Está en la página 1de 93

Pt. 51, App. W 40 CFR Ch.

I (7–1–97 Edition)

APPENDIX W TO PART 51—GUIDELINE ON


AIR QUALITY MODELS
PREFACE
a. Industry and control agencies have long
expressed a need for consistency in the appli-
cation of air quality models for regulatory
purposes. In the 1977 Clean Air Act, Congress
mandated such consistency and encouraged
the standardization of model applications.
The Guideline on Air Quality Models (here-
after, Guideline) was first published in April
1978 to satisfy these requirements by specify-
ing models and providing guidance for their
use. The Guideline provides a common basis
for estimating the air quality concentrations
used in assessing control strategies and de-
veloping emission limits.

346
Environmental Protection Agency Pt. 51, App. W
b. The continuing development of new air 3.3.1 The Model Clearinghouse
quality models in response to regulatory re- 3.3.2 Regional Meteorologists Workshops
quirements and the expanded requirements 4.0 Simple-Terrain Stationary Source Models
for models to cover even more complex prob- 4.1 Discussion
lems have emphasized the need for periodic 4.2 Recommendations
review and update of guidance on these tech- 4.2.1 Screening Techniques
niques. Four primary on-going activities pro- 4.2.2 Refined Analytical Techniques
vide direct input to revisions of the Guide- 5.0 Model Use in Complex Terrain
line. The first is a series of annual EPA 5.1 Discussion
workshops conducted for the purpose of en- 5.2 Recommendations
suring consistency and providing clarifica- 5.2.1 Screening Techniques
tion in the application of models. The second 5.2.2 Refined Analytical Techniques
activity, directed toward the improvement 6.0 Models for Ozone, Carbon Monoxide and
of modeling procedures, is the cooperative Nitrogen Dioxide
agreement that EPA has with the scientific 6.1 Discussion
community represented by the American 6.2 Recommendations
Meteorological Society. This agreement pro- 6.2.1 Models for Ozone
vides scientific assessment of procedures and 6.2.2 Models for Carbon Monoxide
proposed techniques and sponsors workshops 6.2.3 Models for Nitrogen Dioxide (Annual
on key technical issues. The third activity is Average)
the solicitation and review of new models 7.0 Other Model Requirements
from the technical and user community. In 7.1 Discussion
the March 27, 1980 FEDERAL REGISTER, a pro- 7.2 Recommendations
cedure was outlined for the submittal to 7.2.1 Fugitive Dust/Fugitive Emissions
EPA of privately developed models. After ex- 7.2.2 Particulate Matter
tensive evaluation and scientific review, 7.2.3 Lead
these models, as well as those made avail- 7.2.4 Visibility
able by EPA, are considered for recognition 7.2.5 Good Engineering Practice Stack
in the Guideline. The fourth activity is the Height
extensive on-going research efforts by EPA 7.2.6 Long Range Transport (LRT) (i.e., be-
and others in air quality and meteorological yond 50km)
modeling. 7.2.7 Modeling Guidance for Other Govern-
c. Based primarily on these four activities, mental Programs
this document embodies all revisions to the 7.2.8 Air Pathway Analyses (Air Toxics and
Guideline Although the text has been revised Hazardous Waste)
from the original 1978 guide, the present con- 8.0 General Modeling Considerations
tent and topics are similar. As necessary, 8.1 Discussion
new sections and topics are included. EPA 8.2 Recommendations
does not make changes to the guidance on a 8.2.1 Design Concentrations
predetermined schedule, but rather on an as 8.2.2 Critical Receptor Sites
needed basis. EPA believes that revisions of 8.2.3 Dispersion Coefficients
the Guideline should be timely and respon- 8.2.4 Stability Categories
sive to user needs and should involve public 8.2.5 Plume Rise
participation to the greatest possible extent. 8.2.6 Chemical Transformation
All future changes to the guidance will be 8.2.7 Gravitational Settling and Deposition
proposed and finalized in the FEDERAL REG- 8.2.8 Urban/Rural Classification
ISTER. Information on the current status of 8.2.9 Fumigation
modeling guidance can always be obtained 8.2.10 Stagnation
from EPA’s Regional Offices. 8.2.11 Calibration of Models
9.0 Model Input Data
TABLE OF CONTENTS 9.1 Source Data
9.1.1 Discussion
List of Tables 9.1.2 Recommendations
1.0 Introduction 9.2 Background Concentrations
2.0 Overview of Model Use 9.2.1 Discussion
2.1 Suitability of Models 9.2.2 Recommendations (Isolated Single
2.2 Classes of Models Source)
2.3 Levels of Sophistication of Models 9.2.3 Recommendations (Multi-Source
3.0 Recommended Air Quality Models Areas)
3.1 Preferred Modeling Techniques 9.3 Meteorological Input Data
3.1.1 Discussion 9.3.1 Length of Record of Meteorological
3.1.2 Recommendations Data
3.2 Use of Alternative Models 9.3.2 National Weather Service Data
3.2.1 Discussion 9.3.3 Site-Specific Data
3.2.2 Recommendations 9.3.4 Treatment of Calms
3.3 Availability of Supplementary Model- 10.0 Accuracy and Uncertainty of Models
ing Guidance 10.1 Discussion

347
Pt. 51, App. W 40 CFR Ch. I (7–1–97 Edition)
10.1.1 Overview of Model Uncertainty sufficient as the sole basis for demonstrating
10.1.2 Studies of Model Accuracy the adequacy of emission limits for existing
10.1.3 Use of Uncertainty in Decision-Mak- sources. Also, the impacts of new sources
ing that do not yet exist can only be determined
10.1.4 Evaluation of Models through modeling. Thus, models, while
10.2 Recommendations uniquely filling one program need, have be-
11.0 Regulatory Application of Models come a primary analytical tool in most air
11.1 Discussion quality assessments. Air quality measure-
11.2 Recommendations ments though can be used in a complemen-
11.2.1 Analysis Requirements tary manner to dispersion models, with due
11.2.2 Use of Measured Data in Lieu of regard for the strengths and weaknesses of
Model Estimates both analysis techniques. Measurements are
11.2.3 Emission Limits particularly useful in assessing the accuracy
12.0 References of model estimates. The use of air quality
13.0 Bibliography measurements alone however could be pref-
14.0 Glossary of Terms erable, as detailed in a later section of this
APPENDIX A TO APPENDIX W OF PART 51— document, when models are found to be un-
SUMMARIES OF PREFERRED AIR QUALITY acceptable and monitoring data with suffi-
MODELS cient spatial and temporal coverage are
APPENDIX B TO APPENDIX W OF PART 51— available.
SUMMARIES OF ALTERNATIVE AIR QUALITY
c. It would be advantageous to categorize
MODELS
the various regulatory programs and to
APPENDIX C TO APPENDIX W OF PART 51—EX-
apply a designated model to each proposed
AMPLE AIR QUALITY ANALYSIS CHECKLIST
source needing analysis under a given pro-
gram. However, the diversity of the nation’s
LIST OF TABLES topography and climate, and variations in
Table source configurations and operating charac-
Title teristics dictate against a strict modeling
No.
‘‘cookbook.’’ There is no one model capable
5–1a Neutral/Stable Meteorological Matrix for CTSCREEN. of properly addressing all conceivable situa-
5–1b Unstable/Convective Meteorological Matrix for
CTSCREEN.
tions even within a broad category such as
5–2 Preferred Options for the SHORTZ/LONGZ Com- point sources. Meteorological phenomena as-
puter Codes When Used in a Screening Mode. sociated with threats to air quality stand-
5–3 Preferred Options for the RTDM Computer Code ards are rarely amenable to a single mathe-
When Used in a Screening Mode. matical treatment; thus, case-by-case analy-
9–1 Model Emission Input Data for Point Sources. sis and judgment are frequently required. As
9–2 Point Source Model Input Data (Emissions) for PSD modeling efforts become more complex, it is
NAAQS Compliance Demonstrations.
9–3 Averaging Times for Site-Specific Wind and Turbu-
increasingly important that they be directed
lence Measurements. by highly competent individuals with a
broad range of experience and knowledge in
air quality meteorology. Further, they
1.0 INTRODUCTION
should be coordinated closely with special-
a. The Guideline recommends air quality ists in emissions characteristics, air mon-
modeling techniques that should be applied itoring and data processing. The judgment of
to State Implementation Plan (SIP) 1 revi- experienced meteorologists and analysts is
sions for existing sources and to new source essential.
reviews,2 including prevention of significant d. The model that most accurately esti-
deterioration (PSD).3 It is intended for use mates concentrations in the area of interest
by EPA Regional Offices in judging the ade- is always sought. However, it is clear from
quacy of modeling analyses performed by the needs expressed by the States and EPA
EPA, State and local agencies and by indus- Regional Offices, by many industries and
try. The guidance is appropriate for use by trade associations, and also by the delibera-
other Federal agencies and by State agencies tions of Congress, that consistency in the se-
with air quality and land management re- lection and application of models and data
sponsibilities. The Guideline serves to iden- bases should also be sought, even in case-by-
tify, for all interested parties, those tech- case analyses. Consistency ensures that air
niques and data bases EPA considers accept- quality control agencies and the general pub-
able. The guide is not intended to be a com- lic have a common basis for estimating pol-
pendium of modeling techniques. Rather, it lutant concentrations, assessing control
should serve as a basis by which air quality strategies and specifying emission limits.
managers, supported by sound scientific Such consistency is not, however, promoted
judgment, have a common measure of ac- at the expense of model and data base accu-
ceptable technical analysis. racy. This guide provides a consistent basis
b. Due to limitations in the spatial and for selection of the most accurate models
temporal coverage of air quality measure- and data bases for use in air quality assess-
ments, monitoring data normally are not ments.

348
Environmental Protection Agency Pt. 51, App. W
e. Recommendations are made in this data. Chapter 10 covers the uncertainty in
guide concerning air quality models, data model estimates and how that information
bases, requirements for concentration esti- can be useful to the regulatory decision-
mates, the use of measured data in lieu of maker. The last chapter summarizes how es-
model estimates, and model evaluation pro- timates and measurements of air quality are
cedures. Models are identified for some spe- used in assessing source impact and in evalu-
cific applications. The guidance provided ating control strategies.
here should be followed in all air quality i. This appendix W itself contains three ap-
analyses relative to State Implementation pendices: A, B, and C. Thus, when reference
Plans and in analyses required by EPA, is made to ‘‘Appendix A’’, it refers to appen-
State and local agency air programs. The dix A to this appendix W. Appendices B and
EPA may approve the use of another tech- C are referenced in the same way.
nique that can be demonstrated to be more
j. Appendix A contains summaries of re-
appropriate than those recommended in this
fined air quality models that are ‘‘preferred’’
guide. This is discussed at greater length in
for specific applications; both EPA models
section 3.0. In all cases, the model applied to
and models developed by others are included.
a given situation should be the one that pro-
vides the most accurate representation of at- Appendix B contains summaries of other re-
mospheric transport, dispersion, and chemi- fined models that may be considered with a
cal transformations in the area of interest. case-specific justification. Appendix C con-
However, to ensure consistency, deviations tains a checklist of requirements for an air
from this guide should be carefully docu- quality analysis.
mented and fully supported.
2.0 OVERVIEW OF MODEL USE
f. From time to time situations arise re-
quiring clarification of the intent of the a. Before attempting to implement the
guidance on a specific topic. Periodic work- guidance contained in this appendix, the
shops are held with the EPA Regional Mete- reader should be aware of certain general in-
orologists to ensure consistency in modeling formation concerning air quality models and
guidance and to promote the use of more ac- their use. Such information is provided in
curate air quality models and data bases. this section.
The workshops serve to provide further ex-
planations of Guideline requirements to the 2.1 Suitability of Models
Regional Offices and workshop reports are is-
sued with this clarifying information. In ad- a. The extent to which a specific air qual-
dition, findings from on-going research pro- ity model is suitable for the evaluation of
grams, new model submittals, or results source impact depends upon several factors.
from model evaluations and applications are These include: (1) The meteorological and
continuously evaluated. Based on this infor- topographic complexities of the area; (2) the
mation changes in the guidance may be indi- level of detail and accuracy needed for the
cated. analysis; (3) the technical competence of
g. All changes to the Guideline must follow those undertaking such simulation model-
rulemaking requirements since the Guide- ing; (4) the resources available; and (5) the
line is codified in this appendix W of part 51. detail and accuracy of the data base, i.e.,
EPA will promulgate proposed and final emissions inventory, meteorological data,
rules in the FEDERAL REGISTER to amend this and air quality data. Appropriate data
appendix W. Ample opportunity for public should be available before any attempt is
comment will be provided for each proposed made to apply a model. A model that re-
change and public hearings scheduled if re- quires detailed, precise, input data should
quested. not be used when such data are unavailable.
h. A wide range of topics on modeling and However, assuming the data are adequate,
data bases are discussed in the Guideline. the greater the detail with which a model
Chapter 2 gives an overview of models and considers the spatial and temporal vari-
their appropriate use. Chapter 3 provides spe- ations in emissions and meteorological con-
cific guidance on the use of ‘‘preferred’’ air ditions, the greater the ability to evaluate
quality models and on the selection of alter- the source impact and to distinguish the ef-
native techniques. Chapters 4 through 7 pro- fects of various control strategies.
vide recommendations on modeling tech- b. Air quality models have been applied
niques for application to simple-terrain sta- with the most accuracy or the least degree of
tionary source problems, complex terrain uncertainty to simulations of long term
problems, and mobile source problems. Spe- averages in areas with relatively simple to-
cific modeling requirements for selected reg- pography. Areas subject to major topo-
ulatory issues are also addressed. Chapter 8 graphic influences experience meteorological
discusses issues common to many modeling complexities that are extremely difficult to
analyses, including acceptable model compo- simulate. Although models are available for
nents. Chapter 9 makes recommendations for such circumstances, they are frequently site
data inputs to models including source, me- specific and resource intensive. In the ab-
teorological and background air quality sence of a model capable of simulating such

