to service stations ("Stage I loading") has
already been mandated by the EPA requirements for implementation of
reasonably available control technology (RACT).^ A reduction in
emissions generated during delivery of gasoline to individual cus-
tomers at the service stations ("Stage II loading") will be similar-
ly mandated when the CTG document for Stage II loading, expected in
the near future, is published.
6.11 DRYCLEANING
6.11.1 SPATIAL RESOLUTION
Countywide drycleaning emissions should be apportioned to com-
mercial areas or to employment in retail services, after emissions
from plants identified as point sources have been subtracted.
6.11.2 TEMPORAL RESOLUTION
Drycleaning emissions may be assumed to be emitted during day-
light hours, unless more specific information is available locally
regarding the diurnal distribution of drycleaning activity. Operat-
ing hours for small plants may differ from those for plants large
enough to be considered as point sources.
6.11.3 PROJECTIONS
Future drycleaning activity may be assumed to grow in propor-
tion to population, or to employment in retail services. Trends in
usage of different solvents for drycleaning may differ; trade asso-
ciation sources should be consulted for information about such
shifts. Control techniques for solvent evaporation from
6-48
-------
drycleaning where the solvent used is perchlorethylene are described
in a recently issued CTG document, Reference 24.
6.12 DECREASING
6.12.1 SPATIAL RESOLUTION
Countywide degreasing emissions should be apportioned to indus-
trial areas or to employment in appropriate industries, after de-
greasing emissions from point sources have been subtracted. In the
application of LIRAQ to the San Francisco Bay Area, 98 percent of
degreasing emissions were found to come from area sources.2 De-
greasing emissions were ascribed to four employment categories, as
described in Section 6.2.2.2.
6.12.2 TEMPORAL RESOLUTION
Degreasing emissions may be assumed to be correlated with
industrial activity in general and to occur mainly during daylight
hours, unless more specific local information is available.
6.12.3 PROJECTIONS
Future activity in degreasing may be assumed to grow in propor-
tion to area-source industrial activity generally, or in proportion
to employment in the aggregate of employment categories used in
apportioning the degreasing emissions to planning zones and grid
squares. Again, employment in major point sources should be deleted
from the statistics before the area-source growth rates are comput-
ed. Trends in usage of different solvents for degreasing may dif-
fer; trade association sources should be consulted for information
about such shifts. Control techniques for solvent evaporation from
metal degreasing are discussed in a CTG document, Reference 25.
6-49
-------
6.13 NONINDUSTRIAL SURFACE COATING
6.13.1 SPATIAL RESOLUTION
Surface coating emissions from public use of trade paints
should be apportioned to residential, industrial, and commercial/
institutional areas of the land use map, or to appropriate demo-
graphic parameters. In the application of LIRAQ to the San Francis-
co Bay Area, 64 percent of nonindustrial surface coating emissions
were attributed to dwelling units and 3 percent to each of 12
employment categories, as indicated in Table 6-3.2
6.13.2 TEMPORAL RESOLUTION
Seasonal variation is likely to be pronounced, especially for
architectural surface coating, in areas having severe winter condi-
tions. Trade paints are likely to be used most when the weather is
warm; thus, emissions from this source will be relatively more im-
portant in the oxidant season than at other times of the year.
Surface coating emissions from area sources may be assumed uni-
form during daylight hours.
6.13.3 PROJECTIONS
Suitable growth indicators for activity in nonindustrial sur-
face coating are population and residential dwelling units. Trade
association sources should be consulted regarding trends in VOC con-
tent of the trade paints in future years. Control techniques for
VOC evaporation from architectural and miscellaneous coating materi-
als are to be discussed in an EPA CTG document to be issued in the
near future.
6-50
-------
6.14 CUTBACK ASPHALT
6.14.1 SPATIAL RESOLUTION
Emissions by solvent evaporation from the use of cutback as-
phalt, mainly in paving, should be apportioned to residential and
commercial areas.
6.14.2 TEMPORAL RESOLUTION
Cutback asphalt use is likely to be subject to substantial sea-
sonal variation, especially in areas where severe winters may inter-
fere with paving operations. Thus, emissions from this source will,
in many places, be relatively more important during the oxidant sea-
son than at other times of the year.
Since evaporation of the diluent may continue for days or weeks
after a cutback asphalt paving operation is completed, it is prob-
ably sufficiently accurate to assume a uniform distribution of emis-
sions from this source during daylight hours.
6.14.3 PROJECTIONS
According to the CTG document for cutback asphalt use, VOC
emissions from this source will be eliminated by the use of an emul-
sified asphalt containing no VOC diluent.5 Implementation of this
regulation is expected to be complete by 1982. Therefore, for 1982
and subsequent years the total VOC emissions from this source may be
taken as zero.
6.15 PESTICIDES APPLICATIONS
6.15.1 SPATIAL RESOLUTION
Application of pesticides occurs mainly on agriculturally zoned
land, but also (to some extent) on range and forest land. Local and
state agricultural and forestry agencies may be able to supply more
6-51
-------
specific information regarding locations where pesticides are rou-
tinely appl ied.
6.15.2 TEMPORAL RESOLUTION
Seasonal variation in emissions due to pesticides applications
is likely to be pronounced, as the operation occurs mainly during
the growing season.
This activity is ordinarily restricted to daylight hours.
6.15.3 PROJECTIONS
Future activity in pesticides applications may be assumed to
grow (or decline) in proportion to agricultural acreage or agricul-
tural employment. Trends in VOC components of applied pesticides
may be superposed on these activity trends. State agricultural
agencies may be able to furnish relevant information.
6.16 FUEL COMBUSTION
6.16.1 SPATIAL RESOLUTION
The separate area source emissions from residential, commercial,
and industrial fuel combustion should be allocated to individual grid
cells using the corresponding land use areas as surrogate indicators.
6.16.2 TEMPORAL RESOLUTION
Area source fuel combustion contributes relatively minor amounts
of VOC and NOX emissions to the urban total. Further, most of the
fuel used for space heating is burned in the cold months, when weather
is not conducive to oxidant formation. Only that industrial fuel use
not associated with space heating will occur during the summer season.
Suggestions regarding temporal resolution of these emissions are in-
cluded in categories 13, 14, and 15 of Table 6-6.
6-52
-------
6.16.3 PROJECTIONS
Since VOC and NOX emissions from fuel combustion area sources
are commonly low, there is little need for complexity in developing
projection estimates for these sources. The principal factor to be
accounted for in projections is the change in industrial, commer-
cial, and residential fuel demand, which may be estimated with the
assistance of State or local planning officials. Major anticipated
changes in the land use map may also be factored into the projection
scenarios.
6.17 OTHER MISCELLANEOUS AREA SOURCES
6.17.1 DATA COLLECTION
Several minor categories of VOC emissions should be included in
the inventory, namely solid waste disposal (onsite incineration and
open burning), agricultural burning, and structural fires. Emission
estimation techniques for solid waste disposal are given in Section
5.3 of Volume I; techniques for estimating emissions from agricul-
tural burning, structural fires, and other miscellaneous area
sources appear in Section 5.5 of that document. Existing countywide
estimates for these sources will normally be sufficient for use in
the detailed inventory. Since agricultural burning and structural
fires are intermittent sources, there is a possibility that they
should be excluded if the inventory is to represent a typical day.
This should be decided during the inventory strategy planning.
6.17.2 SPATIAL RESOLUTION
In most cases the detailed information necessary to accurately
assign the emissions to individual grid cells is not available, so
6-53
-------
the apportioning can be performed only in a general manner. On-
site incineration emissions should be allocated evenly to residen-
tial, commercial, and industrial areas; open burning emissions
should be allocated to rural areas; agricultural burning should be
allocated to agricultural areas; and structural fires should be dis-
tributed evenly over residential, commercial, and industrial areas.
6.17.3 TEMPORAL RESOLUTION
Agricultural burning is a seasonal activity in which all emis-
sions are released during a short period of the year. Local air
pollution control agencies should be contacted to determine the
normal daily duration of such burning. For other miscellaneous area
sources, it is probably adequate to assume that 75 percent of the
emissions occur uniformly from 6 a.m. to 6 p.m., with 25 percent
occurring uniformly over the remaining 12 hours.
6.17.4 PROJECTIONS
Local air pollution control agencies and solid waste collec-
tion agencies should be contacted to determine future regulations
applicable to on-site incineration and open burning. In many loca-
tions, open burning may be prohibited by law in the future, and
incineration will probably require controls. For structural fires,
emission estimation by population growth is an adequate projection
technique.
6-54
-------
References for Chapter 6.0
1. L.G. Wayne and P.C. Kochis, Tampa Bay Area Photochemical
Oxldant Study: Assessment of the Anthropogenic Hydrocarbon
and Nitrogen Dioxide Emissions in the Tampa Bay Area,
EPA-904/9-7/-016, U.S. Environmental Protection Agency, Region
IV, Air and Hazardous Materials Division, Atlanta, GA,
September 1978.
2. M.C. MacCracken, User's Guide to the LIRAQ Model: An Air
Pollution Model for the San Francisco Bay Area Lawrence
Livermore Laboratory, UCRL-51983, Livermore, CA, 1975.
3. Design Criteria for Stage I Vapor Control Systems for Gasoline
Service Stations, U.S. Environmental Protection Agency, Office
of Air Quality Planning and Standards. Emission Standards and
Engineering Division, Research Triangle Park, NC, November 1975.
4. Compilation of Air Pollutant Emission Factors, Second Edition
(with Supplements 1-7), AP-42, U.S. Environmental Protection
Agency, Office of Air Quality Planning and Standards, Research
Triangle Park, NC, April 1977.
5. Control of Volatile Organic Emissions from Use of Cutback
Asphalt, EPA-450/2-77-037, (QAQPS 1.2-090), U.S. Environmental
Protection Agency, Office of Air Quality Planning and Stan-
dards, Research Triangle Park, NC, December 1977.
6. FAA Air Traffic Activity Reports (Annual), U.S. Department of
Transportation, Federal Aviation Administration, Washington, DC.
7. Airport Activity Statistics for Certified Route Air Carriers
Annual), U.S. Department of Transportation, Federal Aviation
Administration, Washington, DC.
8. Census of U.S. Civil Aircraft (Annual), U.S. Department of
Transportation, Federal Aviation Administration, Washington, DC.
9. Military Air Traffic Report (Annual), U.S. Department of Trans-
portation, Federal Aviation Administration, Washington, DC.
10. Census of Agriculture, Volume I, Area Reports (Annual), U.S.
Department of Commerce, Bureau of the Census, Washington, DC.
6-55
-------
11. C.T. Hare and K.J. Springer, Exhaust Emissions from Uncon-
trolled Vehicles and Related Equipment Using Internal Combus-
tion Engines. EPA Contract No. EHS 70-108, Southwest Research
Institute, San Antonio, TX, May 1973.
12. Ibid, Part 4, Small Air-Cooled Spark Ignition Utility Engines.
13. Motorcycle Statistical Annual. Motorcycle Industry Council,
Inc., Newport Beach, CA.
14. Survey of Motorcycles: Ownerships, Usage, and Maintenance,
Part III, Motorcycle Usage: Street vs. Off-Road. Motorcycle
Industry Council, Inc., Newport Beach, CA, October 1975.
15. Transportation maps of various states, prepared by U.S. Geo-
logical Survey for the Office of Policy and Program Develop-
ment, Federal Railroad Administration, United States Depart-
ment of Transportation, 1976.
16. Annual Railroad Reports prepared for the U.S. Interstate Com-
merce Commission.
17. Yearbook of Railroad Facts, (Annual), Association of American
Railroads, 1920 L Street, Washington, DC.
18. Waterborne Commerce of the United States. (Annual), U.S.
Army Corps of Engineers, Washington, DC.
