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

-------