349
Pt. 51, App. W 40 CFR Ch. I (7–1–97 Edition)
complexities, only a preliminary approxima- Various specific models in these three ge-
tion may be feasible until such time as bet- neric types are discussed in the Guideline.
ter models and data bases become available. c. Physical modeling, the fourth generic
c. Models are highly specialized tools. type, involves the use of wind tunnel or
Competent and experienced personnel are an other fluid modeling facilities. This class of
essential prerequisite to the successful appli- modeling is a complex process requiring a
cation of simulation models. The need for high level of technical expertise, as well as
specialists is critical when the more sophis- access to the necessary facilities. Neverthe-
ticated models are used or the area being in- less, physical modeling may be useful for
vestigated has complicated meteorological complex flow situations, such as building,
or topographic features. A model applied im- terrain or stack downwash conditions, plume
properly, or with inappropriately chosen impact on elevated terrain, diffusion in an
data, can lead to serious misjudgments re- urban environment, or diffusion in complex
garding the source impact or the effective- terrain. It is particularly applicable to such
ness of a control strategy. situations for a source or group of sources in
d. The resource demands generated by use a geographic area limited to a few square
of air quality models vary widely depending kilometers. If physical modeling is available
on the specific application. The resources re- and its applicability demonstrated, it may be
quired depend on the nature of the model and the best technique. A discussion of physical
its complexity, the detail of the data base, modeling is beyond the scope of this guide.
the difficulty of the application, and the The EPA publication ‘‘Guideline for Fluid
amount and level of expertise required. The Modeling of Atmospheric Diffusion,’’4 pro-
costs of manpower and computational facili- vides information on fluid modeling applica-
ties may also be important factors in the se- tions and the limitations of that method.
lection and use of a model for a specific anal- 2.3 Levels of Sophistication of Models
ysis. However, it should be recognized that
under some sets of physical circumstances a. In addition to the various classes of
and accuracy requirements, no present models, there are two levels of sophistica-
model may be appropriate. Thus, consider- tion. The first level consists of general, rel-
ation of these factors should not lead to se- atively simple estimation techniques that
lection of an inappropriate model. provide conservative estimates of the air
quality impact of a specific source, or source
2.2 Classes of Models category. These are screening techniques or
screening models. The purpose of such tech-
a. The air quality modeling procedures dis-
niques is to eliminate the need of further
cussed in this guide can be categorized into more detailed modeling for those sources
four generic classes: Gaussian, numerical, that clearly will not cause or contribute to
statistical or empirical, and physical. Within ambient concentrations in excess of either
these classes, especially Gaussian and nu- the National Ambient Air Quality Standards
merical models, a large number of individual (NAAQS) 5 or the allowable prevention of sig-
‘‘computational algorithms’’ may exist, each nificant deterioration (PSD) concentration
with its own specific applications. While increments.3 If a screening technique indi-
each of the algorithms may have the same cates that the concentration contributed by
generic basis, e.g., Gaussian, it is accepted the source exceeds the PSD increment or the
practice to refer to them individually as increment remaining to just meet the
models. For example, the Industrial Source NAAQS, then the second level of more so-
Complex (ISC) model and the RAM model are phisticated models should be applied.
commonly referred to as individual models. b. The second level consists of those ana-
In fact, they are both variations of a basic lytical techniques that provide more de-
Gaussian model. In many cases the only real tailed treatment of physical and chemical
difference between models within the dif- atmospheric processes, require more detailed
ferent classes is the degree of detail consid- and precise input data, and provide more spe-
ered in the input or output data. cialized concentration estimates. As a result
b. Gaussian models are the most widely they provide a more refined and, at least
used techniques for estimating the impact of theoretically, a more accurate estimate of
nonreactive pollutants. Numerical models source impact and the effectiveness of con-
may be more appropriate than Gaussian trol strategies. These are referred to as re-
models for area source urban applications fined models.
that involve reactive pollutants, but they re- c. The use of screening techniques followed
quire much more extensive input data bases by a more refined analysis is always desir-
and resources and therefore are not as widely able, however there are situations where the
applied. Statistical or empirical techniques screening techniques are practically and
are frequently employed in situations where technically the only viable option for esti-
incomplete scientific understanding of the mating source impact. In such cases, an at-
physical and chemical processes or lack of tempt should be made to acquire or improve
the required data bases make the use of a the necessary data bases and to develop ap-
Gaussian or numerical model impractical. propriate analytical techniques.

350
Environmental Protection Agency Pt. 51, App. W
3.0 RECOMMENDED AIR QUALITY MODELS of use: rural, urban industrial complex, reac-
tive pollutants, mobile sources, complex ter-
a. This section recommends refined model-
rain, visibility, and long range transport.
ing techniques that are preferred for use in
They are undergoing an intensive evaluation
regulatory air quality programs. The status
by category. The evaluation exercises 8 9 10 in-
of models developed by EPA, as well as those
clude statistical measures of model perform-
submitted to EPA for review and possible in-
ance in comparison with measured air qual-
clusion in this guidance, is discussed. The ity data as suggested by the American Mete-
section also addresses the selection of mod- orological Society 11 and, where possible,
els for individual cases and provides rec- peer scientific reviews.12 13 l4
ommendations for situations where the pre- b. When a single model is found to perform
ferred models are not applicable. Two addi- better than others in a given category, it is
tional sources of modeling guidance, the recommended for application in that cat-
Model Clearinghouse 6 and periodic Regional egory as a preferred model and listed in ap-
Meteorologists’ workshops, are also briefly pendix A. If no one model is found to clearly
discussed here. perform better through the evaluation exer-
b. In all regulatory analyses, especially if cise, then the preferred model listed in ap-
other than preferred models are selected for pendix A is selected on the basis of other fac-
use, early discussions among Regional Office tors such as past use, public familiarity, cost
staff, State and local control agencies, in- or resource requirements, and availability.
dustry representatives, and where appro- No further evaluation of a preferred model is
priate, the Federal Land Manager, are in- required if the source follows EPA rec-
valuable and are encouraged. Agreement on ommendations specified for the model in the
the data base to be used, modeling tech- Guideline. The models not specifically rec-
niques to be applied and the overall tech- ommended for use in a particular category
nical approach, prior to the actual analyses, are summarized in appendix B. These models
helps avoid misunderstandings concerning should be compared with measured air qual-
the final results and may reduce the later ity data when they are used for regulatory
need for additional analyses. The use of an applications consistent with recommenda-
air quality checklist, such as presented in tions in section 3.2.
appendix C, and the preparation of a written c. The solicitation of new refined models
protocol help to keep misunderstandings at a which are based on sounder scientific prin-
minimum. ciples and which more reliably estimate pol-
c. It should not be construed that the pre- lutant concentrations is considered by EPA
ferred models identified here are to be per- to be continuous. Models that are submitted
manently used to the exclusion of all others in accordance with the provisions outlined in
or that they are the only models available the FEDERAL REGISTER notice of March 1980
for relating emissions to air quality. The (45 FR 20157) 7 will be evaluated as submitted.
model that most accurately estimates con- These requirements are:
centrations in the area of interest is always i. The model must be computerized and
sought. However, designation of specific functioning in a common Fortran language
models is needed to promote consistency in suitable for use on a variety of computer sys-
model selection and application. tems.
d. The 1980 solicitation of new or different ii. The model must be documented in a
models from the technical community 7 and user’s guide which identifies the mathe-
the program whereby these models are evalu- matics of the model, data requirements and
ated, established a means by which new mod- program operating characteristics at a level
els are identified, reviewed and made avail- of detail comparable to that available for
able in the Guideline. There is a pressing currently recommended models, e.g., the In-
need for the development of models for a dustrial Source Complex (ISC) model.
wide range of regulatory applications. Re- iii. The model must be accompanied by a
fined models that more realistically simu- complete test data set including input pa-
late the physical and chemical process in the rameters and output results. The test data
atmosphere and that more reliably estimate must be included in the user’s guide as well
pollutant concentrations are required. Thus, as provided in computer-readable form.
the solicitation of models is considered to be iv. The model must be useful to typical
continuous. users, e.g., State air pollution control agen-
cies, for specific air quality control prob-
3.1 Preferred Modeling Techniques lems. Such users should be able to operate
the computer program(s) from available doc-
3.1.1 Discussion
umentation.
a. EPA has developed approximately 10 v. The model documentation must include
models suitable for regulatory application. a comparison with air quality data or with
More than 20 additional models were submit- other well-established analytical techniques.
ted by private developers for possible inclu- vi. The developer must be willing to make
sion in the Guideline. These refined models the model available to users at reasonable
have all been organized into eight categories cost or make it available for public access

351
Pt. 51, App. W 40 CFR Ch. I (7–1–97 Edition)
through the National Technical Information techniques recommended in this guide. An
Service; the model cannot be proprietary. alternative to be considered to the perform-
d. The evaluation process will include a de- ance measures contained in Chapter 3 of this
termination of technical merit, in accord- document is set forth in another EPA docu-
ance with the above six items including the ment ‘‘Protocol for Determining the Best
practicality of the model for use in ongoing Performing Model’’. 17 The procedures in both
regulatory programs. Each model will also documents provide a general framework for
be subjected to a performance evaluation for objective decision-making on the accept-
an appropriate data base and to a peer sci- ability of an alternative model for a given
entific review. Models for wide use (not just regulatory application. The documents con-
an isolated case!) found to perform better, tain procedures for conducting both the
based on an evaluation for the same data technical evaluation of the model and the
bases used to evaluate models in appendix A, field test or performance evaluation.
will be proposed for inclusion as preferred b. This section discusses the use of alter-
models in future Guideline revisions. nate modeling techniques and defines three
situations when alternative models may be
3.1.2 Recommendations used.
a. Appendix A identifies refined models
3.2.2 Recommendations
that are preferred for use in regulatory ap-
plications. If a model is required for a par- a. Determination of acceptability of a
ticular application, the user should select a model is a Regional Office responsibility.
model from appendix A. These models may Where the Regional Administrator finds that
be used without a formal demonstration of an alternative model is more appropriate
applicability as long as they are used as indi- than a preferred model, that model may be
cated in each model summary of appendix A. used subject to the recommendations below.
Further recommendations for the applica- This finding will normally result from a de-
tion of these models to specific source prob- termination that (1) A preferred air quality
lems are found in subsequent sections of the model is not appropriate for the particular
Guideline. application; or (2) a more appropriate model
b. If changes are made to a preferred model or analytical procedure is available and is
without affecting the concentration esti- applicable.
mates, the preferred status of the model is b. An alternative model should be evalu-
unchanged. Examples of modifications that ated from both a theoretical and a perform-
do not affect concentrations are those made ance perspective before it is selected for use.
to enable use of a different computer or There are three separate conditions under
those that affect only the format or averag- which such a model will normally be ap-
ing time of the model results. However, when proved for use: (1) If a demonstration can be
any changes are made, the Regional Admin- made that the model produces concentration
istrator should require a test case example estimates equivalent to the estimates ob-
to demonstrate that the concentration esti- tained using a preferred model; (2) if a statis-
mates are not affected. tical performance evaluation has been con-
c. A preferred model should be operated ducted using measured air quality data and
with the options listed in appendix A as the results of that evaluation indicate the
‘‘Recommendations for Regulatory Use.’’ If alternative model performs better for the ap-
other options are exercised, the model is no plication than a comparable model in appen-
longer ‘‘preferred.’’ Any other modification dix A; and (3) if there is no preferred model
to a preferred model that would result in a for the specific application but a refined
change in the concentration estimates like- model is needed to satisfy regulatory re-
wise alters its status as a preferred model. quirements. Any one of these three separate
Use of the model must then be justified on a conditions may warrant use of an alternative
case-by-case basis. model. Some known alternative models that
are applicable for selected situations are
3.2 Use of Alternative Models contained in appendix B. However, inclusion
there does not infer any unique status rel-
3.2.1 Discussion
ative to other alternative models that are
a. Selection of the best techniques for each being or will be developed in the future.
individual air quality analysis is always en- c. Equivalency is established by dem-
couraged, but the selection should be done in onstrating that the maximum or highest,
a consistent manner. A simple listing of second highest concentrations are within 2
models in this guide cannot alone achieve percent of the estimates obtained from the
that consistency nor can it necessarily pro- preferred model. The option to show equiva-
vide the best model for all possible situa- lency is intended as a simple demonstration
tions. An EPA document, ‘‘Interim Proce- of acceptability for an alternative model
dures for Evaluating Air Quality Mod- that is so nearly identical (or contains op-
els’’,15 16 has been prepared to assist in devel- tions that can make it identical) to a pre-
oping a consistent approach when justifying ferred model that it can be treated for prac-
the use of other than the preferred modeling tical purposes as the preferred model. Two

352
Environmental Protection Agency Pt. 51, App. W
percent was selected as the basis for equiva- 3.3.1 The Model Clearinghouse
lency since it is a rough approximation of
the fraction that PSD Class I increments are 3.3.1.1 Discussion
of the NAAQS for SO2, i.e., the difference in a. The Model Clearinghouse is the single
concentrations that is judged to be signifi- EPA focal point for review of air quality
cant. However, notwithstanding this dem- simulation models proposed for use in spe-
onstration, use of models that are not equiv- cific regulatory applications. Details con-
alent may be used when the conditions of cerning the Clearinghouse and its operation
paragraph e of this section are satisfied. are found in the document, ‘‘Model Clearing-
d. The procedures and techniques for deter- house: Operational Plan.’’ 6 Three primary
mining the acceptability of a model for an functions of the Clearinghouse are:
individual case based on superior perform- i. Review of decisions proposed by EPA Re-
ance is contained in the document entitled gional Offices on the use of modeling tech-
‘‘Interim Procedures for Evaluating Air niques and data bases.
Quality Models’’, 15 and should be followed, ii. Periodic visits to Regional Offices to
as appropriate.a Preparation and implemen- gather information pertinent to regulatory
tation of an evaluation protocol which is ac- model usage.
ceptable to both control agencies and regu- iii. Preparation of an annual report sum-
lated industry is an important element in marizing activities of the Clearinghouse in-
such an evaluation. cluding specific determinations made during
e. When no appendix A model is applicable the course of the year.
to the modeling problem, an alternative re-
fined model may be used provided that: 3.3.1.2 Recommendations
i. The model can be demonstrated to be ap- a. The Regional Administrator may re-
plicable to the problem on a theoretical quest assistance from the Model Clearing-
basis; and house after an initial evaluation and deci-
ii. The data bases which are necessary to
sion has been reached concerning the appli-
perform the analysis are available and ade-
cation of a model, analytical technique or
quate; and
data base in a particular regulatory action.
iii. Performance evaluations of the model
The Clearinghouse may also consider and
in similar circumstances have shown that
evaluate the use of modeling techniques sub-
the model is not biased toward underesti-
mitted in support of any regulatory action.
mates; or
Additional responsibilities are: (1) Review
iv. After consultation with the EPA Re-
proposed action for consistency with agency
gional Office, a second model is selected as a
policy; (2) determine technical adequacy; and
baseline or reference point for performance
(3) make recommendations concerning the
and the interim procedures 15 protocol 17 are
technique or data base.
then used to demonstrate that the proposed
model performs better than the reference 3.3.2 Regional Meteorologists Workshops
model.
3.3.2.1 Discussion
3.3 Availability of Supplementary Modeling
Guidance a. EPA conducts an annual in-house work-
shop for the purpose of mutual discussion
a. The Regional Administrator has the au- and problem resolution among Regional Of-
thority to select models that are appropriate fice modeling specialists, EPA research mod-
for use in a given situation. However, there eling experts, EPA Headquarters modeling
is a need for assistance and guidance in the and regulatory staff and representatives
selection process so that fairness and con- from State modeling programs. A summary
sistency in modeling decisions is fostered of the issues resolved at previous workshops
among the various Regional Offices and the was issued in 1981 as ‘‘Regional Workshops
States. To satisfy that need, EPA estab- on Air Quality Modeling: A Summary Re-
lished the Model Clearinghouse and also port.’’ 17 That report clarified procedures not
holds periodic workshops with headquarters, specifically defined in the 1978 version of the
Regional Office and State modeling rep- Guideline and was issued to ensure the con-
resentatives. sistent interpretation of model requirements
from Region to Region. Similar workshops
a Another EPA document, ‘‘Protocol for De- for the purpose of clarifying Guideline proce-
termining the Best Performing Model’’, 17 dures or providing detailed instructions for
contains advanced statistical techniques for the use of those procedures are anticipated
determining which model performs better in the future.
than other competing models. In many cases,
3.3.2.2 Recommendations
this protocol should be considered by users
of the ‘‘Interim Procedures for Evaluating a. The Regional Office should always be
Air Quality Models’’ in preference to the ma- consulted for information and guidance con-
terial currently in Chapter 3 of that docu- cerning modeling methods and interpreta-
ment. tions of modeling guidance, and to ensure