19. Census of Business (1967) Selected Services Area Statistics,
U.S. Department of Commerce, Bureau of the Census, Washington,
DC.
20. Energy Outlook 1978-1990, Exxon Co., U.S.A., May 1978.
21. Residential and Commerical Area Source Emission Inventory
Methodology for the Regional Air Pollution Study"
EPA-450/3-75-008, Environmental Science and Engineering, Inc.,
EPA Contract No. 68-02-1003, Gainesville, FL, March 1973.
22. U.S. and World Energy Outlook Through 1990. U.S. Congres-
sional Research Service, Project Interdependence, Washington,
DC, November 1977.
23. Sales of Fuel Oil and Kerosene, Mineral Industry Surveys.
6-56
-------
24. Control of Volatile Organic Emissions from Perchlorethylene
"Dry Cleaning Systems, EPA-450/2-78-050, U.S. Environmental
Protection Agency, Research Triangle Park, NC, December 1978.
25. Control of Volatile Organic Emissions from Solvent Metal
Cleaning. EPA-45Q/2-77-022 (OAQPS 1.2-079). U.S. Environmental
Protection Agency, Research Triangle Park, NC, November 1977.
6-57
-------
7.0 VOC (AND NOX) ALLOCATION INTO CHEMICAL CLASSES
As discussed in Section 2.6 of Chapter 2.0, most photochemical
models require information about VOC composition in terms of desig-
nated groups or "classes" of compounds. Additionally, in some mod-
els, NOX may have to be specified as NO and N02« Each model's
classification requirements differ somewhat; hence, the following
discussion is necessarily somewhat general in describing how VOC
and NOX are allocated into classes. More specific details on
currently available models are included in Appendix A.
7.1 ALLOCATION OF VOC TO CLASSES
Generally, the basic annual inventory will contain estimates
of either total VOC or nonrnethane VOC, depending on what emission
factor information is used for computing emissions. The basic
approach for allocating VOC into the classes needed by a photochem-
ical model is to employ a set of "split factors" that distribute a
certain fraction of the VOC total into each class.
A simple example demonstrates the concept: assume a source
emits 10 tons of VOC per day, and that the photochemical model
being employed requires VOC emissions to be divided into three
classes. Given split factors for this particular source of 0.2,
0.5, and 0.3, simple multiplication of each factor by total VOC
yields the quantity of VOC in each class, in this case 2, 5, and 3
tons per day. These three totals, for instance, could represent
the quantities of paraffins, olefins, and aromatics, respectively,
composing the total VOC emissions from the particular source.* If
another photochemical model is used that requires six reactive
classes, the mechanics of performing the allocation would be the
*It should be noted that the "reactive classes", as designated by
the model designers, do not always coincide with the chemical
classes implied by their assigned names. For instance, aldehydes
do not fit, chemically, into any of the classes named above, but,
as VOC's, they must be allocated to one or another of these class-
es, to fulfill the requirements of the model. It should also be
7-1
-------
same except that six split factors would be needed rather than
three. This allocation step would, of course, have to be performed
for each emission total developed in the inventory using different
split factors appropriate for each source or source category.
As can be seen from the above example, the VOC allocation step
is not difficult once the split factors are available for each
source. The major difficulty in this process is determining what
split factors are most appropriate. Two basic approaches can be
followed for determining split factors. Ideally, VOC split factors
should be source specific. For example, because of the importance
of gasoline evaporation in VOC inventories, local gasoline composi-
tion data should be obtained corresponding to the summer season in
the modeling area. Additionally, source tests could be performed
to determine VOC species data for each refinery, chemical manufac-
turer, etc., and solvent composition data could be solicited from
each drycleaner, degreaser, etc. From this information, a photo-
chemical modeling specialist could determine appropriate split fac-
tors for each source.
Generally, however, with the exception of a few very large
point or area source emitters, it will not be practical for an
agency to collect such information, due to resource limitations and
because many facilities will not be able or willing to supply sol-
vent composition data. Hence, in most cases, the agency will have
to utilize generalized VOC species data from the literature in
order to develop model-specific VOC split factors. An example of
the kind of species data developed for a particular source catego-
ry—automotive exhaust—is shown in Table 7-1.* A photochemical
noted that photochemical models generally require mole rather
than weight quantities of each class. Hence, as is described
later in this chapter, a weight-to-mole conversion step is also
required during this allocation process.
*Table 7-1 shows, for the sake of example, the estimated composi-
tion of composite exhaust gases from gasoline-fueled vehicles for
a vehicle mix characteristic of the State of California in a
recent year, using fuels available at that time. This sort of
7-2
-------
C/)
LU
I
O
t«— I
*y-
UJ
^>-
<
3T
» — i
or
i
u_
o
a
^1^ fO
•^ CO
UJ
^>~ £^
§ 2
—[— (_5
»— 1 _J
31 _l
Oil
LJ_
a: o
u_
LU
00 H-
LU i— i
/ y"J ( f\
V / V t
CD O_
21
f— O
oo o
X LU
UJ =>
u_
1 1
O LU
•^
s*" l__l
O _l
i — i o
t; °°
ty> CD
0
0-
s:
o
o
t ^
LU
1—
<
s:
• — t
i—
LU
« — i
1
p~^
QJ
i— —
o
f—
C
O
•r—
i— -4->
0 0
E £
QJ -r-
f* frt
_L— W<
O {/)
C QJ
QJ E
O 3
QJ O
a. >
^J
C •(->
QJ .T
O O}
S- •!-
a> QJ
D- 3
t_
rt3
•— +->
3 JC
QJ -r-
r- QJ
O 3
2T
Q
«=C QJ
O T3
C£ 0
«=C 0
00
QJ •
C O
i
3
O QJ OJ
QJ > >
C -i- -i-
E E c E E c oooooi— «j -»->•»->
c^ c^. '[ C}_ 'i c^_ C C C C C | * -4-J 4-i 4_> -4-J C ^— C TO *O
CT» O ^^ »— ' CSJ LO LO r-H t — 1 LT5 O^ tD *3" O^ *O LO CO O *— ' « — * ^"
r-4cvjcMcor^coi-~tO'^-r~^ovooi-~r-o^ovoaio^-i
OJ ^ i— If— 1 «3" P^ r^ CO <— 1 i— 1 i— 1 LO Lf) tf (\J r- < CM
t— 1 ^H (XI
OOOOOOOOOOOOOOOOOOOQO
CM«DC\jrOOJi— lr-IP-«a->
CJ _^"
O) -Q -M
C f— > E
I +J •!-
>> 3 S^
X QQ 1— OJ i
cu a; a; fl3 'O QJ ^— QJ +J QJ &_ SL- t- C ^Q ^— ^— C QJ C
•4-5 C X O.- 4-5 C r— ^> C CT X QJ QJ QJ QJ r*— 03 >} CO C QJ
^ QJ QJ QJ (J O ^j O^ QJ QJ QJ £ £; £ 13 >»> E 4-> ^C" Q-ai o o o r— jr C Qj4->.c e
i i i i i i +-> i- zi i it/3t/>mo4->oOaj4->QJ
21 ^* ^! JT p^* ^^ tj ) Q_ CP ,»^ f— ( |— H t— H K— 1 |— Lf_l 1.1- (^ gr* QJ f*p
C\J C3 *"—> C^J f*O LO CO LO OO ^~ LjT5 CM LO f*^ C\J CO C^J tj0 %— 4 CNJ r-^
*"-' C\J CO CO CO CO C3 C3 rH C\J ^~ CD CD CD CD CD CD CD CD CD CD
CVJ C*J CSJ C\J CVJ C\J CSJ CVJ OJ C\J CM <—H *— * i— - 1 CM CM LO CM CM CM CM
CO CO CO CO CO CO CO CO CO CO CO LO LO LO LO LO CO CO CO CO LO
*3-«a-*fr*j-*s-*d-*j-«d-«d-«a-«*«i-*3-^-«3-*j-«!i-«d-«3-«3-*j-
o^cMOo^LOvo-cocno-
.- co co o CO t\J r-( vo
1-1 CM ro r~
*J^ LO *O
C C C E
O O O C E C 0
•r— f*- »r— OOO •*"
•4.) <4^> 4^ *r— >r— *r— 4_>
tj /O t3 t * ^ ^ ^ % (^
O O U *^ *p* *r— C^.
•f *i^- *r* (^. ^^ ^— >^"
I/I
OOO
4- M_ M- M-
O O O
•r—
CO
O
a.
E
O
XJ
C
3
O
o.
E
O
0
1— 1
CM
.—i
QJ
O
E
ai
s_
QJ
QJ
C£.
a
4->
(
»-
QJ
C)_
QJ
o:
-------
modeling specialist can utilize this type of data, in conjunction
with a knowledge of the model's reactivity requirements, to allo-
cate organic compounds from each source category to the correct VOC
classes, and thus develop split factors tailored to any model.
An example of "lumping" species into classes is also demon-
strated in Table 7-1. In the 7-class scheme shown at the bottom of
this table, paraffins are lumped together as one class (chemical
classification 1), olefins as class 2, aromatics as class 3, etc.
The weight and volume percentages associated with each organic com-
pound are summed for each classification and average molecular
weights computed. The resulting composite percentages after divi-
sion by 100 represent the split factors that can be used for allo-
cating emissions of this particular gas into this particular
7-class scheme. Both weight and volume percentages are shown in
Table 7-1, but for any specific model, only one set or the other
will actually be needed, depending on whether the model requires
emissions information in terms of the mass or number of moles of
VOC classes. An example of the general split factors utilized in
the San Francisco Bay Area's LIRAQ model, which uses a 3-class
scheme, is shown in Table 7-2 (Reference 2).
table is obtained by combining compositions of exhaust samples
from numerous individual automobiles, each weighted to represent
a part of the spectrum of vehicles actually in use in the state.
Similar estimates can be made for other years and other loca-
tions, given the necessary information as to the vehicle mix.
Table 7-1 is taken from the VOC Species Data Manual, published by
EPA in December 1978 (refer to Reference 1).This manual pro-
vides "emission profiles" like the example in Table 7-1 for]175
unique point and area stationary sources, as well as composite
profiles for mobile sources, giving for each a list of individual
VOC species with corresponding percentages, and a composition
summary in terms of seven VOC classes. Where no specific local
information exists, the profiles given in the manual may be use-
ful. However, for major point sources and important area
sources, locally applicable information should always be sought.
7-4
-------
Table 7-2. EXAMPLE VOC ALLOCATION FACTORS BY SOURCE CATEGORY
FOR A 3-CLASS SCHEME3
Source Type
Petroleum production
Refinery operations
Underground storage
Auto fill ing tanks
Fuel consumption
Waste burning fires
Heat treated coatings
Air dried coatings
Drycleaning- petroleum based
Dryclean ing- synthetic
Degreasing- trichloroethyl ene
Degreasing-1 , 1 ,1-trichl oro-
ethane
Rotogravure printing
Fl ex i graphic printing
Rubber, plastics, etc. mfg.
Pharmaceuticals
Misc. solvents
Gasoline exhaust
Gasoline evaporation (mobile)
Diesel exhaust
Gas turbine (jet)
Piston aircraft
HCi
("')
-
23
15
27
38
22
20
14.4
2
-
100
-
7
21
36
27
17
22
24
18
33
26
HC2
(%)
100
77
85
73
30
40
80
85
98
100
-
100
93
78
51.5
70
80
71
76
60
53
HC4
(*)
-
-
-
-
32
38
-
0.6
-
-
-
-
-
1
12.5
3
3
7
-
22
14
66 8
a As used in LIRAQ Model (Refer to Reference 2). Note that in
LIRAQ, in addition to the above split factors, each category is
adjusted by an additional factor (not shown) to account for
varying reactivities associated with each particular species mix.