353
Pt. 51, App. W 40 CFR Ch. I (7–1–97 Edition)
that the air quality model user has available are applicable (e.g., lead), then the long term
the latest most up-to-date policy and proce- models should be used. The conversion from
dures. long term to short term concentration aver-
ages by any transformation technique is not
4.0 SIMPLE-TERRAIN STATIONARY SOURCE acceptable in regulatory applications.
MODELS
5.0 MODEL USE IN COMPLEX TERRAIN
4.1 Discussion
5.1 Discussion
a. Simple terrain, as used in this section, is
considered to be an area where terrain fea- a. For the purpose of the Guideline, com-
tures are all lower in elevation than the top plex terrain is defined as terrain exceeding
of the stack of the source(s) in question. The the height of the stack being modeled. Com-
models recommended in this section are gen- plex terrain dispersion models are normally
erally used in the air quality impact analysis applied to stationary sources of pollutants
of stationary sources for most criteria pol- such as SO2 and particulates.
lutants. The averaging time of the con- b. A major outcome from the EPA Complex
centration estimates produced by these mod- Terrain Model Development project has been
els ranges from 1 hour to an annual average. the publication of a refined dispersion model
b. Model evaluation exercises have been (CTDM) suitable for regulatory application
conducted to determine the ‘‘best, most ap- to plume impaction assessments in complex
propriate point source model’’ for use in sim- terrain. 21 Although CTDM as originally pro-
ple terrain.8 12 However, no one model has duced was only applicable to those hours
been found to be clearly superior. Based on characterized as neutral or stable, a com-
past use, public familiarity, and availability, puter code for all stability conditions,
ISC is the recommended model for a wide CTDMPLUS, 19 together with a user’s
range of regulatory applications. Similar de- guide, 22 and on-site meteorological and ter-
terminations were made for the other refined rain data processors,23 24 is now available.
models that are identified in section 4.2. Moreover, CTSCREEN,19 25 a version of
CTDMPLUS that does not require on-site
4.2 Recommendations meteorological data inputs, is also available
as a screening technique.
4.2.1 Screening Techniques
c. The methods discussed in this section
a. Point source screening techniques are an should be considered in two categories: (1)
acceptable approach to air quality analyses. Screening techniques, and (2) the refined dis-
One such approach is contained in the EPA persion model, CTDMPLUS, discussed below
document ‘‘Screening Procedures for Esti- and listed in appendix A.
mating the Air Quality Impact of Stationary d. Continued improvements in ability to
Sources’’. 18 A computerized version of the accurately model plume dispersion in com-
screening technique, SCREEN, is avail- plex terrain situations can be expected, e.g.,
able.19 20 For the current version of SCREEN, from research on lee side effects due to ter-
see 12.0 References. 20 rain obstacles. New approaches to improve
b. All screening procedures should be ad- the ability of models to realistically simu-
justed to the site and problem at hand. Close late atmospheric physics, e.g., hybrid models
attention should be paid to whether the area which incorporate an accurate wind field
should be classified urban or rural in accord- analysis, will ultimately provide more ap-
ance with section 8.2.8. The climatology of propriate tools for analyses. Such hybrid
the area should be studied to help define the modeling techniques are also acceptable for
worst-case meteorological conditions. Agree- regulatory applications after the appropriate
ment should be reached between the model demonstration and evaluation. 15
user and the reviewing authority on the
choice of the screening model for each analy- 5.2 Recommendations
sis, and on the input data as well as the ulti- a. Recommendations in this section apply
mate use of the results. primarily to those situations where the im-
paction of plumes on terrain at elevations
4.2.2 Refined Analytical Techniques
equal to or greater than the plume center-
a. A brief description of preferred models line during stable atmospheric conditions
for refined applications is found in appendix are determined to be the problem. If a viola-
A. Also listed in appendix A are the model tion of any NAAQS or the controlling incre-
input requirements, the standard options ment is indicated by using any of the pre-
that should be selected when running the ferred screening techniques, then a refined
program, and output options. complex terrain model may be used. Phe-
b. When modeling for compliance with nomena such as fumigation, wind direction
short term NAAQS and PSD increments is of shear, lee-side effects, building wake- or ter-
primary concern, a short term model may rain-induced downwash, deposition, chemical
also be used to provide long term concentra- transformation, variable plume trajectories,
tion estimates. However, when modeling and long range transport are not addressed
sources for which long term standards alone by the recommendations in this section.

354
Environmental Protection Agency Pt. 51, App. W
b. Where site-specific data are used for ei- priate, any of these screening techniques
ther screening or refined complex terrain may be used consistent with the needs, re-
models, a data base of at least 1 full-year of sources, and available data of the user.
meteorological data is preferred. If more b. The Valley Model, COMPLEX I,
data are available, they should be used. Me- SHORTZ/LONGZ, and RTDM should be used
teorological data used in the analysis should only to estimate concentrations at receptors
be reviewed for both spatial and temporal whose elevations are greater than or equal to
representativeness. plume height. For receptors at or below
c. Placement of receptors requires very stack height, a simple terrain model should
careful attention when modeling in complex be used (see Chapter 4). Receptors between
terrain. Often the highest concentrations are stack height and plume height present a
predicted to occur under very stable condi- unique problem since none of the above mod-
tions, when the plume is near, or impinges els were designed to handle receptors in this
on, the terrain. The plume under such condi- narrow regime, the definition of which will
tions may be quite narrow in the vertical, so vary hourly as meteorological conditions
that even relatively small changes in a re- vary. CTSCREEN may be used to estimate
ceptor’s location may substantially affect concentrations under all stability conditions
the predicted concentration. Receptors with- at all receptors located ‘‘on terrain’’ above
in about a kilometer of the source may be stack top, but has limited applicability in
even more sensitive to location. Thus, a multi-source situations. As a result, the esti-
dense array of receptors may be required in mation of concentrations at receptors be-
some cases. In order to avoid excessively tween stack height and plume height should
large computer runs due to such a large be considered on a case-by-case basis after
array of receptors, it is often desirable to consultation with the EPA Regional Office;
model the area twice. The first model run the most appropriate technique may be a
would use a moderate number of receptors function of the actual source(s) and terrain
carefully located over the area of interest. configuration unique to that application.
The second model run would use a more One technique that will generally be accept-
dense array of receptors in areas showing po- able, but is not necessarily preferred for any
tential for high concentrations, as indicated specific application, involves applying both a
by the results of the first model run. complex terrain model (except for the Valley
d. When CTSCREEN or CTDMPLUS is
Model) and a simple terrain model. The Val-
used, digitized contour data must be first
ley Model should not be used for any inter-
processed by the CTDM Terrain Processor 23
mediate terrain receptor. For each receptor
to provide hill shape parameters in a format
between stack height and plume height, an
suitable for direct input to CTDMPLUS.
hour-by-hour comparison of the concentra-
Then the user supplies receptors either
tion estimates from both models is made.
through an interactive program that is part
The higher of the two modeled concentra-
of the model or directly, by using a text edi-
tions should be chosen to represent the im-
tor; using both methods to select receptors
will generally be necessary to assure that pact at that receptor for that hour, and then
the maximum concentrations are estimated used to compute the concentration for the
by either model. In cases where a terrain fea- appropriate averaging time(s). For the sim-
ture may ‘‘appear to the plume’’ as smaller, ple terrain models, terrain may have to be
multiple hills, it may be necessary to model ‘‘chopped off’’ at stack height, since these
the terrain both as a single feature and as models are frequently limited to receptors
multiple hills to determine design con- no greater than stack height.
centrations.
5.2.1.1 Valley Screening Technique
e. The user is encouraged to confer with
the Regional Office if any unresolvable prob- a. The Valley Screening Technique may be
lems are encountered with any screening or used to determine 24-hour averages. This
refined analytical procedures, e.g., meteoro- technique uses the Valley Model with the
logical data, receptor siting, or terrain con- following worst-case assumptions for rural
tour processing issues. areas: (1) P–G stability ‘‘F’’; (2) wind speed of
2.5 m/s; and (3) 6 hours of occurrence. For
5.2.1 Screening Techniques urban areas the stability should be changed
a. Five preferred screening techniques are to ‘‘P–G stability E.’’
currently available to aid in the evaluation b. When using the Valley Screening Tech-
of concentrations due to plume impaction nique to obtain 24-hour average concentra-
during stable conditions: (1) for 24-hour im- tions the following apply: (1) multiple
pacts, the Valley Screening Technique 19 as sources should be treated individually and
outlined in the Valley Model User’s Guide; 26 the concentrations for each wind direction
(2) CTSCREEN,19 as outlined in the summed; (2) only one wind direction should
CTSCREEN User’s Guide; 25 (3) COMPLEX be used (see User’s Guide,26 page 2–15) even if
I; 19 (4) SHORTZ/LONGZ; 19 27 and (5) Rough individual runs are made for each source; (3)
Terrain Dispersion Model (RTDM) 19 90 in its for buoyant sources, the BID option may be
prescribed mode described below. As appro- used, and the option to use the 2.6 stable

355
Pt. 51, App. W 40 CFR Ch. I (7–1–97 Edition)
plume rise factor should be selected; (4) if gional Office should be consulted, and con-
plume impaction is likely on any elevated currence obtained, on the protocol for model-
terrain closer to the source than the dis- ing multiple sources with CTSCREEN to en-
tance from the source to the final plume sure that the worst case is identified and as-
rise, then the transitional (or gradual) plume sessed. The maximum concentration output
rise option for stable conditions should be se- from CTSCREEN represents a worst-case 1-
lected. hour concentration. Time-scaling factors of
c. The standard polar receptor grid found 0.7 for 3-hour, 0.15 for 24-hour and 0.03 for an-
in the Valley Model User’s Guide may not be nual concentration averages are applied in-
sufficiently dense for all analyses if only one ternally by CTSCREEN to the highest 1-hour
geographical scale factor is used. The user
concentration calculated by the model.
should choose an additional set of receptors
at appropriate downwind distances whose 5.2.1.3 COMPLEX I
elevations are equal to plume height minus
10 meters. Alternatively, the user may exer- a. If the area is rural, COMPLEX I may be
cise the ‘‘Valley equivalent’’ option in COM- used to estimate concentrations for all aver-
PLEX I or SCREEN and note the comments aging times. COMPLEX I is a modification of
above on the placement of receptors in com- the MPTER model that incorporates the
plex terrain models. plume impaction algorithm of the Valley
d. When using the ‘‘Valley equivalent’’ op- Model. 19 It is a multiple-source screening
tion in COMPLEX I, set the wind profile ex- technique that accepts hourly meteorologi-
ponents (PL) to 0.0, respectively, for all six cal data as input. The output is the same as
stability classes. the normal MPTER output. When using
COMPLEX I the following options should be
5.2.1.2 CTSCREEN
selected: (1) Set terrain adjustment IOPT
a. CTSCREEN may be used to obtain con- (1)=1; (2) set buoyancy induced dispersion
servative, yet realistic, worst-case estimates IOPT (4)=1; (3) set IOPT (25)=1; (4) set the
for receptors located on terrain above stack terrain adjustment values to 0.5, 0.5, 0.5 0.5,
height. CTSCREEN accounts for the three- 0.0, 0.0, (respectively for six stability class-
dimensional nature of plume and terrain es); and (5) set Z MIN=10.
interaction and requires detailed terrain b. When using the ‘‘Valley equivalent’’ op-
data representative of the modeling domain. tion (only) in COMPLEX I, set the wind pro-
The model description and user’s instruc- file exponents (PL) to 0.0, respectively, for
tions are contained in the user’s guide. 25 The all six stability classes. For all other regu-
terrain data must be digitized in the same
latory uses of COMPLEX I, set the wind pro-
manner as for CTDMPLUS and a terrain
file exponents to the values used in the sim-
processor is available. 23 A discussion of the
ple terrain models, i.e., 0.07, 0.07, 0.10, 0.15,
model’s performance characteristics is pro-
0.35, and 0.55, respectively, for rural model-
vided in a technical paper. 91 CTSCREEN is
designed to execute a fixed matrix of mete- ing.
orological values for wind speed (u), standard c. Gradual plume rise should be used to es-
deviation of horizontal and vertical wind timate concentrations at nearby elevated re-
speeds (σv, σG5w), vertical potential tem- ceptors, if plume impaction is likely on any
perature gradient (dθ/dz), friction velocity elevated terrain closer to the source than
(ux), Monin-Obukhov length (L), mixing the distance from the source to the final
height (zi) as a function of terrain height, plume rise (see section 8.2.5).
and wind directions for both neutral/stable
conditions and unstable convective condi- 5.2.1.4 SHORTZ/LONGZ
tions. Table 5–1 contains the matrix of mete- a. If the source is located in an urbanized
orological variables that is used for each (Section 8.2.8) complex terrain valley, then
CTSCREEN analysis. There are 96 combina- the suggested screening technique is
tions, including exceptions, for each wind di-
SHORTZ for short-term averages or LONGZ
rection for the neutral/stable case, and 108
for long-term averages. SHORTZ and LONGZ
combinations for the unstable case. The
may be used as screening techniques in these
specification of wind direction, however, is
complex terrain applications without dem-
handled internally, based on the source and
terrain geometry. The matrix was developed onstration and evaluation. Application of
from examination of the range of meteoro- these models in other than urbanized valley
logical variables associated with maximum situations will require the same evaluation
monitored concentrations from the data and demonstration procedures as are re-
bases used to evaluate the performance of quired for all appendix B models.
CTDMPLUS. Although CTSCREEN is de- b. Both SHORTZ and LONGZ have a num-
signed to address a single source scenario, ber of options. When using these models as
there are a number of options that can be se- screening techniques for urbanized valley ap-
lected on a case-by-case basis to address plications, the options listed in table 5–2
multi-source situations. However, the Re- should be selected.