7-5
-------
The so-called "carbon bond" mechanism, which is an option in
the Urban Airshed Model, requires an additional step beyond the
determination of split factors for VOC classes; the user must also
subdivide the VOC classes into fractions which will be assumed to
react in different ways in the mechanism's reaction set. Alterna-
tively, if the composition of the VOC emissions is given in terms of
individual compounds, each of these individual species will be
divided up into the corresponding fractions before entry into the
chemical reaction system of the model.
If the agency intends to employ a model that incorporates the
carbon bond mechanism, a photochemical modeling specialist should be
consulted to review all procedures, algorithms, and VOC species/
split factor data prior to initiating any data collection or alloca-
tion effort, and for advice on how to deal with other types of
VOC's. Indeed, such expertise should be engaged in this process
regardless of what model is applied or what approach is used to
allocate VOC into reactive classes. In no case should the agency
develop split factors or carry out such an allocation without
knowing what photochemical model will be run or what classification
scheme is needed to meet the model's reactivity requirements.
7.2 SPECIFICATION OF NOX AS NO AND N02
As indicated in the description of various photochemical models
given in Appendix A, some models do not require that nitrogen oxides
be distinguished as either nitric oxide or nitrogen dioxide. Instead,
these models assume that all NOX is NO, which is the predominant
form of NOX emitted from all combustion processes.
For those models that do require this split to be made, split
factors are applied in the same manner as are VOC split factors.
That is, for each source or source category emitting NOX, two per-
7-6
-------
centages (totaling 100 percent) need to be defined: one correspond-
ing to the fraction of NOX emitted as NO and the other correspond-
ing to the fraction emitted as fM^- In this sense, allocating
NOX into NO and N02 is analogous to utilizing a 2-class scheme
for allocating VOC.
The mode of expression of the necessary split factors may vary
from one model to another. NOX emissions are ordinarily expressed
"as N02", which means that a molecular weight of 46 is attributed
to NO as well as N02» even though the true molecular weight of NO
is 30. Many models take account of this convention by accepting
split factors totaling 100 percent for NOX "as N02", but some
may require NO emissions to be expressed in terms of the true weight
of NO. The true value for NO is 30/46 or 0.65 times the conventional
value of NO "as N02".
If the inventory maintains NO totals as NO, then care should be
taken that a molecular weight of 30 is used when computing moles of
NO for use in the photochemical model. It is important, during the
planning stages, to review the annual inventory to determine how
NOX is reported and to consult with a modeling specialist to find
out how the photochemical model accepts NOX emissions data.
As an example of NOX allocation, consider a power plant that
emits NOX equivalent to 1,000 kg of N02 Per hour. Given split
factors of 95 and 5 percent by weight "as N02", then NO and N02
emissions would be equivalent to 950 and 50 kg "as N02" per hour,
respectively; however, the actual emissions of NO would be only
30/46 of 950, or 620 kg per hour.
!
At present, few references are available that define split
factors for allocating NOX into NO and N02« Reference 3 yields
such data for many types of stationary combustion sources. As a
7-7
-------
rough average, 97 percent (by weight as Nf^) of the NOX emitted
from most boilers will be NO (as
7.3 PROJECTING VOC (AND NOX) SPLIT FACTORS
Just as the quantity of emissions may change in an area from
the base year to any projection year, the composition of these emis-
sions may change, as well. To reflect this, the VOC and NOX split
factors used in the model may need changing in each projection year,
at least for important sources for which such a change can be esti-
mated. One source for which this may be an important consideration
is motor vehicles. With an increasing fraction of the vehicle mix
expected to consist of catalyst-equipped cars, whose emissions con-
tain a higher percentage of methane and light paraffins than uncon-
trolled cars, significantly different VOC split factors may be
appropriate in projection years. Similarly, if significant changes
are expected in the compositions of petroleum products transported
and stored in the modeling area, such changes should be reflected in
the projection year split factors, as well. Of course, for many
sources, no changes in emission composition will be expected. For
instance, no change would be needed for my sources that will use
the same solvents in the base year and projection years (e.g.,
drycleaners using perchloroethylene). Likewise, since no evidence
suggests that NO/N02 ratios in combustion emissions will change in
the near future, the same NOX Split factors could also be used in
projection years.
In no case should different split factors be used in any pro-
jection inventory unless they reflect anticipated changes in the
composition of future emissions. Any other changes ma> cause the
photochemical model to predict changes in ozone concentrations that
are due simply to differences in methodology and are unrelated to
expected real effects of composition changes.
7-8
-------
7.4 COMPATIBILITY WITH INVENTORY DATA AND SOURCE CATEGORIES
An important consideration when allocating VOC into classes is
one that may be easily overlooked. Split factors must be given in
terms compatible with those used to express VOC totals in the basic
inventory. Ordinarily, this means that they should be given as
fractions of total VOC, including methane and any other organic com-
pounds considered unreactive, if such compounds are present in the
emissions for the category under consideration. However, if the
basic inventory has been compiled in terms of nonmethane hydrocar-
bons (NMHC) or reactive volatile organic compounds (RVOC), the split
factors should be given as fractions of these totals rather than of
total VOC.
For example, the split factors shown in Table 7-1 are immedi-
ately applicable to total emissions given as total VOC (including
methane). If they were to be applied (erroneously) to total NMHC
emissions estimates, the resulting emission estimates in each VOC
class would be underestimated, since a smaller total would have
been allocated.
In this case, the given split factors would have to be adjusted
to make them applicable to nonmethane VOC emissions. This can be
readily accomplished by dividing each of the weight (or volume) per-
centages corresponding to each chemical classification by the quan-
tity [(1 - percent methane/100)] where "percent methane" is the
weight (or volume) percent corresponding to classification 6. As an
example of this, referring to Table 7-1, the existing weight percent
associated with classification 1 in motor vehicle exhaust is 28.6.
To convert this to a split factor that would be applicable to a
nonmethane emission weight total, it would be divided by the frac-
tion of nonmethane VOC, namely l-(8.7/100) (since 8.7 is the weight
percentage of methane, classification 6) to yield 31.3, then by 100
7-9
-------
to convert percentage to a fraction, 0.313. The other split factors
would be adjusted accordingly. The sum of all the converted non-
methane split factors should be 1.0 if these conversions have been
performed correctly.
If the existing inventory is not in terms of total VOC or total
nonmethane VOC, and instead is divided into some sort of classifi-
cation scheme that is incompatible with the scheme needed by the
photochemical model, it is questionable whether such an inventory
will be useful as input to the photochemical model. A photochemical
modeling specialist should be consulted if this situation exists.
As suggested in Section 2.5 of Chapter 2.0, another important
issue is split factor-source classification compatibility. It may
often happen that the source categories and subcategories chosen for
the basic inventory fail to distinguish between sources having sub-
stantially different emission compositions, which therefore require
different sets of split factors. For example, area source degreas-
ing may be considered as a single category, but different degreasing
solvents may be used in different plants. There are several pos-
sible solutions to this problem.
First, any individual plants which emit significant amounts of
solvent vapors from degreasing operations can be treated as point
sources, in which case the solvents used at each should be identi-
fied and entered in the point source inventory.
Second, if there are many degreasing operations, each of which
emits only relatively small amounts of solvent vapors, it may be
possible to determine from solvent suppliers how much of each sol-
vent is used in the region. Lacking any information to the con-
trary, the agency may then assume that the emissions of each sol-
vent are uniformly spread through the industrial areas of the
region. In this case degreasing can be treated as a single area
7-10
-------
source category, with split factors in proportion to the overall
supply; or, alternatively, it can be subdivided into several area
source categories, one for each solvent, with the emission totals as
determined from the suppliers.
Third, if there is no locally available information, state or
national supply statistics can be used to provide estimates of the
split factors or the subcategory totals, to be used in the manner
previously explained.
A similar situation may be encountered in dealing with motor
vehicle emissions; split factors may be known for exhaust and evap-
orative VOC's separately, whereas the basic inventory may only
I
provide a single lumped estimate combining these two emission compo-
nents. Since this source is a very important one in most urban
areas, it is best to recalculate these emissions in terms of the two
subcategories, then apply the appropriate split factors to each
category. This enhances the accuracy of the VOC allocation process
and facilitates the use of the model to evaluate control strategies
that may affect exhaust and evaporative emissions differently.
There will also be cases in which compositional information is
simply not available in sufficient detail to permit determination of
split factors for many recognized VOC sources. For instance, a
petroleum refinery may be represented in the inventory by a large
number of point sources (having, for example, different NEDS source
classification codes) while only a single set of split factors is
available for the entire point source category (as, for example, in
Table 7-2).
For less significant VOC sources, such as area source fuel com-
bustion, there is correspondingly less need for accuracy in assign-
ing split factors, since moderate errors in these small contribu-
7-11
-------
tions will result in only very small errors in the individual VOC
class totals. A single set of split factors is, therefore, adequate
for all external combustion sources operating on a given type of
fuel, and no subcategories would be necessary or useful in this
case.
In general, it is important to review the source category list
at an early stage of the planning to ensure that all subcategories
essential for proper allocation of emissions to VOC classes have
been recognized and established for data collection. This is
especially important if special surveys or questionnaires are to be
utilized, because failure to retrieve all the necessary information
in the initial contact can seriously impair the productivity of the
effort.
7-12
-------
References for Chapter 7.0
1. H. W. Bucon, J. Macko, and H. J. Taback, Volatile Organic
Compound (VOC) Species Data Manual, EPA 450/3-78-119 U.S.
Environmental Protection Agency, Research Triangle Park, NC,
December 1978.
2. M. C. MacCracken and G. D. Sauter, eds., Development of Air
Pollution Model for San Francisco Bay Area, Volumes I and II,
Livermore Laboratory, UCRL-51920, Livermore, CA, 1975.
3. R. J. Milligan, et al., Review of NOx Emission Factors for
Stationary Combustion Sources, EPA-450/4-79-021, U.S.
Environmental Protection Agency, September 1979.
7-13
-------
8.0 DATA HANDLING
8.1 GENERAL
The basic distinction between the annual inventory and the
detailed inventory used for photochemical modeling is that, in the
latter, a greater degree of temporal and spatial resolution is
needed and VOC (and NOx) have to be apportioned to classes. The
preceding chapters in this document have largely centered on the
conceptual aspects of incorporating this additional resolution into
the annual emission inventory. Up to this point, however, nothing
has been said concerning how this conversion is best accomplished
from a data management standpoint. Obviously, since many repeti-
tive data manipulations are required to complete the photochemical
modeling inventory in a large urban area, a detailed consideration
of data handling is appropriate prior to initiating the inventory
effort. This chapter focuses on the primary data handling activi-
ties needed in the detailed inventory effort, and discusses how
each data handling step can be facilitated. Because the data
handling considerations are quite different for point, area, and
highway vehicle sources, a separate discussion is included for each
in subsequent paragraphs.
It should be noted that although most data handling will typ-
ically be computerized to the extent possible, some steps will nev-
ertheless still have to be performed manually. This distinction
will be emphasized in the following discussion for those data hand-
ling steps that are better carried out either one way or the other.
The final product of the detailed inventory effort is the
modeler's tape used as input to the photochemical model. A sepa-
rate modeler's tape must be prepared for each base year and projec-
tion scenario that is to be evaluated. A brief discussion of a
generalized modeler's tape format is included in this chapter.
8-1
-------
8.2 POINT SOURCE DATA HANDLING
When assigning point source emissions to specific grid cells,
either a manual or computerized approach can be utilized. Gener-
ally, unless there are very few point sources in the modeling area,
a computerized assignment is more practical. In this approach, a
program is written that compares the UTM (or other) coordinates
stored in each point source record with the coordinates of the grid
cells and determines in which specific grid cell the point is
located. The appropriate grid cell identifier (or coordinates) can
either be written on the point source record (if space is avail-
able) or else a separate correspondence file can be created con-
taining this data. Subsequently, any time a modeler's tape is gen-
erated, this point-to-grid-cell correspondence information can be
accessed to assign point source emissions to grid cells. It would
be possible to generate the point-to-grid-cell correspondence data
during the creation of the modeler's tape; however, this alterna-
tive would have the disadvantage of requiring the coordinate com-
parison step to be repeated for every modeler's tape created.
As mentioned in Section 4.2, the point-to-grid-cell assignment
step can also be performed manually by overlaying an outline of the
grid system onto a map showing the point source locations. If this
is done, the resulting correspondence data will still have to be
loaded onto either the point source records or a machine-readable
correspondence file, as described previously, in order to be
utilized by the program that creates the modeler's tape. The
manual approach, as mentioned previously, has the disadvantage of
being much more time-consuming if numerous point source assignments
are necessary.
Some photochemical models do not require that elevated point
sources be assigned to grid cells. In these models, preprocessor
programs are available that make this assignment based on the point
8-2
-------
source coordinates available from the annual, countywide inventory.
If this is the case, point-to-grid-cell correspondences need not be
determined for these particular sources. Generally, however, since
the modelers may not know in advance which sources will be consid-
ered as elevated, and since computerized assignments will be prac-
ticed in most instances, little extra effort will be expended in
simply making this assignment for all point sources. Thus, this
information will always be available in case it is needed at a
later date.
Hour-by-hour point source emissions are estimated by applying
seasonal, daily, and hourly operating factors to the annual emis-
sions. This concept was discussed in Chapter 4. In the data hand-
ling system, the temporal factors and the resulting hourly emissions
can be stored either in the individual point source records (if
space is available) or on a separate file created for this purpose.
A potential disadvantage of storing hourly emissions on each point
source record is that a great deal of file space is required. An
alternative is to have the program that creates the modeler's tape
compute hourly emissions at the point source level but simply ac-
cumulate hourly emissions at the grid cell level. If this alterna-
tive is employed, VOC and NOX splits would have to be applied to
each point prior to accumulation at the grid cell level in order
not to lose the pollutant split identity of each source category.
A disadvantage of this latter approach is that hourly emissions and
pollutant splits must be recomputed each time a modeler's tape is
created. An example of a file containing point source temporal
factors for calculating hourly emissions from annual emissions is
shown in Table 8-1. The entries for such a file have to be deter-
mined manually using the procedures outlined in Chapter 4. Differ-
ent temporal patterns can be simulated (e.g., in projection inven-
tories) by simply changing the factors in this file to reflect
anticipated operating rate changes.
8-3
-------
Table 8-1. EXAMPLE TEMPORAL FACTOR FILE FOR INDIVIDUAL
POINT SOURCES (EXCERPT)
Source
SCCa Ptb
30200000 02
30200000 02
30200000 02
30200000 02
30200000 02
30200000 02
30200000 03
30200000 03
30200000 03
30200000 03
30200000 03
30200000 03
Temporal Factors0
1.43
2.29
2.78
5.23
5.23
1.76
2.99
2.99
5.77
4.28
2.29
3.21
5.23
5.23
2.99
3.63
5.77
4.28
27.0
2.29
3.92
5.23
5.23
32.0
2.99
4.28
5.77
4.28
2.29
4.58
5.23
5.23
2.99
4.28
5.77
4.28
2.29
5.23
5.23
5.23
2.99
4.28
5.77
3.63
2.29
5.23
5.23
3.76
2.99
5.00
5.00
2.99
Coded
S
D
1H
2H
3H
4H
S
D
1H
2H
3H
4H
aPlant identification by SCC code (eight digits).
bpoint source identification by number within plant.
(Note that state, county and plant ID's are not shown
and should be included for complete identification.)
Six successive lines constitute one point source record.
Definitions: Seasonal, percentage of annual activity
occurring during chosen quarter-year; daily, percentage
of seasonal activity occurring on selected day; hourly,
percentage of daily activity occurring on selected hour.
Six consecutive hourly values appear on one line.
dCode: S, seasonal; D, daily; 1H, hourly, 0001 to 0600.
2H, 0601 to 1200. 3H, 1201 to 1800. 4H, 1801 to 2400.
8-4
-------
The allocation of VOC (and NOX) into classes is similar to
the allocation of annual emissions into hourly increments. Basi-
cally, as is conceptually described in Chapter 4, the VOC (and
NOX) emissions from each point source are multiplied by "split
factors" to allocate them into classes. A separate file of split
factors can be created for this purpose, an example of which is
shown in Table 8-2, or the split factors can be stored as part of
the point source records. Molecular weights should be stored
similarly.
It should be noted that the file shown in Table 8-2 gives
split factors by source category (i.e., at the SCC level) and not
for each point source. This approach is generally recommended,
since (1) specific split factors will not be known for most indivi-
dual facilities, and (2) considerably less file space will be need-
ed. A disadvantage of this approach is that it makes it difficult
to reflect specific split factor information that may be available
for individual operations.
As in calculating hourly emissions, estimates of VOC (and NOx)
by class can either be (1) computed prior to the generation of the
modeler's tape and stored in the point source records or in a sepa-
rate file, or (2) computed during the creation of the modeler's
tape, to conserve file space. A disadvantage of the latter approach
is that VOC (and NOx) emissions by class have to be recomputed each
time a modeler's tape is created. In any projection inventory, new
VOC (and NOx) split factors can be reflected simply by changing the
split factor files and applying them to the projected VOC (and NOx)
emission totals.
Of course, if the model being used employs carbon bond chemis-
try, special algorithms generally will have to be executed subse-
quent to applying VOC split factors. These algorithms are fairly
simple; hence, no special data handling problems should be encoun-
tered in incorporating them into the VOC allocation routine.
8-5
-------
Table 8-2. EXAMPLE "SPLIT FACTOR" FILE (EXCERPT)
Source
Category3
sec
30600801f
30600802
30600803
40300106
40300107
40300152
40300205
30000606
30000608
Pollutant
Codeb
HC
HC
HC
HC
HC
HC
HC
NX
NX
Class lc
SFd MWe
34.5
13.8
5.0
13.9
13.9
1.1
1.1
85.0
85.0
46
46
52
58
58
38
38
30
30
Class 2C
SF MW
56.9
75.0
84.0
73.5
73.5
57.3
57.3
15.0
15.0
61
81
87
61
61
68
68
46
46
Class 3C
SF MW
7.9
3.9
6.6
11.2
11.2
37.0
37.0
71
72
72
72
72
31
31
Class4c
SF MW
.7
7.3
4.4
1.4
1.4
4.6
4.6
96
92
96
92
92
101
101
aSource category by SCC code (eight digits).
bCode: HC, VOC; NX, NOx.
CC1asses are defined as follows:
for VOC, class 1 is nonreactives; class 2, paraffins;
class 3, olefins; class 4, aromatics
for NOx, class 1 is NO; class 2, N02
^Split factor, percent of total, by weight.
eAverage molecular weight.
^Each line constitutes a record, either for VOC
or for NOx, for one source category.
-------
8.3 AREA SOURCE DATA HANDLING
The major difference between area source data handling and
point source data handling concerns the way emissions are estimated
at the grid cell level. Since point source locations are typically
known to the nearest tenth of a kilometer, it is easy to assign
them to specific grid cells. Area source emissions, however, are
typically only resolved to the county (or equivalent) level in an-
nual inventories and, thus, must be disaggregated to the grid cell
level using apportioning procedures such as are described in Chap-
ter 6.
Area source apportioning factors can be stored in a special
file, an example excerpt of which is shown in Table 8-3. This file
is basically a matrix of apportioning factor values by grid cell.
In this table the surrogate indicators are designated along the
top, and the grid coordinates are presented along the side. The
values loaded into the table represent the fraction of the county-
wide total of each variable that is located within each particular
grid cell. (Such a table would have to be prepared for each county
equivalent level for which area source emissions are resolved in
the annual inventory.) Hence, in order to determine emissions from
a particular area source in a given grid cell, the calculation pro-
gram (1) determines what surrogate indicator is appropriate for the
source in question (this information would be written into or
supplied to the emission calculation routine), (2) accesses the
apportioning factor file to determine what fraction corresponds to
the grid cell/surrogate indicator combination in question, and then
(3) multiplies that fraction by the countywide emission total for
the particular area source.
The sequence of steps described in the previous paragraph ap-
plies to inventories in which each area source category is appor-
tioned by only one surrogate indicator. If an area source category
is to be apportioned by multiple surrogate indicators, the same
8-7
-------
Table 8-3. EXAMPLE FILE OF GRID CELL APPORTIONING FACTORS FOR
AREA SOURCES (EXCERPT)
Grid Cell
Coordinates3
272,784
274,784
276,784
278,784
280,784
252,786
254,786
256,786
258,786
260,786
Apportioning Factors for
SIlb SI2b SI3b SI4b SI5b SI6b
.001
.001
.001
.001
.001
.001
.011
.013
.001
.001
.001
.001
.001
.001
.001
.002
.011
.014
.001
.001
.001
.001
.001
.001
.001
.002
.012
.015
.001
.001
.0
.0
.0
.0
- .0
.0
.0
.0
.0
.100
.004
.004
.004
.004
.003
.004
.0
.0
.004
.004
.0
.0
.0
.0
.0
.0
.045
.270
.009
.0
aUTM coordinates of grid cell, SW corner, Km.
^Apportioning factors for this example are based on
the following surrogate indicators: SI1, employment;
SI2, commercial employment; SI3, dwelling units;
SI4, general aviation; 515, open burning;
SI6, vehicle miles traveled.
The entry in each case is the fraction of the total
indicated activity which occurs in the grid cell.
8-8
-------
procedure can be followed by creating new subcategories corres-
ponding to the level of activity to be apportioned by each indica-
tor as discussed in Section 6.2.2.1.
Thus, referring again to the example given in Section 6.2.2.2,
if area source degreasing emissions are 42 tons per day of VOC and
these are estimated to arise 10 percent from MFG4 activities, 60
percent from MFG5 activities, 20 percent from retail service activ-
ities, and 10 percent from other service activities, the area
source degreasing category can be partitioned into four subcate-
gories, having respective totals of 4.2, 25.2, 8.4, and 4.2 tons
per day. Each subcategory would then be apportioned to the grid
cell level by its appropriate surrogate indicator as previously
described. These subcategories would appear in the area source
apportioning factor file, but not necessarily in the emissions
inventory provided by the operating agency.
Where the agency chooses to allocate area source emissions
based on zonal statistics on population, employment, etc., instead
of using land use maps, a data handling procedure is necessary to
convert the zonal level apportioning information to the grid cell
level. As discussed in Section 6.2.2.2, the first steps in this
process are to overlay a map showing the grid system boundaries
over a map showing the zonal boundaries, and then determine or
estimate fractions of zonal areas lying within specific grid cells.
These area! fractions are then loaded into a machine-readable file
to serve as zone-to-grid-cell correspondence values. This file, in
turn, can be used to generate grid cell apportioning factors by (1)
multiplying the surrogate indicator values available at the zonal
level (e.g., from forecasting models) by the areal fractions for
each zone, (2) summing over all zones, and (3) normalizing, as
shown in the equations given in Section 6.2.2.2. The latter steps
should be computerized because of the great amount of data handling
involved when hundreds of zones and grid cells are involved.