356
Environmental Protection Agency Pt. 51, App. W
5.2.1.5 RTDM (Screening Mode) 5.2.2 Refined Analytical Techniques
a. RTDM with the options specified in a. When the results of the screening analy-
table 5–3 may be used as a screening tech- sis demonstrate a possible violation of
nique in rural complex terrain situations NAAQS or the controlling PSD increments, a
without demonstration and evaluation. more refined analysis may need to be con-
b. The RTDM screening technique can pro- ducted.
vide a more refined concentration estimate b. The Complex Terrain Dispersion Model
if on-site wind speed and direction char- Plus Algorithms for Unstable Situations
acteristic of plume dilution and transport (CTDMPLUS) is a refined air quality model
are used as input to the model. In complex that is preferred for use in all stability con-
terrain, these winds can seldom be estimated ditions for complex terrain applications.
accurately from the standard surface (10m CTDMPLUS is a sequential model that re-
level) measurements. Therefore, in order to quires five input files: (1) General program
increase confidence in model estimates, EPA specifications; (2) a terrain data file; (3) a re-
recommends that wind data input to RTDM ceptor file; (4) a surface meteorological data
should be based on fixed measurements at file; and (5) a user created meteorological
stack top height. For stacks greater than profile data file. Two optional input files
100m, the measurement height may be lim- consist of hourly emissions parameters and a
ited to 100m in height relative to stack base. file containing upper air data from rawin-
However, for very tall stacks, see guidance sonde data files, e.g., a National Climatic
in section 9.3.3.2. This recommendation is Data Center TD–6201 file, unless there are no
broadened to include wind data representa- hours categorized as unstable in the record.
tive of plume transport height where such The model description and user instructions
data are derived from measurements taken are contained in Volume 1 of the User’s
with remote sensing devices such as SODAR. Guide. 22 Separate publications 23 24 describe
The data from both fixed and remote meas- the terrain preprocessor system and the me-
urements should meet quality assurance and teorological preprocessor program. In Part I
recovery rate requirements. The user should of a technical article 92 is a discussion of the
also be aware that RTDM in the screening model and its preprocessors; the model’s per-
mode accepts the input of measured wind formance characteristics are discussed in
speeds at only one height. The default values Part II of the same article.93 The size of the
for the wind speed profile exponents shown CTDMPLUS executable file on a personal
in table 5–3 are used in the model to deter- computer is approximately 360K bytes. The
mine the wind speed at other heights. RTDM model produces hourly average concentra-
uses wind speed at stack top to calculate the tions of stable pollutants, i.e., chemical
plume rise and the critical dividing stream- transformation or decay of species and set-
line height, and the wind speed at plume tling/deposition are not simulated. To obtain
transport level to calculate dilution. RTDM concentration averages corresponding to the
treats wind direction as constant with NAAQS, e.g., 3- or 24-hour, or annual aver-
height. ages, the user must execute a postprocessor
c. RTDM makes use of the ‘‘critical divid- program such as CHAVG. 19 CTDMPLUS is
ing streamline’’ concept and thus treats applicable to all receptors on terrain ele-
plume interactions with terrain quite dif- vations above stack top. However, the model
ferently from other models such as SHORTZ contains no algorithms for simulating build-
and COMPLEX I. The plume height relative ing downwash or the mixing or recirculation
to the critical dividing streamline deter- found in cavity zones in the lee of a hill. The
mines whether the plume impacts the ter- path taken by a plume through an array of
rain, or is lifted up and over the terrain. The hills cannot be simulated. CTDMPLUS does
receptor spacing to identify maximum im- not explicitly simulate calm meteorological
pact concentrations is quite critical depend- periods, and for those situations the user
ing on the location of the plume in the verti- should follow the guidance in section 9.3.4.
cal. Analysis of the expected plume height The user should follow the recommendations
relative to the height of the critical dividing in the User’s Guide under General Program
streamline should be performed for differing Specifications for: (1) Selecting mixed layer
meteorological conditions in order to help heights, (2) setting minimum scalar wind
develop an appropriate array of receptors. speed to 1 m/s, and (3) scaling wind direction
Then it is advisable to model the area twice with height. Close coordination with the Re-
according to the suggestions in section 5.2. gional Office is essential to insure a consist-
ent, technically sound application of this
5.2.1.6 Restrictions
model.
a. For screening analyses using the Valley c. The performance of CTDMPLUS is
Screening Technique, COMPLEX I or RTDM, greatly improved by the use of meteorologi-
a sector greater than 221⁄2° should not be al- cal data from several levels up to plume
lowed. Full ground reflection should always height. However, due to the vast range of
be used in the Valley Screening Technique source-plume-hill geometries possible in
and COMPLEX I. complex terrain, detailed requirements for

357
Pt. 51, App. W 40 CFR Ch. I (7–1–97 Edition)
meteorological monitoring in support of re- These measurements should be obtained up
fined analyses using CTDMPLUS should be to the representative plume height(s) of in-
determined on a case-by-case basis. The fol- terest (i.e., the plume height(s) under those
lowing general guidance should be consid- conditions important to the determination
ered in the development of a meteorological of the design concentration). The representa-
monitoring protocol for regulatory applica- tive plume height(s) of interest should be de-
tions of CTDMPLUS and reviewed in detail termined using an appropriate complex ter-
by the Regional Office before initiating any rain screening procedure (e.g., CTSCREEN)
monitoring. As appropriate, the On-Site Me- and should be documented in the monitoring/
teorological Program Guidance document 66 modeling protocol. The necessary meteoro-
should be consulted for specific guidance on logical measurements should be obtained
siting requirements for meteorological tow- from an appropriately sited meteorological
ers, selection and exposure of sensors, etc. As tower augmented by SODAR if the represent-
more experience is gained with the model in ative plume height(s) of interest exceed
a variety of circumstances, more specific 100m. The meteorological tower need not ex-
guidance may be developed. ceed the lesser of the representative plume
d. Site specific meteorological data are height of interest (the highest plume height
critical to dispersion modeling in complex if there is more than one plume height of in-
terrain and, consequently, the meteorologi- terest) or 100m.
cal requirements are more demanding than
g. Locating towers on nearby terrain to ob-
for simple terrain. Generally, three different
meteorological files (referred to as surface, tain stack height or plume height measure-
profile, and rawin files) are needed to run ments for use in profiles by CTDMPLUS
CTDMPLUS in a regulatory mode. should be avoided unless it can clearly be
e. The surface file is created by the mete- demonstrated that such measurements
orological preprocessor (METPRO) 24 based would be representative of conditions affect-
on on-site measurements or estimates of ing the plume.
solar and/or net radiation, cloud cover and h. The rawin file is created by a second me-
ceiling, and the mixed layer height. These teorological preprocessor (READ62) 24 based
data are used in METPRO to calculate the on NWS (National Weather Service) upper
various surface layer scaling parameters air data. The rawin file is used in
(roughness length, friction velocity, and CTDMPLUS to calculate vertical potential
Monin-Obukhov length) which are needed to temperature gradients for use in estimating
run the model. All of the user inputs re- plume penetration in unstable conditions.
quired for the surface file are based either on The representativeness of the off-site NWS
surface observations or on measurements at upper air data should be evaluated on a case-
or below 10m. by-case basis.
f. The profile data file is prepared by the i. In the absence of an appropriate refined
user with on-site measurements (from at model, screening results may need to be used
least three levels) of wind speed, wind direc- to determine air quality impact and/or emis-
tion, turbulence, and potential temperature. sion limits.

TABLE 5–1A—NEUTRAL/STABLE METEOROLOGICAL MATRIX FOR CTSCREEN

Variable Specific values

U (m/s) ..................................... 1.0 2.0 3.0 4.0 5.0


σv (m/s) .................................... 0.3 0.75
σw (m/s) .................................... 0.08 0.15 0.30 0.75
DQ/Dz (K/m) ............................. 0.01 0.02 0.035
WD (Wind direction optimized internally for each meteorological combination)
Exceptions:
(1) If U ≤ 2 m/s and σv ≥ 0.3 m/s, then include σw = 0.04 m/s.
(2) If σw = 0.75 m/s and U ≥ 3.0 m/s, then DU/Dz is limited to ≤ 0.01 K/m.
(3) If U ≥ 4 m/s, then σw ≥ 0.15 m/s.
(4) σw ≤ σv

TABLE 5–1B—UNSTABLE/CONVECTIVE METEOROLOGICAL MATRIX FOR CTSCREEN

Variable Specific values

U (m/s) ................................... 1.0 2.0 3.0 4.0 5.0


ux (m/s) ................................... 0.1 0.3 0.5

358
Environmental Protection Agency Pt. 51, App. W

TABLE 5–1B—UNSTABLE/CONVECTIVE METEOROLOGICAL MATRIX FOR CTSCREEN—Continued


L (m) ....................................... ¥10 ¥50 ¥90
DU/ z (K/m) 0.030 (potential temperature gradient above zi)
D
zi (m) ...................................... 0.5h 1.0h 1.5h
(where h = terrain height)

TABLE 5–2—PREFERRED OPTIONS FOR THE SHORTZ/LONGZ COMPUTER CODES WHEN USED IN A
SCREENING MODE

Option Selection

I Switch 9 ........................ ......................................... If using NWS data, set = 0, If using site-specific data,
check with the Regional Office.
I Switch 17 ...................... ......................................... Set = 1 (urban option).
GAMMA 1 ....................... ......................................... Use default values (0.6 entrainment coefficient).
GAMMA 2 ....................... ......................................... Always default to ‘‘stable’’.
XRY ................................ ......................................... Set = 0 (50m rectilinear expansion distance).
NS, VS, FRQ (SHORTZ)
(particle size, etc.) Do not use (applicable only in flat terrain).
NUS, VS, FRQ (LONGZ)
ALPHA ............................ ......................................... Select 0.9.
SIGEPU
(dispersion parameters) Use Cramer curves (default); if site-specific turbulence
data are available, see Regional Office for advice.
SIGAPU
P (wind profile) ............... ......................................... Select default values given in table 2–2 of User’s In-
structions; if site-specific data are available, see Re-
gional Office for advice.

TABLE 5–3—PREFERRED OPTIONS FOR THE RTDM COMPUTER CODE WHEN USED IN A SCREENING
MODE
Parameter Variable Value Remarks

PR001–003 .................. SCALE ................ ...................................................... Scale factors assuming horizontal distance is
in kilometers, vertical distance is in feet,
and wind speed is in meters per second.
PR004 .......................... ZWIND1 .............. Wind measurement height ........... See section 5.2.1.4.
ZWIND2 .............. Not used ....................................... Height of second anemometer.
IDILUT ................ 1 ................................................... Dilution wind speed scaled to plume height.
ZA ....................... 0 (default) ..................................... Anemometer-terrain height above stack
base.
PR005 .......................... EXPON ............... 0.09, 0.11, 0.12, 0.14, 0.2, 0.3 Wind profile exponents.
(default).
PR006 .......................... ICOEF ................. 3 (default) ..................................... Briggs Rural/ASME 139 dispersion param-
eters.
PR009 .......................... IPPP ................... 0 (default) ..................................... Partial plume penetration; not used.
PR010 .......................... IBUOY ................ 1 (default) ..................................... Buoyancy-enhanced dispersion is used.
ALPHA ................ 3.162 (default) .............................. Buoyancy-enhanced dispersion coefficient.
PR011 .......................... IDMX ................... 1 (default) ..................................... Unlimited mixing height for stable conditions.
PR012 .......................... ITRANS .............. 1 (default) ..................................... Transitional plume rise is used.
PR013 .......................... TERCOR ............ 6*0.5 (default) .............................. Plume patch correction factors.
PR014 .......................... RVPTG ............... 0.02, 0.035 (default) ..................... Vertical potential temperature gradient values
for stabilities E and F.
PR015 .......................... ITIPD .................. 1 ................................................... Stack-tip downwash is used.
PR020 .......................... ISHEAR .............. 0 (default) ..................................... Wind shear; not used.
PR022 .......................... IREFL ................. 1 (default) ..................................... Partial surface reflection is used.
PR023 .......................... IHORIZ ............... 2 (default) ..................................... Sector averaging.
SECTOR ............. 6*22.5 (default) ............................ Using 22.5° sectors.

359
Pt. 51, App. W 40 CFR Ch. I (7–1–97 Edition)

TABLE 5–3—PREFERRED OPTIONS FOR THE RTDM COMPUTER CODE WHEN USED IN A SCREENING
MODE—Continued
Parameter Variable Value Remarks

PR016 to 019; 021; and IY, IZ, IRVPTG, 0 ................................................... Hourly values of turbulence, vertical potential
024. IHVPTG; IEPS; temperature gradient, wind speed profile
IEMIS. exponents, and stack emissions are not
used.

6.0 MODELS FOR OZONE, CARBON MONOXIDE entire urban areas. To ensure proper execu-
AND NITROGEN DIOXIDE tion of this numerical model, users must sat-
isfy the extensive input data requirements
6.1 Discussion for the model as listed in appendix A and the
a. Models discussed in this section are ap- users guide. Users are also referred to the
plicable to pollutants often associated with ‘‘Guideline for Regulatory Application of the
mobile sources, e.g., ozone (O3), carbon mon- Urban Airshed Model’’ 29 for additional data
oxide (CO) and nitrogen dioxide (NO2). Where requirements and procedures for operating
stationary sources of CO and NO2 are of con- this model.
cern, the reader is referred to sections 4 and b. The empirical model, City-specific
5 EKMA,19 30–33 has limited applicability for
b. A control agency with jurisdiction over urban ozone analyses. Model users should
areas with significant ozone problems and consult the appropriate Regional Office on a
which has sufficient resources and data to case-by-case basis concerning acceptability
use a photochemical dispersion model is en- of this modeling technique.
couraged to do so. Experience with and eval-
c. Appendix B contains some additional
uations of the Urban Airshed Model show it
to be an acceptable, refined approach, and models that may be applied on a case-by-
better data bases are becoming available case basis for photochemical or reactive pol-
that support the more sophisticated analyt- lutant modeling. Other photochemical mod-
ical procedures. However, empirical models els, including multi-layered trajectory mod-
(e.g., EKMA) fill the gap between more so- els, that are available may be used if shown
phisticated photochemical dispersion models to be appropriate. Most photochemical dis-
and proportional (rollback) modeling tech- persion models require emission data on in-
niques and may be the only applicable proce- dividual hydrocarbon species and may re-
dure if the available data bases are insuffi- quire three dimensional meteorological in-
cient for refined dispersion modeling. formation on an hourly basis. Reasonably so-
c. Models for assessing the impact of car- phisticated computer facilities are also often
bon monoxide emissions are needed for a required. Because the input data are not uni-
number of different purposes, e.g., to evalu- versally available and studies to collect such
ate the effects of point sources, congested data are very resource intensive, there are
intersections and highways, as well as the only limited evaluations of those models.
cumulative effect on ambient CO concentra- d. For those cases which involve estimat-
tions of all sources of CO in an urban ing the impact on ozone concentrations due
area.94 95
to stationary sources of VOC and NOX,
d. Nitrogen oxides are reactive and also an
whether for permitting or other regulatory
important contribution to the photo-
cases, the model user should consult the ap-
chemical ozone problem. They are usually of
most concern in areas of high ozone con- propriate Regional Office on the accept-
centrations. Unless suitable photochemical ability of the modeling technique.
dispersion models are used, assumptions re- e. Proportional (rollback/forward) model-
garding the conversion of NO to NO2 are re- ing is not an acceptable procedure for evalu-
quired when modeling. Site-specific conver- ating ozone control strategies.
sion factors may be developed. If site-specific
conversion factors are not available or pho- 6.2.2 Models for Carbon Monoxide
tochemical models are not used, NO2 model- a. For analyzing CO impacts at roadway
ing should be considered only a screening intersections, users should follow the proce-
procedure. dures in the ‘‘Guideline for Modeling Carbon
6.2 Recommendations Monoxide from Roadway Intersections’’. 34
The recommended model for such analyses is
6.2.1 Models for Ozone CAL3QHC. 35 This model combines CALINE3
a. The Urban Airshed Model (UAM)19 28 is (already in appendix A) with a traffic model
recommended for photochemical or reactive to calculate delays and queues that occur at
pollutant modeling applications involving signalized intersections. In areas where the