8-9
-------
The estimation of hourly area source emissions is accomplished
using essentially the same data handling procedures as are de-
scribed in Section 8.2 for point sources. Basically, a file of
seasonal, daily, and hourly temporal factors is created (similar in
format to Table 8-1) that can be multiplied by the annual area
source emissions to generate hourly emission estimates. Typically,
one set of temporal operating factors will be loaded into this file
for each area source category, and will be applicable for the en-
tire modeling area. The determination of the appropriate factors
for the area source temporal factor file is a manual procedure, and
is described in Section 6.3.
The allocation of area source VOC (and NOx) into classes is
also similar to the allocation of point source emissions into
classes. Generally, a file of area source split factors can be
created for this purpose (similar to that shown in Table 8-2) or,
alternately, these split factors can be stored in the area source
records themselves if space is available. These split factors are
simply multiplied by the VOC (and NOx) totals to estimate emissions
by class. If a model employing carbon bond chemistry is employed,
special algorithms may have to be applied subsequent to this step
to estimate the number of moles of various carbon bonds.
In area source projection inventories, anticipated changes in
land use patterns, temporal patterns, and VOC (and NOx) split fac-
tors can be readily accommodated by simply changing the values in
all the above files and then applying them to the projected inven-
tory of annual, countywide emissions. Any changes in these files
would, of course, have to be manually determined for each projec-
tion year.
8.4 HIGHWAY VEHICLE DATA HANDLING
As described in Chapter 5, much of the development of the
detailed photochemical inventory of highway vehicle emissions is
8-10
-------
accomplished using available DOT and EPA computer packages. Spe-
cifically, either the UTPS or PLANPAC/BACKPAC batteries will typi-
cally be employed for developing link/zone VMT estimates; HWYEMIS1,
APRAC-2, or SAPOLLUT will be used as network emission models to
calculate link/intrazonal VMT emissions; and MOBILE1 will be used
to generate highway vehicle emission factors used in the network
models. Because each of these computer packages is described in
Chapter 5, and since complete user's documentation is available
thereon, no additional discussion is included here. The reader
should consult the references cited in Chapter 5 for data handling
details on these available packages.
The transportation, emission factor, and network emission
models described in Chapter 5 generally yield temporally resolved
highway vehicle emission estimates at the link/zonal level. Cur-
rent network models can also provide gridded emission estimates;
however, this capability is of limited usefulness as considerably
more spatial resolution is required in detailed photochemical in-
ventories than these models provide. Hence, the agency responsible
for the detailed highway vehicle inventory will have to develop
link/zone-to-grid-cell conversion routines before creating the
modeler's tape. Emissions for each link are apportioned to grid
squares by assuming that the emissions are uniformly distributed
along a straight line. The fraction of the link-segment that falls
within a given grid square determines the fraction of the total
emissions for that link that fall in that square. The contribu-
tions of all links are summed for each square. Zonal emissions are
aggregated at the grid cell level by developing an equivalency
table designating the fraction of each zone lying within each grid
cell, and assigning emissions accordingly.
Another function not performed by the available FHWA/EPA
transportation packages is the allocation of VOC (and NOx) emis-
sions into classes. This step can be carried out similarly to the
-------
procedure described in Section 8.2, with separate split factors
used for each vehicle category for which an emission total is car-
ried along by the network emission calculation model. Ideally, as
is done in some areas, VMT and emissions are calculated separately
for each major vehicle type (e.g., LDV, HDV, etc.), in which case
vehicle-type-specific VOC and NOx split factors can be applied. In
other cases, however, network models may supply only composite
emissions for all vehicles, in which case composite split factors
will have to be applied based on the fraction of travel by each
vehicle type.
When making the VOC and NOx splits for light-duty vehicles, it
is also important to note whether the inventory distinguishes ex-
haust from evaporative emissions. Since the VOC compositions of
the exhaust and evaporative components are significantly different,
this distinction is desirable and should be maintained through the
VOC allocation process. MOBILE1, as an option, classifies highway
vehicle emissions into the exhaust and evaporative components.
8.5 MODELER'S TAPE CREATION
The final emission inventory product needed in data-intensive
photochemical models is the modeler's tape. This is a file of
hourly, gridded emissions for each pollutant (including VOC and NOx
classes) that is formatted to interface with the photochemical
model. As mentioned previously, separate modeler's tapes are needed
for each inventory scenario that will be modeled.
Each photochemical model requires different input information
for its modeler's tape and formats vary appreciably. In addition,
some models have preprocessor programs that convert certain data
items to model format. Also, less data-intensive models may accept
daily rather than hourly gridded emissions, together with diurnal
distributions as estimated for broad groups of categories. Because
8-12
-------
of the wide variety of models available, specific details of the
input data and format requirements are beyond the scope of this
document.
A generalized modeler's tape format is shown in Figures 8-1
and 8-2. Figure 8-1 shows a typical format for elevated point
sources, while Figure 8-2 shows a ground level source format. The
distinction between elevated and ground level sources is made in
some models in order to incorporate some measure of vertical reso-
lution. There are no hard-and-fast rules regarding what sources
need to be considered as elevated sources on the modeler's tape.
To some extent, this decision may be made by the photochemical
modeler after reviewing point source data. Because a great amount
of file space is required on the modeler's tape to store all of the
information needed on elevated sources, it is generally desirable
to include many of the smaller, low level point sources as ground
level sources.
It should be noted that the ground level file on the modeler's
tape combines low level point, area, and highway vehicle emissions
for each hour and for each grid. Hence, the program that creates
the modeler's tape must be capable of merging emissions from these
three types of sources at the grid cell level. Elevated point
sources may or may not have to be assigned to the grid cell level
during the creation of the modeler's tape, depending on the model
and depending on the availability of a preprocessor program that
will make this assignment. In the example elevated source format
shown in Figure 8-1, only the UTM coordinates for each point source
are needed, which a preprocessor program subsequently uses to as-
sign emissions to the proper grid cell. Model documentation should
be consulted during the planning stages to determine what data are
needed in the modeler's tape, in what format, and what steps are
accomplished by the preprocessor program available with the model
being used.
8-13
-------
81
"
Is
CL. W")
1
1
1
mm
CM
§
0
cc.
\
—
on
y £
J- 1/)
•"T
LU
1
1
1
—
•M
CVJ
v^S
gg
O
2=
,_,
•3-
z
^P
f- 3
(/> £
u.
"
u_
°3
Ko
g?
§ C/>
1
1
H
1
_
wm
•M
••
MMI
»••
h*"1
•^
-
_
_
^
MM
HH
^
fS
C/>
HYDROCARBON
CLASSES,
MOLES
SS
£S
X
9^
5S
H-
a: o
ss
3UjS
O^S
cdS<^-
z
^^"
(O
> OJ
O) O
i— (O
S- 10
o s-
q- O)
S- T5
3 O
•MS:
o
3 c
s- -i-
-»->
oo in
QJ
cu u
^- S-
•r- 3
U. O
00
i— O
O-D-
E
to
X
I
CO
01
S-
Ol
8-14
-------
800
V}
Cd t-«
a;
>
CD
3
O
CD
s-
o ai
H- Q-
ea
CU h-
3 I/)
4-> S-
O to
r— tO
O.T-
fO UJ
X
IXI
I
CO
a>
^
O)
8-15
-------
APPENDIX A
URBAN SCALE PHOTOCHEMICAL
AIR QUALITY
SIMULATION MODELS
-------
URBAN SCALE PHOTOCHEMICAL AIR QUALITY SIMULATION MODELS
Urban scale photochemical air quality simulation models are
briefly discussed in the following pages. All of these models are
or soon will be available for general use, either from the National
Technical Information Service (NTIS), EPA, or the California Air
Resources Board.
The models are characterized according to (1) the theoretical
approach taken toward simulation of the atmospheric system; (2) the
chemical reaction mechanism used to simulate the rates of appearance
and disappearance of the reactive chemical species present in the
atmosphere; and (3) the format provided for input of emissions in-
formation.
The theoretical approach is, basically, either Lagrangian or
Eulerian. In the Lagrangian approach, computational complexity is
reduced through decoupling of the chemical and fluid-dynamic aspects
of the simulation; this is achieved by focusing attention on the
chemical processes within individual moving air parcels. In the
Eulerian approach, advantage is taken of the extensive capabilities
of modern computer systems in order to simulate both chemical and
fluid-dynamic aspects in all parts of the atmosphere simultaneously.
Lagrangian models are often called "trajectory" models because of
the requirement of identifying plausible trajectories for the air
parcels which are to be simulated; Eulerian models are often called
"grid" models, because the fluid-dynamic aspects are defined with
respect to a gridded map covering the simulated region.
Chemical reaction mechanisms are defined in terms of a limited
set of reactive chemical species which are presumed to be subject
to a limited set of "elementary" chemical reactions. The species
involved include ozone precursors, i.e., volatile organic compounds
A-l
-------
and oxides or nitrogen; secondary pollutants, including ozone, that
are produced by the chemical reactions; and intermediate or "non-
accumulating" species, such as free atoms of oxygen and free hy-
droxyl radicals, which on theoretical grounds are believed to be
present, but at concentrations so low that ordinary chemical pro-
cedures do not detect them. The format for representation of the
reactive VOC's is of particular importance to the user, since he
will be required to apply different sets of "split factors" for use
in different models, depending on the chemical species used in the
models to simulate the reactive VOC's. The NOx format is equally
important but less troublesome, as only two species are involved--
nitric oxide (NO), and nitrogen dioxide (N02).
Gridded information on emissions is required for all photochem-
ical air quality simulation models. To use a Lagrangian model in
reference to a selected trajectory, emissions input is needed only
for the area immediately surrounding the trajectory. However, for
most practical applications, enough trajectories will be simulated
so that almost all of the areas in the available emissions grid will
be involved; therefore, it is preferable to provide the complete
gridded emissions inventory at the outset. To use an Eulerian
model, the complete gridded inventory must always be provided.
Diurnal emissions are represented, in the input to Eulerian
models, by providing detailed hourly information for each grid cell
and for each separately identified point source. For input to
Lagrangian models, in some cases, diurnal variation may be simulated
by estimated hourly distributions that apply to sources in par-
ticular categories, as described in the user's guides. In such
cases, the gridded emissions input is usually formatted in terms of
emissions for a 24-hour period.
A-2
-------
The purpose of this appendix is to describe, briefly, the main
features of currently available models. For further details, the
user's guides provided with the models are usually adequate;
however, the originating firms or organizations will often provide
consultation with regard to particular problems that may be brought
to their attention.
Model Designation: REM/TRACE
Developed by: Pacific Environmental Services, Inc.
1930 14th Street
Santa Monica, California 90404
(213) 450-1800
Model Description
REM and TRACE are trajectory models in which a single vertical
cell is assumed to reach from ground level to the mixing height,
with instantaneous mixing. The two models have different chemical
reaction mechanisms. TRACE includes a module which simulates the
effect of horizontal diffusion of pollutants out of the moving cell.
Computer requirements are moderate.
Chemistry
REM contains a 34-reaction mechanism based on two classes of
hydrocarbons; less reactive and more reactive (methane excluded).
TRACE contains a 76-reaction mechanism based on two classes of
hydrocarbons, represented in terms of surrogate species, propylene
and butane. Methane is excluded. This mechanism is the one used in
EPA's Empirical Kinetic Modeling Approach (EKMA).
A-3
-------
Emissions Source Categories
Both REM and TRACE accept emissions from categories labeled (1)
freeways, (2) streets, (3) area sources. All emissions except those
from motor vehicle traffic are included among the "area" sources.
Emission from point sources are attributed to the grid square in
which they originate, unless their effective stack height is greater
than the mixing height (in which case they are deleted).
Emissions in each category are specified in terms of a daily
total of non-methane hydrocarbon for each grid square and a diurnal
distribution in terms of percent of the daily total emitted in each
hour. For each grid square, the fraction of more reactive hydrocar-
bon in each category is also entered. NOx emissions are similarly
entered in terms of daily totals and hourly percentage distribu-
tions; they are taken to be 100 percent NO. CO may also be entered,
although it has little or no effect on the concentrations of photo-
chemical pollutants.
All point sources are included in overall emissions for each
grid square.
References:
1. L. G. Wayne, A. Kokin, M. I. Weisburd (1973) Controlled Evalua-
tion of the Reactive Environmental Simulation Model (REM), Vol-
ume I: Final Report. Pacific Environmental Services, Inc.,
Santa Monica, California, EPA R4-73-013a, NTIS No. PB 220 456/8.
2. A. Kokin, L. G. Wayne, and M. I. Weisburd (1973) Controlled
Evaluation of the Reactive Environmental Simulation Model (REM),
Volume II: User's Guide. Pacific Environmental Services, Inc.,
Santa Monica, California, EPA No. R4-73-013b, NTIS No. PB 220
457/6.
3. P. J. Drivas and L. G. Wayne (1978) Sensitivity Tests of a
Largrangian Photochemical Air Quality Simulation Model. Paper
No. 78-10.3, APCA Annual Meeting, Houston, Texas.
A-4
-------
Model Designation: DIFKIN/ELSTAR
Developed by: Environmental Research and Technology, Inc.
2625 Townsgate Road
Westlake Village, California 91360
(213) 889-5313
Model Description
DIFKIN and ELSTAR are trajectory models in which a columnar air
parcel comprising one to five vertically stacked cells is assumed to
reach from the ground level to the mixing height. The two models
have different chemical reaction mechanisms, and ELSTAR includes
algorithms to simulate the loss of pollutants at the ground. Com-
puter requirements are moderate. (DIFKIN is now considered
obsolete.)
Chemistry
DIFKIN contains a 16-reaction mechanism with one class of reac-
tive hydrocarbon.
ELSTAR contains a 62-reaction mechanism with six VOC classes,
designated as alkanes, ethylene, other alkenes, aromatics,
formaldehyde, and other aldehydes.
Emission Source Categories
Both DIFKIN and ELSTAR accept emissions from categories labeled
freeways, streets, and area sources. In addition, point sources
are individually considered in ELSTAR. Diurnal distributions of
hourly emissions are specifed for freeways, streets, power plants
and refineries. NOx emissions are taken to be 90 percent NO. CO
may also be entered, although it has little or no effect on the
concentration of photochemical pollutants.
A-5
-------
All point sources are included in overall emissions for each
grid square.
References:
1. A. Q. Eschenroeder, J. R. Martinez, and R. A. Nordsieck (1972)
Evaluation of a Diffusion Model for Photochemical Smog
Simulation, Final Report. General Research Corporation, Santa
Barbara, California, EPA No. R4-73-012a.
2. J. R. Martinez, (1972) User's Guide to Diffusion/Kinetics
(DIFKIN) Code. General Research Corporation, Santa Barbara,
California, EPA NO. R4-73-012b.
3. Lurman, F., D. Godden, A.C. Lloyd, R.A. Nordsieck (Oct. 1978).
The Adaptation of a Lagrangian Photochemical Air Quality
Simulation Model to the St. Louis Region and the Regional Air
Pollution Study (RAPS) Data Base, Vols. I & II, Draft Report-
EPA Contract No. 68-02-2765, Research Triangle Park, N.C.
A-6
-------
Model Designation: MADCAP
Developed by: Form and Substance, Inc.
875 Westlake Blvd., Suite 212
Westlake Village, California 91361
(213) 889-0514
Model Description
MADCAP has been called a "hybrid" model, in that it simulates
chemical reactions by Lagrangian principles while retaining an
Eulerian framework for air flow. It allows for several layers of
vertically stacked cells between ground level and mixing height.
Computer requirements are rather extensive.
Chemistry
MADCAP contains a 58-reaction mechanism based on three VOC
classes. These are designated HC1 (olefins and relatively reactive
aromatics), HC2 (alkanes, naphthenes, less reactive aromatics and
some oxygenates), and HC3 (aldehydes and certain ketones).
Emission Source Categories
Tabulated hourly emissions are entered for each cell of the
Eulerian framework, including the elevated cells, which receive
emissions from point sources with large enough effective stack
heights. The emissions are tabulated in terms of the three VOC
classes, NO, NO?, and (optionally) CO.
Elevated Point Source Requirements
Effective stack height of point sources necessary to input
hourly gridded emissions into elevated grid cells.
A-7
-------
References:
1. R. C. Sklarew, J. C. Wilson, and A. J. Fabrick (1977)
Verification of the MADCAP Model of Photochemical Air Pollution
in the San Diego Air Basin. Form and Substance, Westlake
Village, California, Technical Report 10-77-01.
2. R. J. Gelinas, and P. D. Skewes-Cox (1977) Tropospheric
Photochemical Mechanisms. J. Phys. Chem., 81, 2468-2479.
A-8
-------
Model Designation: LIRAQ
Developed by: Lawrence Livermore Laboratory
University of California
P. 0. Box 808
Livermore, California 94550
(415) 422-1826
Model Description
LIRAQ is an Eulerian framework model having one layer of well-
mixed vertical cells between ground level and the mixing height,
with each cell corresponding to one emission grid square. Loss of
pollutants at the ground can be accommodated. Computer requirements
are very extensive.
Chemistry
LIRAQ contains a 56-reaction chemical mechanism based orr three
VOC classes. These are designated HC1 (alkenes and relatively reac-
tive aromatics); HC2 (alkanes, less reactive aromatics and some oxy-
genates); and HC4 (aldehydes, some aromatics, and ketones). Methane
is excluded.
Emission Source Categories
Tabulated hourly emissions are entered for categories known as
(1) airport, (2) major-industrial, (3) population-dependent, and (4)
vehicular. Pollutants entered are particulates, HC1, HC2, HC4, NO,
S02» and CO. All NOx is assumed to be NO.
A-9
-------
Elevated Point Source Requirements
Although the model has no vertical resolution, a distinction is
made between surface and elevated emissions (higher than 100 ft); in
the model, elevated emissions do not immediately impact their
vertical grid cells.
References:
1. M. C. MacCracken, and G. D. Sauter (1975) Development of an Air
Pollution Model for the San Francisco Bay Area, Final Report to
the National Science Foundation. Lawrence Livermore Laboratory,
Livermore, California, UCRL-51920, Volumes 1 and 2.
2. M. C. MacCracken (1975) User's Guide to the LIRAQ Model: An Air
Pollution Model for the San Francisco Bay Area. Lawrence Liver-
more Laboratory, Livermore, California, UCRL-51983.
3. Livermore Regional Air Quality Model (LIRAQ) Application to St.
Louis. Lawrence Livermore Laboratory, Livermore, California,
UCRL-52432.
4. M.C. MacCracken, et al. (1978) The Livermore Regional Air
Quality Model: I Concept and Development. J. Appl. Meteorol.,
17., 264-272.
5. W.H. Deuwer, M.C. MacCracken, and J.J. Walton (1978) The
Livermore Regional Air Quality Model: II Verification and Sample
Application in the San Francisco Bay Area. J. Appl. Meteorol.,
L7, 273-311.
A-10
-------
Model Designation: Urban Airshed Model
Developed by: System Applications, Inc.
950 Northgate Drive
San Rafael, California 94903
(415) 472-4011
Model Description
The Urban Airshed Model is an Eulerian framework model having
(optionally) several vertically stacked cells overlying each
emission grid cell. It requires the calculation of self-consistent
three-dimensional wind fields and accommodates assumptions regarding
loss of pollutants to the ground. Computer requirements are very
extensive.
Chemistry
The most recent chemical mechanism reported in connection with
the Urban Airshed Model employs 75 reactions, using a so-called
"carbon bond" approach in which each carbon atom is treated accord-
ing to its bond type.
Emission Source Categories
All source categories are distinguished as either elevated or
ground level. Elevated point sources are large VOC or NOX emit-
ters having effective stack heights above some established cutoff
level. Ground level sources include minor (non-elevated) point
sources, highway vehicles, and other area sources. VOC classes
input to the model include single bonded carbon atoms, double carbon
bonds, aromatic rings, carbonyl bonds, and ethylene-like double
carbon bonds. Other pollutants required are NO, N02, CO, S02,
and particulate.
A-ll
-------
Elevated Point Source Requirements
Input requirements include stack parameters for large point
sources as well as emission data, to allow the program to allocate
emissions to elevated grid cells.
References:
1. Ames, J. et al (1978), "The User's Manual for the SAI Airshed
Model," draft report for Environmental Science Research Labora-
tory, Office of Research and Development, U.S. Environmental
Protection Agency, EM78-89, Systems Applications, Inc., San
Rafael, California.
2. G. E. Anderson, S. R. Hayes, M. J. Hillyer, J. P. Killus, and P.
V. Mundkur (1977) Air Quality in the Denver Metropolitan Region
1974-2000. Systems Applications, Inc., San Rafael, California,
EPA No. 908/1-77-002.
3. Jerskey, T.N., and J.H. Seinfeld (1976), "Continued Research in
Mesoscale Air Pollution Simulation Modeling--Volume IV: Exami-
nation of the Feasibility of Modeling Photochemical Aerosol
Dynamics," EPA-600/4-76-016d, Systems Applications, Incorporated,
San Rafael, California.
4. Jerskey, T.N., et al. (1976), "Continued Research in Mesoscale
Air Pollution Simulation Modeling—Vol. VII: Mathematical
Modeling of Urban Aerosol Dynamics," EF76-144, Systems Applica-
tions, Incorporated, San Rafael, California.
5. Killus, J.P., et al. (1977), "Continued Research in Mesoscale
Air Pollution Simulation Modeling—Vol. V: Refinements in
Numerical Analysis, Transport, Chemistry, and Pollutant Removal,"
F77-142, Systems Applications, Incorporated, San Rafael,
Cal iform'a.
6. Lamb, R.G., et al. (1977), "Continued Research in Mesoscale Air
Pollution Simulation Model ing—Vol. VI: Further Studies in the
Modeling of Microscale Phenomena," EF77-143, Systems Applica-
tions, Incorporated, San Rafael, California.
7. Lamb, R.G. (1976), "Continued Research in Mesoscale Air Pollu-
tion Simulation Model ing--Vol ume III: Modeling of Microscale
Phenomena," EPA-600/4-76016c, Systems Applications, Incorpo-
rated, San Rafael, California.
A-12
-------
8. Lamb, R.G., W.H. Chen, and J.H. Seinfeld (1975), "Numerico-
Empirical Analysis of Atmospheric Diffusion Theories," J. Atmos.
Sci., Vol. 32, pp. 1794-1807.
9. Liu, M.K., et al. (1976a), "Continued Research in Mesoscale Air
Pollution Simulation Modeling--Volume I: Analysis of Model
Validity and Sensitivity and Assessment of Prior Evaluation
Studies," EPA-600/4-76-016a, Systems Applications, Incorpo-
rated, San Rafael, California.
10. Reynolds, S.D., et al. (1979), "Photochemical Modeling of Trans-
portation Control Strategies. Volume I, Model Development,
Performance Evaluation, and Strategy Assessment," draft final
report for Office of Research FHWA, U.S. Department of Transpor-
tation, EF79-37, Systems Applications, Incorporated, San Rafael,
California.
11. Reynolds, S.D., et al. (1978), "Application of the SAI Airshed
Model to the Evaluation of Alternative Population Growth Fore-
casts for the South Coast Air Basin," EF78-124, Systems Applica-
tions, Incorporated, San Rafael, California.