360
Environmental Protection Agency Pt. 51, App. W
use of either TEXIN2 or CALINE4 has pre- where the maximum collective impact from
viously been established, its use may con- the new plus existing sources occurs.
tinue. The capability exists for these inter- d. In urban areas, a proportional model
section models to be used in either a screen- may be used as a preliminary assessment to
ing or refined mode. The screening approach evaluate control strategies to meet the
is described in reference 34; a refined ap- NAAQS for multiple minor sources, i.e.
proach may be considered on a case-by-case minor point, area and mobile sources of NOX;
basis. The latest version of the MOBILE (mo- concentrations resulting from major point
bile source emission factor) model should be sources should be estimated separately as
used for emissions input to intersection mod- discussed above, then added to the impact of
els. the minor sources. An acceptable screening
b. For analyses of highways characterized technique for urban complexes is to assume
by uninterrupted traffic flows, CALINE3 is that all NOX is emitted in the form of NO2
recommended, with emissions input from the and to use a model from appendix A for non-
latest version of the MOBILE model. reactive pollutants to estimate NO2 con-
c. The recommended model for urban centrations. A more accurate estimate can
areawide CO analyses is RAM or Urban be obtained by: (1) Calculating the annual
Airshed Model (UAM); see appendix A. Infor- average concentrations of NOX with an urban
mation on SIP development and require- model, and (2) converting these estimates to
NO2 concentrations using an empirically de-
ments for using these models can be found in
rived annual NO2/NOX ratio. A value of 0.75 is
references 34, 96, 97 and 98.
recommended for this ratio. However, a spa-
d. Where point sources of CO are of con- tially averaged annual NO2/NOX ratio may be
cern, they should be treated using the determined from an existing air quality
screening and refined techniques described in monitoring network and used in lieu of the
section 4 or 5 of the Guideline. 0.75 value if it is determined to be represent-
ative of prevailing ratios in the urban area
6.2.3 Models for Nitrogen Dioxide (Annual
by the reviewing agency. To ensure use of
Average)
appropriate locally derived annual NO2/NOX
a. A tiered screening approach is rec- ratios, monitoring data under consideration
ommended to obtain annual average esti- should be limited to those collected at mon-
mates of NO2 from point sources for New itors meeting siting criteria defined in 40
Source Review analysis, including PSD, and CFR part 58, appendix D as representative of
for SIP planning purposes. This multi-tiered ‘‘neighborhood’’, ‘‘urban’’, or ‘‘regional’’
approach is conceptually shown in Figure 6– scales. Furthermore, the highest annual spa-
1 and described in paragraphs b and c of this tially averaged NO2/NOX ratio from the most
section. Figure 6–1 is as follows: recent 3 years of complete data should be
used to foster conservatism in estimated im-
FIGURE 6–1—MULTI-TIERED SCREENING AP- pacts.
PROACH FOR ESTIMATING ANNUAL NO2 CON-
e. To demonstrate compliance with NO2
PSD increments in urban areas, emissions
CENTRATIONS FROM POINT SOURCES
from major and minor sources should be in-
cluded in the modeling analysis. Point and
Tier 1: Assume Total Conversion of NO to NO2
area source emissions should be modeled as
Tier 2: Multiply Annual NOX Estimate by Empirically discussed above. If mobile source emissions
Derived NO2/NOX Ratio. do not contribute to localized areas of high
ambient NO2 concentrations, they should be
b. For Tier 1 (the initial screen), use an ap- modeled as area sources. When modeled as
propriate Gaussian model from appendix A area sources, mobile source emissions should
to estimate the maximum annual average be assumed uniform over the entire highway
concentration and assume a total conversion link and allocated to each area source grid
of NO to NO2. If the concentration exceeds square based on the portion of highway link
the NAAQS and/or PSD increments for NO2, within each grid square. If localized areas of
proceed to the 2nd level screen. high concentrations are likely, then mobile
c. For Tier 2 (2nd level) screening analysis, sources should be modeled as line sources
multiply the Tier 1 estimate(s) by an empiri- with the preferred model ISCLT.
cally derived NO2/NOX value of 0.75 (annual f. More refined techniques to handle spe-
national default).36 An annual NO2/NOX ratio cial circumstances may be considered on a
differing from 0.75 may be used if it can be case-by-case basis and agreement with the
shown that such a ratio is based on data reviewing authority should be obtained.
likely to be representative of the location(s) Such techniques should consider individual
where maximum annual impact from the in- quantities of NO and NO2 emissions, atmos-
dividual source under review occurs. In the pheric transport and dispersion, and atmos-
case where several sources contribute to con- pheric transformation of NO to NO2. Where
sumption of a PSD increment, a locally de- they are available, site-specific data on the
rived annual NO2/NOX ratio should also be conversion of NO to NO2 may be used. Photo-
shown to be representative of the location chemical dispersion models, if used for other

361
Pt. 51, App. W 40 CFR Ch. I (7–1–97 Edition)
pollutants in the area, may also be applied niques. EPA is continuing to evaluate the
to the NOX problem. performance of a number of proprietary and
public domain models for intermittent and
7.0 OTHER MODEL REQUIREMENTS non-stack emission releases. Until EPA com-
pletes its evaluation, it is premature to rec-
7.1 Discussion
ommend specific models for air pathway
a. This section covers those cases where analyses of intermittent and non-stack re-
specific techniques have been developed for leases in the Guideline.
special regulatory programs. Most of the e. Regional scale models are used by EPA
programs have, or will have when fully de- to develop and evaluate national policy and
veloped, separate guidance documents that assist State and local control agencies. Two
cover the program and a discussion of the such models are the Regional Oxidant Model
tools that are needed. The following para- (ROM) 101 102 103 and the Regional Acid Deposi-
graphs reference those guidance documents, tion Model (RADM). 104 Due to the level of re-
when they are available. No attempt has sources required to apply these models, it is
been made to provide a comprehensive dis- not envisioned that regional scale models
cussion of each topic since the reference doc- will be used directly in most model applica-
uments were designed to do that. This sec- tions.
tion will undergo periodic revision as new
programs are added and new techniques are 7.2 Recommendations
developed.
7.2.1 Fugitive Dust/Fugitive Emissions
b. Other Federal agencies have also devel-
oped specific modeling approaches for their a. Fugitive dust usually refers to the dust
own regulatory or other requirements. An put into the atmosphere by the wind blowing
example of this is the three-volume manual over plowed fields, dirt roads or desert or
issued by the U. S. Department of Housing sandy areas with little or no vegetation. Re-
and Urban Development, ‘‘Air Quality Con- entrained dust is that which is put into the
siderations in Residential Planning.’’ 37 Al- air by reason of vehicles driving over dirt
though such regulatory requirements and roads (or dirty roads) and dusty areas. Such
manuals may have come about because of sources can be characterized as line, area or
EPA rules or standards, the implementation volume sources. Emission rates may be based
of such regulations and the use of the model- on site-specific data or values from the gen-
ing techniques is under the jurisdiction of eral literature.
the agency issuing the manual or directive. b. Fugitive emissions are usually defined
c. The need to estimate impacts at dis- as emissions that come from an industrial
tances greater than 50km (the nominal dis- source complex. They include the emissions
tance to which EPA considers most Gaussian resulting from the industrial process that
models applicable) is an important one espe- are not captured and vented through a stack
cially when considering the effects from sec- but may be released from various locations
ondary pollutants. Unfortunately, models within the complex. Where such fugitive
submitted to EPA have not as yet undergone emissions can be properly specified, the ISC
sufficient field evaluation to be rec- model, with consideration of gravitational
ommended for general use. Existing data settling and dry deposition, is the rec-
bases from field studies at mesoscale and ommended model. In some unique cases a
long range transport distances are limited in model developed specifically for the situa-
detail. This limitation is a result of the ex- tion may be needed.
pense to perform the field studies required to c. Due to the difficult nature of character-
verify and improve mesoscale and long range izing and modeling fugitive dust and fugitive
transport models. Particularly important emissions, it is recommended that the pro-
and sparse are meteorological data adequate posed procedure be cleared by the appro-
for generating three dimensional wind fields. priate Regional Office for each specific situa-
Application of models to complicated terrain tion before the modeling exercise is begun.
compounds the difficulty. EPA has com-
pleted limited evaluation of several long 7.2.2 Particulate Matter
range transport (LRT) models against two a. The particulate matter NAAQS, promul-
sets of field data. The evaluation results are gated on July 1, 1987 (52 FR 24634), includes
discussed in the document, ‘‘Evaluation of only particles with an aerodynamic diameter
Short-Term Long-Range Transport Mod- less than or equal to a nominal 10 microm-
els.’’ 99 100 For the time being, long range and eters (PM–10). EPA promulgated regulations
mesoscale transport models must be evalu- for PSD increments measured as PM–10 on
ated for regulatory use on a case-by-case June 3, 1993 (58 FR 31621), which are codified
basis. at §§ 51.166(c) and 52.21(c).
d. There are several regulatory programs b. Screening techniques like those identi-
for which air pathway analysis procedures fied in section 4 are also applicable to PM–10
and modeling techniques have been devel- and to large particles. It is recommended
oped. For continuous emission releases, ISC that subjectively determined values for
forms the basis of many analytical tech- ‘‘half-life’’ or pollutant decay not be used as

362
Environmental Protection Agency Pt. 51, App. W
a surrogate for particle removal. Conserv- lead air quality standard around specified
ative assumptions which do not allow re- lead point sources. For other areas reporting
moval or transformation are suggested for a violation of the lead standard, § 51.85 re-
screening. Proportional models (rollback/for- quires an analysis of the area in the vicinity
ward) may not be applied for screening anal- of the monitor reporting the violation. The
ysis, unless such techniques are used in con- NAAQS for lead is a quarterly (three month)
junction with receptor modeling. average, thus requiring the use of modeling
c. Refined models such as those in section techniques that can provide long-term con-
4.0 are recommended for PM–10 and large centration estimates.
particles. However, where possible, particle b. The SIP should contain an air quality
size, gas-to-particle formation, and their ef- analysis to determine the maximum quar-
fect on ambient concentrations may be con- terly lead concentration resulting from
sidered. For urban-wide refined analyses major lead point sources, such as smelters,
CDM 2.0 (long term) or RAM (short term) gasoline additive plants, etc. For these appli-
should be used. ISC is recommended for point cations the ISC model is preferred, since the
sources of small particles and for source-spe- model can account for deposition of particles
cific analyses of complicated sources. No and the impact of fugitive emissions. If the
model recommended for general use at this source is located in complicated terrain or is
time accounts for secondary particulate for- subject to unusual climatic conditions, a
mation or other transformations in a man- case-specific review by the appropriate Re-
ner suitable for SIP control strategy dem- gional Office may be required.
onstrations. Where possible, the use of recep- c. In modeling the effect of traditional line
tor models 38 39 105 106 107 in conjunction with sources (such as a specific roadway or high-
dispersion models is encouraged to more pre- way) on lead air quality, dispersion models
cisely characterize the emissions inventory applied for other pollutants can be used. Dis-
and to validate source specific impacts cal- persion models such as CALINE3 have been
culated by the dispersion model. A SIP de- widely used for modeling carbon monoxide
velopment guideline,108 model reconciliation emissions from highways. However, where
guidance,106 and an example model applica- deposition is of concern, the line source
tion 109 are available to assist in PM–10 anal- treatment in ISC may be used. Also, where
yses and control strategy development. there is a point source in the middle of a sub-
d. Under certain conditions, recommended stantial road network, the lead concentra-
dispersion models are not available or appli- tions that result from the road network
cable. In such circumstances, the modeling should be treated as background (see section
approach should be approved by the appro- 9.2); the point source and any nearby major
priate Regional Office on a case-by-case roadways should be modeled separately using
basis. For example, where there is no rec- the ISC model.
ommended air quality model and area d. To model an entire major urban area or
sources are a predominant component of to model areas without significant sources of
PM–10, an attainment demonstration may be lead emissions, as a minimum a proportional
based on rollback of the apportionment de- (rollback) model may be used for air quality
rived from two reconciled receptor models, if analysis. The rollback philosophy assumes
the strategy provides a conservative dem- that measured pollutant concentrations are
onstration of attainment. At this time, anal- proportional to emissions. However, urban or
yses involving model calculations for dis- other dispersion models are encouraged in
tances beyond 50km and under stagnation these circumstances where the use of such
conditions should also be justified on a case- models is feasible.
by-case basis (see sections 7.2.6 and 8.2.10). e. For further information concerning the
e. As an aid to assessing the impact on am- use of models in the development of lead im-
bient air quality of particulate matter gen- plementation plans, the documents ‘‘Supple-
erated from prescribed burning activities, mentary Guidelines for Lead Implementa-
reference 110 is available. tion Plans,’’ 40 and ‘‘Updated Information on
7.2.3 Lead Approval and Promulgation of Lead Imple-
mentation Plans,’’ 41 should be consulted.
a. The air quality analyses required for
lead implementation plans are given in 7.2.4. Visibility
§§ 51.83, 51.84 and 51.85. Sections 51.83 and a. The visibility regulations as promul-
51.85 require the use of a modified rollback
gated in December 1980 b require consider-
model as a minimum to demonstrate attain-
ation of the effect of new sources on the visi-
ment of the lead air quality standard but the
bility values of Federal Class I areas. The
use of a dispersion model is the preferred ap-
state of scientific knowledge concerning
proach. Section 51.83 requires the analysis of
identifying, monitoring, modeling, and con-
an entire urban area if the measured lead
trolling visibility impairment is contained
concentration in the urbanized area exceeds
in an EPA report ‘‘Protecting Visibility: An
a quarterly (three month) average of 4.0 µg/
m3. Section 51.84 requires the use of a disper-
sion model to demonstrate attainment of the b § 51.300–307.