12. Reynolds, S.D., et al. (1976), "Continued Research in Mesoscale
Air Pollution Simulation Modeling: Volume II: Refinements in the
Treatments of Chemistry, Meteorology, and Numerical Integration
Procedures," EPA-600/4-76-016b, Systems Applications, Incorpo-
rated, San Rafael, California.
13. Whitten, G.Z., et al. (1979). Modeling of Simulated Photochemi-
cal Smog with Kinetic Mechanisms. Volume 1, Interim Report;
Volume 2, Interim Report Appendix. Systems Applications, Inc.,
San Rafael, California, EPA No. 600/3-79-001 a and b.
A-13
-------
Model Designation: Caltech Model
Developed by: California Institute of Technology
1201 E. California Boulevard
Pasadena, California 91125
(213) 795-6811
Model Description
The Caltech Model utilizes an Eulerian framework, having
(optionally) several vertically stacked cells overlying each
emissions grid cell. It requires the calculation of self-consistent
three-dimensional wind fields and accommodates assumptions regarding
the loss of pollutants to the ground. Computer requirements are
very extensive.
Chemistry
The most recent chemical mechanism is said to involve 58 reac-
tions and six main classes of VOC's designated as paraffins, aroma-
tics, ethylene, other olefins, formaldehyde, and other aldehydes.
Emission Source Categories
Tabulated hourly emissions are entered for as many categories
as needed. For major point sources, stack height and heat flux
parameters must be provided.
Reference:
1. G. J. McRae and J. H. Seinfeld (1979). Final Report to the
California Air Resources Board, in preparation. (California
Institute of Technology).
A-14
-------
APPENDIX B
ACRONYMS AND ABBREVIATIONS
GLOSSARY OF TERMS
-------
ACRONYMS AND ABBREVIATIONS
ABAG
ADT
AP-42
APRAC2
AQCR
CBD
CTG
DOT
EIS/P&R
EKMA
EPA
GVW
HC
HDD
HDG
HWYEMIS1
I/M
LOT
LDT1
LDT2
LDV
LOADVN
MC
MOBILE1
MPO
NEDS
NMHC
NOX
OBERS
0-D
Association of Bay Area Governments
Average Daily Traffic
"Compilation of Air Pollutant Emission Factors,"
Publication No. AP-42, U.S. Enviromental
Protection Agency, Research Triangle Park, NC
A link-based traffic network emissions model, more
data-intensive than HWYEMIS1
Air Quality Control Region
Central Business District
Control Techniques Guideline
The U.S. Department of Transportation
Emission Inventory System/Permits and Regulation
Subsystem
Empirical Kinetic Modeling Approach
The U.S. Environmental Protection Agency
Gross Vehicle Weight
Hydrocarbons
Heavy Duty Diesel Vehicle
Heavy Duty Vehicle, Gasoline Fueled
A link-based traffic network emissions model
Inspection and Maintenance
Light Duty Trucks
Light duty, gasoline-powered trucks, having gross
vehicle weight 6,000 pounds or less
Light duty, gasoline-powered trucks, having gross
vehicle weight more than 6,000 pounds but not
more than 8,500 pounds
Light Duty Vehicle
A computer program in the PLANPAC/BACKPAC urban
transportation program planning battery of
computer program that is used in the trip
assignment process
Motorcycle
A computer model, approved by EPA, for estimating
motor vehicle emission factors; preferred over the
earlier "modal model"
Metropolitan Planning Organization
National Emissions Data System
Nonmethane Hydrocarbons
Oxides of Nitrogen
Office of Business Economics, Research Service
Origin-Destination
Ozone
B-l
-------
ACRONYMS AND ABBREVIATIONS
PLANPAC/BACKPAC - An urban transportation planning battery of
computer programs available through the Federal
Highway Administration. This battery lacks some
of the multimodal planning capabilities included
in the more recent Urban Transportation Planning
Systems.
RACT - Reasonably Available Control Technology
RVOC - Reactive Volatile Organic Compounds
SAPOLLUT - A link-based traffic network emissions model, now
largely superseded by HWYEMIS1 and APRAC2
SMSA - Standard Metropolitan Statistical Area
SOX - Oxides of Sulfur
SPEED - A computer program in the PLANPAC/BACKPAC urban
transportation planning battery of computer
programs used to determine vehicle speed based on
the volume/capacity relationship and other land
and other land use and facility characteristics.
TRPTAB - A program in the PLANPAC/BACKPAC system, used to
generate the number of zone-to-zone traffic
interchanges
TSP - Total Suspended Particulates
UROAD - A computer program in urban transportation
planning systems used in the trip assignment
process.
USGS - United States Geological Survey
UTM - Universal Transverse Mercator
UTPS - Urban Transportation Planning System
VMT - Vehicle Miles Traveled
VOC - Volatile Organic Compound
B-2
-------
GLOSSARY OF TERMS
Activity level - Any variable parameter associated with the opera-
tion of a source of emissions which is proportional to the
quantity of pollutant emitted.
Aldehyde - A carbon compound containing the aldehyde functional
grouping (-CHO).
Algorithm - Instruction for computing; usually in reference to
computer programs.
Allocation factor -Used for distributing emissions spatially or temporally.
Ambient measurement - Measurement of concentrations of pollutants in
the ambient atmosphere.
Area source - The aggregate of all emission sources in a particular
source category, except those treated in an emission inventory
as point sources (q.v.). All mobile source categories are area
sources.
Area type - A general land use descriptor. A frequent classifica-
tion system is CBD, fringe, suburban, and rural.
Aromatic - A carbon compound containing aromatic nuclei, sometimes
called "benzene rings."
Average daily traffic - The average number of vehicles passing a
specified point during a 24-hour period.
Background concentration - The concentration of a pollutant in the
ambient air of a region as measured by monitors unaffected by
sources within the region (i.e., by "upwind" monitors).
Baseline projection - Estimate of emissions expected in future
years, based on a conservative scenario of growth and emission
controls. Baseline emission controls for a given projection
year include only those controls that have been legally man-
dated at the time of preparing the projection.
Baseline projection inventory - An inventory of emissions expected
in a future year, based on best available projections of popu-
lation and employment and assuming that only currently mandated
emission controls will be in force (i.e., no new control
strategies).
Breathing loss - Loss of vapors from storage tanks due to diurnal
warming and cooling.
B-3
-------
Capacity - The maximum number of vehicles reasonably expected to
pass over a section of highway in a specified time under
specified conditions. Also known as "level of service."
Carbon bond mechanism - A chemical mechanism intended for use in
photochemical simulation models. It differs from other current
mechanisms in the manner of lumping VOC species.
Catalyst - A material which, without being consumed, promotes a
chemical reaction. Afterburners containing catalysts are used
for control of VOC emissions from gasoline-fueled vehicles.
Centroid - An assumed point in a traffic zone that represents the
origin or destination of all trips to or from that zone.
Centroid connector - A synthetic roadway link representing travel on
local or collector streets, connecting a zone centroid to the
actual highway network.
Chemical reaction rate - The rate at which a reactant is consumed or
a product generated by a chemical reaction, usually given as
rate of change of concentration.
Cold start - For catalyst-equipped vehicles, startup of an engine
which has not been operated during the previous hour. For
other vehicles, startup of an engine which has not been
operated during the previous 4 hours.
Composition - The proportion of various species or classes of
species present in a mixture of chemical substances (e.g., the
composition of VOC present in a polluted atmosphere can be
expressed in terms of the fractions of paraffins, olefins,
aromatics, and other chemical groups).
Control strategy projection inventory - An inventory of emissions,
for a future year, which differs from the baseline inventory in
that it takes into account the expected impact of a proposed
control strategy.
Deterioration rate - The estimated linear rate of change in average
emissions from a motor vehicle exhaust system over 10,000 miles
or 1 year of driving.
Diurnal pattern - The variation of emission rate or activity source
with time of day; usually expressed as the fraction or percen-
tage of the daily total occurring in each hour of the day.
Double bond - A type of carbon-atom linkage which exists in olefins,
but not in paraffins.
B-4
-------
Effective stack height - The height of the horizontal plume center-
line above the base of the stack, equal to physical stack
height plus height of plume rise.
Emission factor - For point sources, the factor relating uncontrol-
led emissions to activity levels. For area sources, the factor
relating net emissions to activity levels.
Emission factor model - A computer model for estimating emission
factors, usually in reference to motor vehicle emissions.
Emission inventory - A compilation of information relating to sources
of pollutant emissions, including location, source strength,
control devices, stack parameters, and often additional details.
Emissions, cold start - Motor vehicle emissions occurring within
about 505 seconds after a cold start (q.v.).
Emissions, diurnal - (See Emissions, vehicle related).
Emissions, hot soak - Motor vehicle emissions originating in evapo-
ration of fuel from the carburetor after engine shutdown.
Emissions, hot transient - Motor vehicle emissions occurring within
about 505 seconds after a hot start, (q.v.)
Emissions, travel related - Motor vehicle emissions which occur while
a vehicle is in a hot, stable operating mode. Also referred to
as "exhaust emissions" and "hot stable mode emissions".
Emissions, trip-end related - Motor vehicle emissions related to
engine startup and shutdown. They include "cold startup" and
"hot transient" emissions, occurring on startup, and "hot soak"
emissions, which occur on shutdown.
Emissions, vehicle related - Motor vehicle emissions caused by
diurnal changes in ambient temperature, independent of vehicle
operation. Also called "diurnal" emissions.
Empirical Kinetic Modeling Approach - A source-receptor relationship
developed by EPA for estimating the overall reduction of
volatile organic compound emissions needed in an urban area,
based on existing oxidant levels and VOC/NOx ratios.
Evaporative losses - VOC emissions caused by the vaporization of
materials, such as gasoline and solvents.
Facility type - Roadway classification (e.g., freeway, major
arterial, etc.).
B-5
-------
Forecast - Prediction of future conditions or occurrences. A forecast
differs from a projection in that the latter may be contingent on
various assumptions (for example, in control-strategy scenarios).
Gasoline marketing - The operations and systems associated with the
transportation of gasoline from refineries to bulk terminals,
bulk storage, service stations, and vehicle tanks.
Grid cell - One square of a grid system (q.v.).
Grid model - An air quality simulation model which provides estimates
of pollutant concentrations for a gridded network of receptors,
using assumptions regarding the exchange of air between hypothe-
tical box-like cells in the atmosphere above an emission grid
system. Mathematically, this is known as an "Eulerian" model.
Grid spacing - In a grid system used in photochemical modeling, the
distance between successive grid points, which is equal to the
length of each side of a grid square.
Grid square - Same as "grid cell." See grid system.
Grid system - A network formed by two mutually perpendicular sets of
parallel lines dividing an area into square cells called grid
cells or grid squares.
Hot start - Startup of a catalyst-equipped vehicle which has been
operated within the previous hour.
Hybrid model - A traffic network emissions model in which estimates
of the three major components of automotive emissions are pre-
pared by separate methodologies.
Hydrocarbon - (1) In chemistry, any compound containing only carbon
and hydrogen. Common types of hydrocarbons are paraffins
(alkanes), olefins (alkenes), and aromatics. (2) In air
pollution usage, equivalent to "volatile organic compound"
(VOC) and recently being supplanted by that term, since some
VOC's are not, chemically, hydrocarbons.
Impedance - In traffic assignment, a factor representing resistance
to travel.
Land use planning model - A system of computer programs constituting
tools for forecasting geographical distribution of various uses
of land in a planning area.
Land use projection - Estimate of land use in a future year (often
given in terms of land use maps representing the projected con-
ditions).
B-6
-------
Land use trip generation rate - A rate establishing the ratio between
number of trips and a land use or socio-economic indicator.
Link - In traffic assignment, a section of a highway network defined
by a node at each end.
Lumping - In chemical mechanisms, the stratagem of representing
certain components by surrogate or hypothetical species, in
order to reduce the assumed number of elementary reactions to
a manageable number.
Methane - The simplest hydrocarbon species; often excluded from VOC
measurements or inventories because it is essentially unreactive
in atmospheric photochemical reactions.
Mobile source - Any self-propelled source of air pollutants, such as
motor vehicles, vessels, locomotives and aircraft.
Modeler's tape - Magnetic tape carrying data formatted for direct in-
put to a computer model, such as an air quality simulation model.
Mole - An amount of a chemical species whose weight is the molecular
weight in weight units; e.g., for NO, having molecular weight =
30, a gram mole is 30 grams, a pound mole is 30 pounds, etc.
Mole percent - The number of moles of a given component in 100 moles
of a mixture. In gaseous mixtures, equivalent to volume percent.
National Emission Data System - An automatic data processing system
developed by EPA for storage and retrieval of source and emis-
sion data.
Nitric oxide (or nitrogen monoxide) - One of the two oxides of ni-
trogen which are collectively referred to as NOx (q.v.). The
amount of NO in NOx is often reported in terms of the equiva-
lent weight of N02, in which case its true weight is only
30/46 of the reported weight.
Nitrogen dioxide - One of the two oxides of nitrogen which are
collectively referred to as NOx (q.v.). The total weight of
NOx is often reported "as N02", which is not the true weight
of the mixture but the weight which would be attained if all
the NO were converted to N02-
Node - A numbered point of a transportation model network, desig-
nating the end-point of a link and usually representing a
street intersection.
B-7
-------
Nonmethane - Excluding methane.
Nonmethane hydrocarbon - All hydrocarbons, or all VOC, except
methane.
Office of Business Economics, Research Service - (Used in reference
to projections prepared jointly by the U.S. Department of Com-
merce, Bureau of Economic Affairs, Office of Business Economics
and the U.S. Department of Agriculture, Economic Research
Service, for the U.S. Water Resources Council, April 1974.)
Olefin - A carbon compound characterized by a type of unsaturation
known as a "double bond".
Operating pattern - (See diurnal pattern).
Origin-destination survey - A detail study of existing travel demand
and characteristics made by home and driver interviews.
Oxides of nitrogen - In air pollution usage, this comprises nitric
oxide (NO) and nitrogen dioxide (N0£); usually expressed in
terms of the equivalent amount of N0~2.
Ozone - A toxic substance formed in the photochemical reactions
involving volatile organic compounds and oxides of nitrogen;
the principal chemical component of the photochemical oxidant
formed in photochemical air pollution.
Ozone precursors - Volatile organic compounds and oxides of nitro-
gen, as air pollutant emissions and as air contaminants. Under
the influence of ultraviolet light from the sun, these pre-
cursors react to form photochemical oxidants, including ozone.
Paraffin - A saturated, nonaromatic hydrocarbon compound.
Photochemistry - The chemistry of reactions which are promoted by
light.
Plume rise - Height of the horizontal centerline of a plume above
the stack top or other point of origin of the plume.
Point source - Generally, any stationary emission source for which
individual records are maintained for emission inventory pur-
poses; distinguished from "area source," often by a criterion
involving emission rate, such as 100 tons per year.
Profile - (1) Chemical composition of a VOC mixture, in terms of
individual compounds (VOC species) or classes of compounds (VOC
classes); (2) Time-dependence of the concentration of a com-
pound or class of compounds, depicted graphically.
B-8
-------
Reactive class - See "Species class."
Reactivity - A measure of the rate at which an organic compound will
react, in the presence of sunlight and oxides of nitrogen, to
promote the formation of photochemical oxidants.
Reactivity class - See "Species class."
Receptor - A hypothetical sensor or monitoring instrument, usually a
unit of a hypothetical network overlaid on the map of an area
being modeled. In Eulerian "grid" models, one receptor is
usually assumed at the center of each grid square.
Rollback model - A method of estimating degree of control of
emissions needed to achieve a specified air quality in a
region, based on the assumption that pollutant concentrations
will be proportional to emissions in the region.
Seasonal adjustment - Used with reference to annual average rates of
pollutant emissions, this is the factor needed to calculate
daily or hourly average rates for one season (in the case of
ozone, summer rates are most commonly required).
Secondary traffic - In some transportation models, traffic on
streets not included as links in the model.
Simulation model (air quality) - A model which estimates pollutant
concentrations by a series of calculations meant to simulate
physical and chemical processes in the atmosphere, rather than
by direct application of mathematical or statistical formulas.
Single bond - The type of carbon-atom linkage characteristic of
paraffins.
Solvent - Any liquid organic compound used for dissolving other
substances.
Source/receptor relationship - A model or relationship that predicts
ambient pollutant concentrations based on precursor emission
strengths. Photochemical simulation models are among source/
receptor relationships for ozone.
Spatial resolution - (1) The process of determining or estimating
what emissions may be associated with individual grid cells or
other subcounty areas, given totals for a larger area such as a
county. (2) The degree to which a source can be pinpointed
geographically in an emission inventory.
B-9
-------
Species - With regard to VOC, a species is any particular volatile
organic compound (e.g., methane; 2-hexene; 1,1,1-trichloroe-
thane). With regard to NOX, a species is either nitric oxide
(NO) or nitrogen dioxide
Species class - Any subset of VOC species, combined in accordance
with rules specified by the input instructions for a photo-
chemical simulation model. Also called'Veactive class" or
"reactivity class."
Speed, input - In traffic assignment, the speed included on the link
description.
Speed, output - In traffic assignment, the speed on a link after
capacity restraint.
Split factor - The factor by which total VOC emissions in a given
category must be multiplied to give VOC emissions belonging to
a certain class of compounds, as required for use in a photo-
chemical simulation model. Also, the factor by which NOx emis-
sions must be multiplied to determine NO or N0£ emissions.
Stack parameters - Parameters characteristic of a stack and its
associated plume, as required for input to some models; typi-
cally, stack height, inner diameter, volume flow rate, temper-
ature of gas (needed to calculate plume rise).
Stationary source - Any non-mobile source of air pollutants.
Surrogate indicator - (1) For spatial resolution, a quantity whose
areal distribution is known or has been estimated and may be
assumed to be similar to that of the emissions from some source
category whose areal distribution is unknown. (2) For growth,
a quantity for which official growth projections are available
and whose growth may be assumed to be similar to that of ac-
tivity in some source category for which projections are needed.
Temporal resolution - (1) The process of determining or estimating
what emissions may be associated with various seasons of the
year, days of the week or hours of the day, given annual totals
or averages. (2) A measure of the smallest time interval with
which emissions can be associated in an inventory.
Traffic assignment - In a transportation model, the process of de-
termining the route of travel, allocating zone-to-zone trips by
these routes, and accumulating traffic volumes by link.
Traffic volume - A measure of the total number of vehicle miles
traveled in unit time on a given link of a traffic network.
B-10
-------
Trajectory - The path (over the map of a region) described by a
hypothetical parcel of air moved by winds. The air parcel is
identified as being at a given location at a given time, and
the trajectory connects its hypothetical locations at earlier
and later times of day.
Trajectory model - An air quality simulation model which provides
estimates of pollutant concentrations at selected points and
times on the trajectories of hypothetical air parcels which
move over an emission grid system. Mathematically, this is
known as a "Lagrangian" model.
Transportation planning model - A system of computer programs con-
stituting tools for forecasting needs for and performance of
future transportation systems in an urban area.
Trip distribution - In traffic assignment, the process by which
movement of trips between zones is estimated.
Trip end - Either a trip origin or trip destination; or, if the
Gravity Model has been used in the transportation modeling, a
trip production or trip attraction.
Trip generation model - A model representing the relationship
between trip makers and trip making. The product is trip ends
per traffic zone.
Trip table - A table showing trips between zones. The trips may be
separated by mode, by purpose, by time period, and by vehicle
type.
Universal Transverse Mercator - A coordinate system for identifying
locations of sources (or other features) on a map, based on
distances in kilometers from established reference points on
the earth's surface.
Urban Transportation Planning System - An urban transportation plan-
ning battery of computer programs distributed jointly by the
Urban Mass Transit Administration and the Federal Highway Ad-
ministration. It is the most recent battery of programs from
the Department of Transportation.
Validation - (1) In principle, a demonstration of the validity of a
model, such as an air quality simulation model. (2) In air
pollution usage, more often an evaluation of model performance
by means of a comparison between model estimates and observed
air quality data. In this sense, often called "verification."
B-ll
-------
Vehicle miles traveled - An estimated total of number of miles
traveled by all vehicles, or by all vehicles of a given
category, in a specified region for a specified period of time;
often used as a surrogate indicator for spatial resolution of
motor vehicle emissions.
Vehicle mix - Composition of vehicular traffic in terms of the
fraction of vehicle miles traveled by various classes of
vehicles.
Vertical resolution - In regard to meteorological parameters and
concentrations of pollutants in ambient air, the provision (in
a model) of means for taking into account various values at
different heights above ground.
Volatile organic compounds - Any hydrocarbon or other carbon
compound which can be present in the gas phase in the atmos-
phere, with the exception of carbon monoxide, carbon dioxide,
carbonic acid, carbonates, and metallic carbides.
Volume percent - The number of volumes of a given component in 100
volumes of a mixture. In gaseous mixtures, equivalent to mole
percent.
Weight percent - The number of weight or mass units of a given
component in 100 units of a mixture.
Zone - A subdivision of a study area, constituting the smallest
geographic area for which data are aggregated and basic
analyses made.
B-12
-------
TECHNICAL REPORT DATA
(Please read Instructions on the reverse before completing)
1. REPORT NO.
450/4-79-018
2.
3. RECIPIENT'S ACCESSION NO.
4. TITLE AND SUBTITLE
Procedures for the
For Volatile
Inventory Requirements for Photochemical Air Quality
Models - — - --
Preparation of Emission Inventories
Organic Compounds, Volume IT: Emission
6. PERFORMING ORGANIZATION CODE
7 AUTHOR(S)
8 PERhORMING ORGANIZATION REPORT NO.
5 REPORT DATE
September 1Q7Q
9 PERFORMING ORGANIZATION NAME AND ADDRESS
U.S. Environmental Protection Agency
Office of Air Quality Planning and Standards,
Monitoring and Data Analysis Division (MD-14)
Research Triangle Park, NC 27711
10. PROGRAM ELEMENT NO.
11 CONTRACT/GRANT NO.
12. SPONSORING AGENCY NAME AND ADDRESS
13. TYPE OF REPORT AND PERIOD COVERED
14. SPONSORING AGENCY CODE
15. SUPPLEMENTARY NOTES
Project Officer - Thomas F.
16. ABSTRACT
This is a companion document to Volume I, which describes procedures for
compiling the annual countywide inventory of volatile organic compound (VOC) emissions.
Volume II describes procedures for converting the annual countywide emission inventory
to the detailed inventory needed for photochemical models. The detailed inventory
contains hourly gridded emissions (by species class for VOC and NOX) for a typical
weekday during the oxidant season.
17.
KEY WORDS AND DOCUMENT ANALYSIS
DESCRIPTORS
b.IDENTIFIERS/OPEN ENDED TERMS
c. COS AT I F;ield/Group
Apportioning
Emissions Inventory
Gridding
Hydrocarbons
Nitrogen Oxides
Photochemical Models
Projections
Spatial Resolution
Temporal Resolution
Species Resolution
Species
Volatile Organics
18. DISTRIBUTION STATEMENT
19. SECURITY CLASS (ThisReport)
21. NO. OF PAGES
232
20. SECURITY CLASS (Thispage)
22. PRICE
EPA Form 2220-1 (Rev. 4-77) PREVIOUS EDITION is OBSOLETE
------- |