363
Pt. 51, App. W 40 CFR Ch. I (7–1–97 Edition)
EPA Report to Congress’’.42 In 1985, EPA pro- Gaussian models are considered accurate for
mulgated Federal Implementation Plans setting emission limits. Since in many cases
(FIPs) for states without approved visibility PSD analyses may show that Class I areas
provisions in their SIPs. A monitoring plan may be threatened at distances greater than
was established as part of the FIPs.c 50km from new sources, some procedure is
b. Guidance and a screening model, needed to (1) determine if a significant im-
VISCREEN, is contained in the EPA docu- pact will occur, and (2) identify the model to
ment ‘‘Workbook for Plume Visual Impact be used in setting an emission limit if the
Screening and Analysis (Revised).’’ 43 Class I increments are threatened (models
VISCREEN can be used to calculate the po- for this purpose should be approved for use
tential impact of a plume of specified emis- on a case-by-case basis as required in section
sions for specific transport and dispersion 3.2). This procedure and the models selected
conditions. If a more comprehensive analysis for use should be determined in consultation
is required, any refined model should be se- with the EPA Regional Office and the appro-
lected in consultation with the EPA Re- priate Federal Land Manager (FLM). While
gional Office and the appropriate Federal the ultimate decision on whether a Class I
Land Manager who is responsible for deter- area is adversely affected is the responsibil-
mining whether there is an adverse effect by ity of the permitting authority, the FLM has
a plume on a Class I area. an affirmative responsibility to protect air
c. PLUVUE II, listed in appendix B, may be quality related values that may be affected.
applied on a case-by-case basis when refined b. If LRT is determined to be important,
plume visibility evaluations are needed. then estimates utilizing an appropriate re-
Plume visibility models have been evaluated fined model for receptors at distances great-
against several data sets.44, 45 er than 50 km should be obtained.
7.2.5 Good Engineering Practice Stack MESOPUFF II, listed in appendix B, may be
Height applied on a case-by-case basis when LRT es-
timates are needed. Additional information
a. The use of stack height credit in excess on applying this model is contained in the
of Good Engineering Practice (GEP) stack EPA document ‘‘A Modeling Protocol For
height or credit resulting from any other dis- Applying MESOPUFF II to Long Range
persion technique is prohibited in the devel- Transport Problems’’. 111
opment of emission limitations by §§ 51.118
and 51.164. The definitions of GEP stack 7.2.7 Modeling Guidance for Other
height and dispersion technique are con- Governmental Programs
tained in § 51.100. Methods and procedures for
a. When using the models recommended or
making the appropriate stack height cal-
discussed in the Guideline in support of pro-
culations, determining stack height credits
grammatic requirements not specifically
and an example of applying those techniques
covered by EPA regulations, the model user
are found in references 46, 47, 48, and 49.
b. If stacks for new or existing major should consult the appropriate Federal or
sources are found to be less than the height State agency to ensure the proper applica-
defined by EPA’s refined formula for deter- tion and use of that model. For modeling as-
mining GEP height, d then air quality im- sociated with PSD permit applications that
pacts associated with cavity or wake effects involve a Class I area, the appropriate Fed-
due to the nearby building structures should eral Land Manager should be consulted on
be determined. Detailed downwash screening all modeling questions.
procedures 18 for both the cavity and wake b. The Offshore and Coastal Dispersion
regions should be followed. If more refined (OCD) model 112 was developed by the Min-
concentration estimates are required, the In- erals Management Service and is rec-
dustrial Source Complex (ISC) model con- ommended for estimating air quality impact
tains algorithms for building wake calcula- from offshore sources on onshore, flat ter-
tions and should be used. Fluid modeling can rain areas. The OCD model is not rec-
provide a great deal of additional informa- ommended for use in air quality impact as-
tion for evaluating and describing the cavity sessments for onshore sources. Sources lo-
and wake effects. cated on or just inland of a shoreline where
fumigation is expected should be treated in
7.2.6 Long Range Transport (LRT) (i.e., accordance with section 8.2.9.
beyond 50km) c. The Emissions and Dispersion Modeling
System (EDMS) 113 was developed by the Fed-
a. Section 165(e) of the Clean Air Act re- eral Aviation Administration and the United
quires that suspected significant impacts on States Air Force and is recommended for air
PSD Class I areas be determined. However, quality assessment of primary pollutant im-
50km is the useful distance to which most pacts at airports or air bases. Regulatory ap-
plication of EDMS is intended for estimating
c § 51.300–307. the cumulative effect of changes in aircraft
d The EPA refined formula height is defined operations, point source, and mobile source
as H + 1.5L (see Reference 46). emissions on pollutant concentrations. It is

364
Environmental Protection Agency Pt. 51, App. W
not intended for PSD, SIP, or other regu- 8.2 Recommendations
latory air quality analyses of point or mobile
sources at or peripheral to airport property 8.2.1 Design Concentrations
that are independent of changes in aircraft
8.2.1.1 Design Concentrations for Criteria
operations. If changes in other than aircraft Pollutants With Deterministic Standards
operations are associated with analyses, a
model recommended in Chapter 4, 5, or 6 a. An air quality analysis for SO2, CO, Pb,
should be used. and NO2 is required to determine if the
source will (1) Cause a violation of the
7.2.8 Air Pathway Analyses (Air Toxics and NAAQS, or (2) cause or contribute to air
Hazardous Waste) quality deterioration greater than the speci-
fied allowable PSD increment. For the
a. Modeling is becoming an increasingly former, background concentration (see sec-
important tool for regulatory control agen- tion 9.2) should be added to the estimated
cies to assess the air quality impact of re- impact of the source to determine the design
leases of toxics and hazardous waste mate- concentration. For the latter, the design
rials. Appropriate screening techniques 114 115 concentration includes impact from all in-
for calculating ambient concentrations due crement consuming sources.
to various well-defined neutrally buoyant b. If the air quality analyses are conducted
toxic/hazardous pollutant releases are avail- using the period of meteorological input data
able. recommended in section 9.3.1.2 (e.g., 5 years
b. Several regulatory programs within of NWS data or 1 year of site-specific data),
EPA have developed modeling techniques then the design concentration based on the
and guidance for conducting air pathway highest, second-highest short term con-
analyses as noted in references 116–129. ISC centration or long term average, whichever
forms the basis of the modeling procedures is controlling, should be used to determine
for air pathway analyses of many of these emission limitations to assess compliance
regulatory programs and, where identified, is with the NAAQS and to determine PSD in-
crements.
appropriate for obtaining refined ambient
c. When sufficient and representative data
concentration estimates of neutrally buoy-
exist for less than a 5-year period from a
ant continuous air toxic releases from tradi-
nearby NWS site, or when on-site data have
tional sources. Appendix A to the Guideline been collected for less than a full continuous
contains additional models appropriate for year, or when it has been determined that
obtaining refined estimates of continuous air the on site data may not be temporally rep-
toxic releases from traditional sources. Ap- resentative, then the highest concentration
pendix B contains models that may be used estimate should be considered the design
on a case-by-case basis for obtaining refined value. This is because the length of the data
estimates of denser-than-air intermittent record may be too short to assure that the
gaseous releases, e.g., DEGADIS; 130 guidance conditions producing worst-case estimates
for the use of such models is also avail- have been adequately sampled. The highest
able. 131 value is then a surrogate for the concentra-
c. Many air toxics models require input of tion that is not to be exceeded more than
chemical properties and/or chemical engi- once per year (the wording of the determinis-
neering variables in order to appropriately tic standards). Also, the highest concentra-
characterize the source emissions prior to tion should be used whenever selected worst-
dispersion in the atmosphere; reference 132 is case conditions are input to a screening
one source of helpful data. In addition, EPA technique. This specifically applies to the
has numerous programs to determine emis- use of techniques such as outlined in
sion factors and other estimates of air toxic ‘‘Screening Procedures for Estimating the
Air Quality Impact of Stationary Sources,
emissions. The Regional Office should be
Revised’’. 18 Specific guidance for CO may be
consulted for guidance on appropriate emis-
found in the ‘‘Guideline for Modeling Carbon
sion estimating procedures and any uncer-
Monoxide from Roadway Intersections’’. 34
tainties that may be associated with them.
d. If the controlling concentration is an
8.0 GENERAL MODELING CONSIDERATIONS annual average value and multiple years of
data (on-site or NWS) are used, then the de-
8.1 Discussion sign value is the highest of the annual aver-
ages calculated for the individual years. If
a. This section contains recommendations the controlling concentration is a quarterly
concerning a number of different issues not average and multiple years are used, then
explicitly covered in other sections of this the highest individual quarterly average
guide. The topics covered here are not spe- should be considered the design value.
cific to any one program or modeling area e. As long a period of record as possible
but are common to nearly all modeling anal- should be used in making estimates to deter-
yses. mine design values and PSD increments. If

365
Pt. 51, App. W 40 CFR Ch. I (7–1–97 Edition)
more than 1 year of site-specific data is the given situation than are algorithms con-
available, it should be used. tained in the preferred models.
c. Buoyancy-induced dispersion (BID), as
8.2.1.2 Design Concentrations for Criteria identified by Pasquill, 54 is included in the
Pollutants With Expected Exceedance preferred models and should be used where
Standards buoyant sources, e.g., those involving fuel
combustion, are involved.
a. Specific instructions for the determina-
tion of design concentrations for criteria pol- 8.2.4 Stability Categories
lutants with expected exceedance standards,
ozone and PM–10, are contained in special a. The Pasquill approach to classifying sta-
guidance documents for the preparation of bility is generally required in all preferred
SIPs for those pollutants. 86 108 For all SIP re- models (Appendix A). The Pasquill method,
visions the user should check with the Re- as modified by Turner, 55 was developed for
gional Office to obtain the most recent guid- use with commonly observed meteorological
ance documents and policy memoranda con- data from the National Weather Service and
cerning the pollutant in question. is based on cloud cover, insolation and wind
speed.
8.2.2 Critical Receptor Sites b. Procedures to determine Pasquill stabil-
ity categories from other than NWS data are
a. Receptor sites for refined modeling found in subsection 9.3. Any other method to
should be utilized in sufficient detail to esti- determine Pasquill stability categories must
mate the highest concentrations and possible be justified on a case-by-case basis.
violations of a NAAQS or a PSD increment. c. For a given model application where sta-
In designing a receptor network, the empha- bility categories are the basis for selecting
sis should be placed on receptor resolution dispersion coefficients, both σy and σz should
and location, not total number of receptors. be determined from the same stability cat-
The selection of receptor sites should be a egory. ‘‘Split sigmas’’ in that instance are
case-by-case determination taking into con- not recommended.
sideration the topography, the climatology, d. Sector averaging, which eliminates the
monitor sites, and the results of the σy term, is generally acceptable only to de-
initialscreening procedure. For large sources termine long term averages, such as seasonal
(those equivalent to a 500MW power plant) or annual, and when the meteorological
and where violations of the NAAQS or PSD input data are statistically summarized as in
increment are likely, 360 receptors for a the STAR summaries. Sector averaging is,
polar coordinate grid system and 400 recep- however, commonly acceptable in complex
tors for a rectangular grid system, where the terrain screening methods.
distance from the source to the farthest re-
ceptor is 10km, are usually adequate to iden- 8.2.5 Plume Rise
tify areas of high concentration. Additional
a. The plume rise methods of Briggs 56 57 are
receptors may be needed in the high con-
incorporated in the preferred models and are
centration location if greater resolution is
recommended for use in all modeling appli-
indicated by terrain or source factors.
cations. No provisions in these models are
8.2.3 Dispersion Coefficients made for fumigation or multistack plume
rise enhancement or the handling of such
a. Gaussian models used in most applica- special plumes as flares; these problems
tions should employ dispersion coefficients should be considered on a case-by-case basis.
consistent with those contained in the pre- b. Since there is insufficient information
ferred models in appendix A. Factors such as to identify and quantify dispersion during
averaging time, urban/rural surroundings, the transitional plume rise period, gradual
and type of source (point vs. line) may dic- plume rise is not generally recommended for
tate the selection of specific coefficients. use. There are two exceptions where the use
Generally, coefficients used in appendix A of gradual plume rise is appropriate: (1) In
models are identical to, or at least based on, complex terrain screening procedures to de-
Pasquill-Gifford coefficients 50 in rural areas termine close-in impacts; (2) when calculat-
and McElroy-Pooler 51 coefficients in urban ing the effects of building wakes. The build-
areas. ing wake algorithm in the ISC model incor-
b. Research is continuing toward the devel- porates and automatically (i.e., internally)
opment of methods to determine dispersion exercises the gradual plume rise calcula-
coefficients directly from measured or ob- tions. If the building wake is calculated to
served variables. 52 53 No method to date has affect the plume for any hour, gradual plume
proved to be widely applicable. Thus, direct rise is also used in downwind dispersion cal-
measurement, as well as other dispersion co- culations to the distance of final plume rise,
efficients related to distance and stability, after which final plume rise is used.
may be used in Gaussian modeling only if a c. Stack tip downwash generally occurs
demonstration can be made that such param- with poorly constructed stacks and when the
eters are more applicable and accurate for ratio of the stack exit velocity to wind speed

366
Environmental Protection Agency Pt. 51, App. W
is small. An algorithm developed by Briggs ticulate matter sources can be quantified
(Hanna et al.) 57 is the recommended tech- and settling and deposition are problems.
nique for this situation and is found in the
point source preferred models. 8.2.8 Urban/Rural Classification
d. Where aerodynamic downwash occurs a. The selection of either rural or urban
due to the adverse influence of nearby struc- dispersion coefficients in a specific applica-
tures, the algorithms included in the ISC tion should follow one of the procedures sug-
model 58 should be used. gested by Irwin 59 and briefly described
below. These include a land use classifica-
8.2.6 Chemical Transformation
tion procedure or a population based proce-
a. The chemical transformation of SO2 dure to determine whether the character of
emitted from point sources or single indus- an area is primarily urban or rural.
trial plants in rural areas is generally as- b. Land Use Procedure: (1) Classify the
sumed to be relatively unimportant to the land use within the total area, Ao, cir-
estimation of maximum concentrations cumscribed by a 3km radius circle about the
when travel time is limited to a few hours. source using the meteorological land use
However, in urban areas, where synergistic typing scheme proposed by Auer 60; (2) if land
effects among pollutants are of considerable use types I1, I2, C1, R2, and R3 account for 50
consequence, chemical transformation rates percent or more of Ao, use urban dispersion
may be of concern. In urban area applica- coefficients; otherwise, use appropriate rural
tions, a half-life of 4 hours 55 may be applied dispersion coefficients.
to the analysis of SO2 emissions. Calcula- c. Population Density Procedure: (1) Com-
tions of transformation coefficients from pute the average population density, p̄ per
site-specific studies can be used to define a square kilometer with Ao as defined above;
‘‘half-life’’ to be used in a Gaussian model (2) If p̄ is greater than 750 people/km2, use
with any travel time, or in any application, urban dispersion coefficients; otherwise use
if appropriate documentation is provided. appropriate rural dispersion coefficients.
Such conversion factors for pollutant half- d. Of the two methods, the land use proce-
life should not be used with screening analy- dure is considered more definitive. Popu-
ses. lation density should be used with caution
b. Complete conversion of NO to NO2 and should not be applied to highly industri-
should be assumed for all travel time when alized areas where the population density
simple screening techniques are used to may be low and thus a rural classification
model point source emissions of nitrogen ox- would be indicated, but the area is suffi-
ides. If a Gaussian model is used, and data ciently built-up so that the urban land use
are available on seasonal variations in maxi- criteria would be satisfied. In this case, the
mum ozone concentrations, the Ozone Limit- classification should already be ‘‘urban’’ and
ing Method 36 is recommended. In refined urban dispersion parameters should be used.
analyses, case-by case conversion rates based e. Sources located in an area defined as
on technical studies appropriate to the site urban should be modeled using urban disper-
in question may be used. The use of more so- sion parameters. Sources located in areas de-
phisticated modeling techniques should be fined as rural should be modeled using the
justified for individual cases. rural dispersion parameters. For analyses of
c. Use of models incorporating complex whole urban complexes, the entire area
chemical mechanisms should be considered should be modeled as an urban region if most
only on a case-by-case basis with proper of the sources are located in areas classified
demonstration of applicability. These are as urban.
generally regional models not designed for
the evaluation of individual sources but used 8.2.9 Fumigation
primarily for region-wide evaluations. Visi- a. Fumigation occurs when a plume (or
bility models also incorporate chemical multiple plumes) is emitted into a stable
transformation mechanisms which are an in- layer of air and that layer is subsequently
tegral part of the visibility model itself and mixed to the ground either through convec-
should be used in visibility assessments. tive transfer of heat from the surface or be-
cause of advection to less stable surround-
8.2.7 Gravitational Settling and Deposition
ings. Fumigation may cause excessively high
a. An ‘‘infinite half-life’’ should be used for concentrations but is usually rather short-
estimates of particle concentrations when lived at a given receptor. There are no rec-
Gaussian models containing only expo- ommended refined techniques to model this
nential decay terms for treating settling and phenomenon. There are, however, screening
deposition are used. procedures (see ‘‘Screening Procedures for
b. Gravitational settling and deposition Estimating the Air Quality Impact of Sta-
may be directly included in a model if either tionary Sources’’ 18) that may be used to ap-
is a significant factor. One preferred model proximate the concentrations. Considerable
(ISC) contains a settling and deposition algo- care should be exercised in using the results
rithm and is recommended for use when par- obtained from the screening techniques.

367
Pt. 51, App. W 40 CFR Ch. I (7–1–97 Edition)
b. Fumigation is also an important phe- estimate the concentration at an exact loca-
nomenon on and near the shoreline of bodies tion for a specific increment of time. Such
of water. This can affect both individual uncertainties make calibration of short term
plumes and area-wide emissions. When fumi- models of questionable benefit. Therefore,
gation conditions are expected to occur from short term model calibration is unaccept-
a source or sources with tall stacks located able.
on or just inland of a shoreline, this should
be addressed in the air quality modeling 9.0 MODEL INPUT DATA
analysis. The Shoreline Dispersion Model
a. Data bases and related procedures for es-
(SDM) listed in appendix B may be applied
timating input parameters are an integral
on a case-by-case basis when air quality esti-
part of the modeling procedure. The most ap-
mates under shoreline fumigation conditions
propriate data available should always be se-
are needed.133 Information on the results of
lected for use in modeling analyses. Con-
EPA’s evaluation of this model together
centrations can vary widely depending on
with other coastal fumigation models may
the source data or meteorological data used.
be found in reference 134. Selection of the ap-
Input data are a major source of inconsist-
propriate model for applications where
encies in any modeling analysis. This section
shoreline fumigation is of concern should be
attempts to minimize the uncertainty asso-
determined in consultation with the Re-
ciated with data base selection and use by
gional Office.
identifying requirements for data used in
8.2.10 Stagnation modeling. A checklist of input data require-
ments for modeling analyses is included as
a. Stagnation conditions are characterized appendix C. More specific data requirements
by calm or very low wind speeds, and vari- and the format required for the individual
able wind directions. These stagnant mete- models are described in detail in the users’
orological conditions may persist for several guide for each model.
hours to several days. During stagnation
conditions, the dispersion of air pollutants, 9.1 Source Data
especially those from low-level emissions
sources, tends to be minimized, potentially 9.1.1 Discussion
leading to relatively high ground-level con-
centrations. a. Sources of pollutants can be classified as
b. When stagnation periods such as these point, line and area/volume sources. Point
are found to occur, they should be addressed sources are defined in terms of size and may
in the air quality modeling analysis. vary between regulatory programs. The line
WYNDvalley, listed in appendix B, may be sources most frequently considered are road-
applied on a case-by-case basis for stagna- ways and streets along which there are well-
tion periods of 24 hours or longer in valley- defined movements of motor vehicles, but
type situations. Caution should be exercised they may be lines of roof vents or stacks
when applying the model to elevated point such as in aluminum refineries. Area and
sources. Users should consult with the appro- volume sources are often collections of a
priate Regional Office prior to regulatory ap- multitude of minor sources with individually
plication of WYNDvalley. small emissions that are impractical to con-
sider as separate point or line sources. Large
8.2.11 Calibration of Models area sources are typically treated as a grid
network of square areas, with pollutant
a. Calibration of long term multi-source
emissions distributed uniformly within each
models has been a widely used procedure
grid square.
even though the limitations imposed by sta-
b. Emission factors are compiled in an EPA
tistical theory on the reliability of the cali-
publication commonly known as AP–42 62; an
bration process for long term estimates are
indication of the quality and amount of data
well known. 61 In some cases, where a more
on which many of the factors are based is
accurate model is not available, calibration
also provided. Other information concerning
may be the best alternative for improving
emissions is available in EPA publications
the accuracy of the estimated concentra-
relating to specific source categories. The
tions needed for control strategy evalua-
Regional Office should be consulted to deter-
tions.
mine appropriate source definitions and for
b. Calibration of short term models is not
guidance concerning the determination of
common practice and is subject to much
emissions from and techniques for modeling
greater error and misunderstanding. There
the various source types.
have been attempts by some to compare
short term estimates and measurements on 9.1.2 Recommendations
an event-by-event basis and then to calibrate
a model with results of that comparison. a. For point source applications the load or
This approach is severely limited by uncer- operating condition that causes maximum
tainties in both source and meteorological ground-level concentrations should be estab-
data and therefore it is difficult to precisely lished. As a minimum, the source should be

368
Environmental Protection Agency Pt. 51, App. W
modeled using the design capacity (100 per- g. Pollution control equipment param-
cent load). If a source operates at greater eters. For each boiler served and each pollut-
than design capacity for periods that could ant affected, the type of emission control
result in violations of the standards or PSD equipment, the year of its installation, its
increments, this load e should be modeled. design efficiency and mass emission rate, the
Where the source operates at substantially data of the last test and the tested effi-
less than design capacity, and the changes in ciency, the number of hours of operation
the stack parameters associated with the op- during the latest year, and the best engineer-
erating conditions could lead to higher ing estimate of its projected efficiency if
ground level concentrations, loads such as 50 used in conjunction with coal combustion;
percent and 75 percent of capacity should data for any anticipated modifications or ad-
also be modeled. A range of operating condi- ditions.
tions should be considered in screening anal- h. Data for new boilers or stacks. For all
yses; the load causing the highest concentra- new boilers and stacks under construction
tion, in addition to the design load, should and for all planned modifications to existing
be included in refined modeling. For a power boilers or stacks, the scheduled date of com-
plant, the following paragraphs b through h pletion, and the data or best estimates avail-
of this section describe the typical kind of able for paragraphs b through g of this sec-
data on source characteristics and operating tion above following completion of construc-
conditions that may be needed. Generally, tion or modification.
input data requirements for air quality mod- i. In stationary point source applications
els necessitate the use of metric units; where for compliance with short term ambient
English units are common for engineering standards, SIP control strategies should be
usage, a conversion to metric is required. tested using the emission input shown on
b. Plant layout. The connection scheme be- table 9–1. When using a refined model,
tween boilers and stacks, and the distance sources should be modeled sequentially with
and direction between stacks, building pa- these loads for every hour of the year. To
rameters (length, width, height, location and evaluate SIPs for compliance with quarterly
orientation relative to stacks) for plant and annual standards, emission input data
structures which house boilers, control shown in table 9–1 should again be used.
equipment, and surrounding buildings within Emissions from area sources should gen-
a distance of approximately five stack erally be based on annual average condi-
heights. tions. The source input information in each
c. Stack parameters. For all stacks, the model user’s guide should be carefully con-
stack height and inside diameter (meters), sulted and the checklist in appendix C should
and the temperature (K) and volume flow also be consulted for other possible emission
rate (actual cubic meters per second) or exit data that could be helpful. PSD NAAQS com-
gas velocity (meters per second) for oper- pliance demonstrations should follow the
ation at 100 percent, 75 percent and 50 per- emission input data shown in table 9–2. For
cent load. purposes of emissions trading, new source re-
d. Boiler size. For all boilers, the associ- view and demonstrations, refer to current
ated megawatts, 106 BTU/hr, and pounds of EPA policy and guidance to establish input
steam per hour, and the design and/or actual data.
fuel consumption rate for 100 percent load j. Line source modeling of streets and high-
for coal (tons/hour), oil (barrels/hour), and ways requires data on the width of the road-
natural gas (thousand cubic feet/hour). way and the median strip, the types and
e. Boiler parameters. For all boilers, the amounts of pollutant emissions, the number
percent excess air used, the boiler type (e.g., of lanes, the emissions from each lane and
wet bottom, cyclone, etc.), and the type of the height of emissions. The location of the
firing (e.g., pulverized coal, front firing, ends of the straight roadway segments
etc.). should be specified by appropriate grid co-
f. Operating conditions. For all boilers, the ordinates. Detailed information and data re-
type, amount and pollutant contents of fuel, quirements for modeling mobile sources of
the total hours of boiler operation and the pollution are provided in the user’s manuals
boiler capacity factor during the year, and for each of the models applicable to mobile
the percent load for peak conditions. sources.
k. The impact of growth on emissions
e Malfunctions which may result in excess should be considered in all modeling analy-
emissions are not considered to be a normal ses covering existing sources. Increases in
operating condition. They generally should emissions due to planned expansion or
not be considered in determining allowable planned fuel switches should be identified.
emissions. However, if the excess emissions Increases in emissions at individual sources
are the result of poor maintenance, careless that may be associated with a general indus-
operation, or other preventable conditions, it trial/commercial/residential expansion in
may be necessary to consider them in deter- multi-source urban areas should also be
mining source impact. treated. For new sources the impact of

369
Pt. 51, App. W 40 CFR Ch. I (7–1–97 Edition)
growth on emissions should generally be con- emissions which were not subject to
sidered for the period prior to the start-up preconstruction review, and emissions due to
date for the source. Such changes in emis- sources with permits to construct that have
sions should treat increased area source not yet started operation.
emissions, changes in existing point source

370
Environmental Protection Agency Pt. 51, App. W

TABLE 9–1— MODEL EMISSION INPUT DATA FOR POINT SOURCES 1


Operating level (MMBtu/ Operating factor (e.g., hr/
Averaging time Emission limit (#/MMBtu) 2 × ×
hr) 2 yr, hr/day)

Stationary Point Source(s) Subject to SIP Emission Limit(s) Evaluation for Compliance with Ambient Standards
(Including Areawide Demonstrations)

Annual & quarterly ........... Maximum allowable emis- Actual or design capacity Actual operating factor
sion limit or federally en- (whichever is greater), or averaged over most re-
forceable permit limit. federally enforceable cent 2 years.3
permit condition.
Short term ........................ Maximum allowable emis- Actual or design capacity Continuous operation, i.e.,
sion limit or federally en- (whichever is greater), or all hours of each time
forceable permit limit. federally enforceable period under consider-
permit condition 4. ation (for all hours of the
meteorological data
base).5

Nearby Background Source(s)—Same input requirements as for stationary point source(s) above.

Other Background Source(s)—If modeled (see section 9.2.3), input data requirements are defined below.

Annual & quarterly ........... Maximum allowable emis- Annual level when actually Actual operating factor
sion limit or federal en- operating, averaged over averaged over the most
forceable permit limit. the most recent 2 recent 2 years.3
years 3.
Short term ........................ Maximum allowable emis- Annual level when actually Continuous operation, i.e.,
sion limit or federally en- operating, averaged over all hours of each time
forceable permit limit. the most recent 2 period under consider-
years 3. ation (for all hours of the
meteorological data
base).5
1 The model input data requirements shown on this table apply to stationary source control strategies for STATE IMPLEMEN-
TATION PLANS. For purposes of emissions trading, new source review, or prevention of significant deterioration, other model
input criteria may apply. Refer to the policy and guidance for these programs to establish the input data.
2 Terminology applicable to fuel burning sources; analogous terminology (e.g., #/throughput) may be used for other types of
sources.
3 Unless it is determined that this period is not representative.
4 Operating levels such as 50 percent and 75 percent of capacity should also be modeled to determine the load causing the
highest concentration.
5 If operation does not occur for all hours of the time period of consideration (e.g., 3 or 24 hours) and the source operation is
constrained by a federally enforceable permit condition, an appropriate adjustment to the modeled emission rate may be made
(e.g., if operation is only 8:00 a.m. to 4:00 p.m. each day, only these hours will be modeled with emissions from the source.
Modeled emissions should not be averaged across non-operating time periods.)

TABLE 9–2—POINT SOURCE MODEL INPUT DATA (EMISSIONS) FOR PSD NAAQS COMPLIANCE
DEMONSTRATIONS
Operating level (MMBtu/ Operating factor (e.g., hr/
Averaging time Emission limit (#/MMBtu) 1 × ×
hr) 1 yr, hr/day)

Proposed Major New or Modified Source

Annual & quarterly ........... Maximum allowable emis- Design capacity or feder- Continuous operation (i.e.,
sion limit or federally en- ally enforceable permit 8760 hours).2
forceable permit limit. condition.
Short term (≤ 24 hours) ... Maximum allowable emis- Design capacity or feder- Continuous operation (i.e.,
sion limit or federally en- ally enforceable permit all hours of each time
forceable permit limit. condition.3 period under consider-
ation) (for all hours of the
meteorological data
base).2

Nearby Background Source(s) 4

Annual & quarterly ........... Maximum allowable emis- Actual or design capacity Actual operating factor
sion limit or federally en- (whichever is greater), or averaged over the most
forceable permit limit. federally enforceable recent 2 years.5 7
permit condition.

371
Pt. 51, App. W 40 CFR Ch. I (7–1–97 Edition)

TABLE 9–2—POINT SOURCE MODEL INPUT DATA (EMISSIONS) FOR PSD NAAQS COMPLIANCE
DEMONSTRATIONS—Continued
Operating level (MMBtu/ Operating factor (e.g., hr/
Averaging time Emission limit (#/MMBtu) 1 × ×
hr) 1 yr, hr/day)

Short term (≤ 24 hours) ... Maximum allowable emis- Actual or design capacity Continuous operation (i.e.,
sion limit or federally en- (whichever is greater), or all hours of each time
forceable permit limit. federally enforceable period under consider-
permit condition.3 ation) (for all hours of the
meteorological data
base).2

Other Background Source(s) 6

Annual & quarterly ........... Maximum allowable emis- Annual level when actually Actual operating factor
sion limit or federally en- operating, averaged over averaged over the most
forceable permit limit. the most recent 2 years.5 recent 2 years.5 7
Short term (≤ 24 hours) ... Maximum allowable emis- Annual level when actually Continuous operation (i.e.,
sion limit or federally en- operating, averaged over all hours of each time
forceable permit limit. the most recent 2 years.5 period under consider-
ation) (for all hours of the
meteorological data
base).2
1 Terminology applicable to fuel burning sources; analogous terminology (e.g., #/throughput) may be used for other types of
sources.
2hnsp;If operation does not occur for all hours of the time period of consideration (e.g., 3 or 24 hours) and the source oper-
ation is constrained by a federally enforceable permit condition, an appropriate adjustment to the modeled emission rate may be
made (e.g., if operation is only 8:00 a.m. to 4:00 p.m. each day, only these hours will be modeled with emissions from the
source. Modeled emissions should not be averaged across non-operating time periods.
3 Operating levels such as 50 percent and 75 percent of capacity should also be modeled to determine the load causing the
highest concentration.
4 Includes existing facility to which modification is proposed if the emissions from the existing facility will not be affected by the
modification. Otherwise use the same parameters as for major modification.
5 Unless it is determined that this period is not representative.
6 Generally, the ambient impacts from non-nearby background sources can be represented by air quality data unless adequate
data do not exist.
7 For those permitted sources not yet in operation or that have not established an appropriate factor, continuous operation (i.e.,
8760 hours) should be used.

9.2 Background Concentrations 9.2.2 Recommendations (Isolated Single


Source)
9.2.1 Discussion
a. Two options (paragraph b or c of this
a. Background concentrations are an es- section) are available to determine the back-
sential part of the total air quality con- ground concentration near isolated sources.
centration to be considered in determining b. Use air quality data collected in the vi-
source impacts. Background air quality in- cinity of the source to determine the back-
cludes pollutant concentrations due to: (1) ground concentration for the averaging
natural sources; (2) nearby sources other times of concern.f Determine the mean back-
than the one(s) currently under consider- ground concentration at each monitor by ex-
ation; and (3) unidentified sources. cluding values when the source in question is
b. Typically, air quality data should be impacting the monitor. The mean annual
used to establish background concentrations background is the average of the annual con-
in the vicinity of the source(s) under consid- centrations so determined at each monitor.
eration. The monitoring network used for For shorter averaging periods, the meteoro-
background determinations should conform logical conditions accompanying the con-
centrations of concern should be identified.
to the same quality assurance and other re-
Concentrations for meteorological condi-
quirements as those networks established for
tions of concern, at monitors not impacted
PSD purposes. 63 An appropriate data valida- by the source in question, should be averaged
tion procedure should be applied to the data for each separate averaging time to deter-
prior to use. mine the average background value. Mon-
c. If the source is not isolated, it may be itoring sites inside a 90° sector downwind of
necessary to use a multi-source model to es- the source may be used to determine the
tablish the impact of nearby sources. Back-
ground concentrations should be determined
f For purposes of PSD, the location of mon-
for each critical (concentration) averaging
time. itors as well as data quality assurance proce-
dures must satisfy requirements listed in the
PSD Monitoring Guidelines. 63

372
Environmental Protection Agency Pt. 51, App. W
area of impact. One hour concentrations may ceptors of interest and the complex topo-
be added and averaged to determine longer graphic characteristics of the area. Tem-
averaging periods. poral representativeness is a function of the
c. If there are no monitors located in the year-to-year variations in weather condi-
vicinity of the source, a ‘‘regional site’’ may tions.
be used to determine background. A ‘‘re- b. Model input data are normally obtained
gional site’’ is one that is located away from either from the National Weather Service or
the area of interest but is impacted by simi- as part of an on-site measurement program.
lar natural and distant man-made sources. Local universities, Federal Aviation Admin-
istration (FAA), military stations, industry
9.2.3 Recommendations (Multi-Source Areas) and pollution control agencies may also be
a. In multi-source areas, two components sources of such data. Some recommendations
of background should be determined. for the use of each type of data are included
b. Nearby Sources: All sources expected to in this section 9.3.
cause a significant concentration gradient in 9.3.1 Length of Record of Meteorological
the vicinity of the source or sources under Data
consideration for emission limit(s) should be
explicitly modeled. For evaluation for com- 9.3.1.1 Discussion
pliance with the short term and annual am-
bient standards, the nearby sources should a. The model user should acquire enough
meteorological data to ensure that worst-
be modeled using the emission input data
case meteorological conditions are ade-
shown in table 9–1 or 9–2. The number of such
quately represented in the model results.
sources is expected to be small except in un-
The trend toward statistically based stand-
usual situations. The nearby source inven-
ards suggests a need for all meteorological
tory should be determined in consultation
conditions to be adequately represented in
with the reviewing authority. It is envi-
the data set selected for model input. The
sioned that the nearby sources and the
number of years of record needed to obtain a
sources under consideration will be evalu-
stable distribution of conditions depends on
ated together using an appropriate appendix
the variable being measured and has been es-
A model. timated by Landsberg and Jacobs 64 for var-
c. The impact of the nearby sources should ious parameters. Although that study indi-
be examined at locations where interactions cates in excess of 10 years may be required to
between the plume of the point source under achieve stability in the frequency distribu-
consideration and those of nearby sources tions of some meteorological variables, such
(plus natural background) can occur. Signifi- long periods are not reasonable for model
cant locations include: (1) the area of maxi- input data. This is due in part to the fact
mum impact of the point source; (2) the area that hourly data in model input format are
of maximum impact of nearby sources; and frequently not available for such periods and
(3) the area where all sources combine to that hourly calculations of concentration for
cause maximum impact. These locations long periods are prohibitively expensive. A
may be identified through trial and error recent study 65 compared various periods
analyses. from a 17-year data set to determine the
d. Other Sources: That portion of the back- minimum number of years of data needed to
ground attributable to all other sources (e.g., approximate the concentrations modeled
natural sources, minor sources and distant with a 17-year period of meteorological data
major sources) should be determined by the from one station. This study indicated that
procedures found in section 9.2.2 or by appli- the variability of model estimates due to the
cation of a model using table 9–1 or 9–2. meteorological data input was adequately
reduced if a 5-year period of record of mete-
9.3 Meteorological Input Data
orological input was used.
a. The meteorological data used as input to
a dispersion model should be selected on the 9.3.1.2 Recommendations
basis of spatial and climatological (tem- a. Five years of representative meteoro-
poral) representativeness as well as the abil- logical data should be used when estimating
ity of the individual parameters selected to concentrations with an air quality model.
characterize the transport and dispersion Consecutive years from the most recent,
conditions in the area of concern. The rep- readily available 5-year period are preferred.
resentativeness of the data is dependent on: The meteorological data may be data col-
(1) the proximity of the meteorological mon- lected either onsite or at the nearest Na-
itoring site to the area under consideration; tional Weather Service (NWS) station. If the
(2) the complexity of the terrain; (3) the ex- source is large, e.g., a 500MW power plant,
posure of the meteorological monitoring the use of 5 years of NWS meteorological
site; and (4) the period of time during which data or at least 1 year of site-specific data is
data are collected. The spatial representa- required.
tiveness of the data can be adversely affected b. If one year or more, up to five years, of
by large distances between the source and re- site-specific data is available, these data are

373
Pt. 51, App. W 40 CFR Ch. I (7–1–97 Edition)
preferred for use in air quality analyses. agencies may be used if such data are equiva-
Such data should have been subjected to lent in accuracy and detail to the NWS data.
quality assurance procedures as described in
section 9.3.3.2. 9.3.3 Site-Specific Data
c. For permitted sources whose emission
limitations are based on a specific year of 9.3.3.1 Discussion
meteorological data that year should be
a. Spatial or geographical representative-
added to any longer period being used (e.g., 5
ness is best achieved by collection of all of
years of NWS data) when modeling the facil-
ity at a later time. the needed model input data at the actual
site of the source(s). Site-specific measured
9.3.2 National Weather Service Data data are therefore preferred as model input,
provided appropriate instrumentation and
9.3.2.1 Discussion quality assurance procedures are followed
a. The National Weather Service (NWS) and that the data collected are representa-
meteorological data are routinely available tive (free from undue local or ‘‘micro’’ influ-
and familiar to most model users. Although ences) and compatible with the input re-
the NWS does not provide direct measure- quirements of the model to be used. How-
ments of all the needed dispersion model ever, direct measurements of all the needed
input variables, methods have been devel- model input parameters may not be possible.
oped and successfully used to translate the This section discusses suggestions for the
basic NWS data to the needed model input. collection and use of on-site data. Since the
Direct measurements of model input param- methods outlined in this section are still
eters have been made for limited model stud-
being tested, comparison of the model pa-
ies and those methods and techniques are be-
rameters derived using these site-specific
coming more widely applied; however, most
model applications still rely heavily on the data should be compared at least on a spot-
NWS data. check basis, with parameters derived from
b. There are two standard formats of the more conventional observations.
NWS data for use in air quality models. The
short term models use the standard hourly 9.3.3.2 Recommendations: Site-specific Data
weather observations available from the Na- Collection
tional Climatic Data Center (NCDC). These a. The document ‘‘On-Site Meteorological
observations are then ‘‘preprocessed’’ before Program Guidance for Regulatory Modeling
they can be used in the models. ‘‘STAR’’ Applications’’ 66 provides recommendations
summaries are available from NCDC for long
on the collection and use of on-site meteoro-
term model use. These are joint frequency
logical data. Recommendations on charac-
distributions of wind speed, direction and P–
G stability category. They are used as direct teristics, siting, and exposure of meteoro-
input to models such as the long term ver- logical instruments and on data recording,
sion of ISC. 58 processing, completeness requirements, re-
porting, and archiving are also included.
9.3.2.2 Recommendations This publication should be used as a supple-
a. The preferred short term models listed ment to the limited guidance on these sub-
in appendix A all accept as input the NWS jects now found in the ‘‘Ambient Monitoring
meteorological data preprocessed into model Guidelines for Prevention of Significant De-
compatible form. Long-term (monthly sea- terioration’’. 63 Detailed information on qual-
sonal or annual) preferred models use NWS ity assurance is provided in the ‘‘Quality As-
‘‘STAR’’ summaries. Summarized concentra- surance Handbook for Air Pollution Meas-
tion estimates from the short term models urement Systems: Volume IV’’. 67 As a mini-
may also be used to develop long-term aver- mum, site-specific measurements of ambient
ages; however, concentration estimates air temperature, transport wind speed and
based on the two separate input data sets direction, and the parameters to determine
may not necessarily agree. Pasquill-Gifford (P–G) stability categories
b. Although most NWS measurements are should be available in meteorological data
made at a standard height of 10 meters, the sets to be used in modeling. Care should be
actual anemometer height should be used as
taken to ensure that meteorological instru-
input to the preferred model.
ments are located to provide representative
c. National Weather Service wind direc-
tions are reported to the nearest 10 degrees. characterization of pollutant transport be-
A specific set of randomly generated num- tween sources and receptors of interest. The
bers has been developed for use with the pre- Regional Office will determine the appro-
ferred EPA models and should be used to en- priateness of the measurement locations.
sure a lack of bias in wind direction assign- b. All site-specific data should be reduced
ments within the models. to hourly averages. Table 9–3 lists the wind
d. Data from universities, FAA, military related parameters and the averaging time
stations, industry and pollution control requirements.

374
Environmental Protection Agency Pt. 51, App. W
c. Solar Radiation Measurements. Total i. In general, the wind speed used in deter-
solar radiation should be measured with a re- mining plume rise is defined as the wind
liable pyranometer, sited and operated in ac- speed at stack top.
cordance with established on-site meteoro- j. Specifications for wind measuring in-
logical guidance. 66 struments and systems are contained in the
d. Temperature Measurements. Tempera- ‘‘On-Site Meteorological Program Guidance
ture measurements should be made at stand- for Regulatory Modeling Applications’’. 66
ard shelter height (2m) in accordance with k. Stability Categories. The P–G stability
established on-site meteorological guid- categories, as originally defined, couple
ance. 66 near-surface measurements of wind speed
with subjectively determined insolation as-
e. Temperature Difference Measurements.
sessments based on hourly cloud cover and
Temperature difference (ή) measurements ceiling height observations. The wind speed
for use in estimating P–G stability cat- measurements are made at or near 10m. The
egories using the solar radiation/delta-T insolation rate is typically assessed using
(SRDT) methodology (see Stability Cat- observations of cloud cover and ceiling
egories) should be obtained using two height based on criteria outlined by Turn-
matched thermometers or a reliable thermo- er. 50 It is recommended that the P–G stabil-
couple system to achieve adequate accuracy. ity category be estimated using the Turner
f. Siting, probe placement, and operation method with site-specific wind speed meas-
of >T systems should be based on guidance ured at or near 10m and representative cloud
found in Chapter 3 of reference 66, and such cover and ceiling height. Implementation of
guidance should be followed when obtaining the Turner method, as well as considerations
vertical temperature gradient data for use in in determining representativeness of cloud
plume rise estimates or in determining the cover and ceiling height in cases for which
critical dividing streamline height. site-specific cloud observations are unavail-
g. Wind Measurements. For refined model- able, may be found in section 6 of reference
ing applications in simple terrain situations, 66. In the absence of requisite data to imple-
if a source has a stack below 100m, select the ment the Turner method, the SRDT method
stack top height as the wind measurement or wind fluctuation statistics (i.e., the σE and
height for characterization of plume dilution σA methods) may be used.
and transport. For sources with stacks ex- l. The SRDT method, described in section
tending above 100m, a 100m tower is sug- 6.4.4.2 of reference 66, is modified slightly
gested unless the stack top is significantly from that published by Bowen et al. (1983) 136
above 100m (i.e., ≥200m). In cases with stack and has been evaluated with three on-site
tops ≥200m, remote sensing may be a feasible data bases. 137 The two methods of stability
alternative. In some cases, collection of classification which use wind fluctuation
stack top wind speed may be impractical or statistics, the σE and σA methods, are also de-
scribed in detail in section 6.4.4 of reference
incompatible with the input requirements of
66 (note applicable tables in section 6). For
the model to be used. In such cases, the Re-
additional information on the wind fluctua-
gional Office should be consulted to deter-
tion methods, see references 68–72.
mine the appropriate measurement height.
m. Hours in the record having missing data
h. For refined modeling applications in should be treated according to an established
complex terrain, multiple level (typically data substitution protocol and after valid
three or more) measurements of wind speed data retrieval requirements have been met.
and direction, temperature and turbulence Such protocols are usually part of the ap-
(wind fluctuation statistics) are required. proved monitoring program plan. Data sub-
Such measurements should be obtained up to stitution guidance is provided in section 5.3
the representative plume height(s) of inter- of reference 66.
est (i.e., the plume height(s) under those con- n. Meteorological Data Processors. The fol-
ditions important to the determination of lowing meteorological preprocessors are rec-
the design concentration). The representa- ommended by EPA: RAMMET, PCRAMMET,
tive plume height(s) of interest should be de- STAR, PCSTAR, MPRM, 135 and METPRO. 24
termined using an appropriate complex ter- RAMMET is the recommended meteorologi-
rain screening procedure (e.g., CTSCREEN) cal preprocessor for use in applications em-
and should be documented in the monitoring/ ploying hourly NWS data. The RAMMET for-
modeling protocol. The necessary meteoro- mat is the standard data input format used
logical measurements should be obtained in sequential Gaussian models recommended
from an appropriately sited meteorological by EPA. PCRAMMET 138 is the PC equivalent
tower augmented by SODAR if the represent- of the mainframe version (RAMMET). STAR
ative plume height(s) of interest exceed is the recommended preprocessor for use in
100m. The meteorological tower need not ex- applications employing joint frequency dis-
ceed the lesser of the representative plume tributions (wind direction and wind speed by
height of interest (the highest plume height stability class) based on NWS data. PCSTAR
if there is more than one plume height of in- is the PC equivalent of the mainframe ver-
terest) or 100m. sion (STAR). MPRM is the recommended

375
Pt. 51, App. W 40 CFR Ch. I (7–1–97 Edition)
preprocessor for use in applications employ- tion equal to the previous hour’s wind direc-
ing on-site meteorological data. The latest tion. Such cases are then treated in a pre-
version (MPRM 1.3) has been configured to scribed manner when estimating short term
implement the SRDT method for estimating concentrations.
P–G stability categories. MPRM is a general
purpose meteorological data preprocessor 9.3.4.2 Recommendations
which supports regulatory models requiring a. Hourly concentrations calculated with
RAMMET formatted data and STAR Gaussian models using calms should not be
formatted data. In addition to on-site data, considered valid; the wind and concentration
MPRM provides equivalent processing of estimates for these hours should be dis-
NWS data. METPRO is the required meteoro- regarded and considered to be missing. Criti-
logical data preprocessor for use with cal concentrations for 3-, 8-, and 24-hour
CTDMPLUS. All of the above mentioned averages should be calculated by dividing
data preprocessors are available for the sum of the hourly concentration for the
downloading from the SCRAM BBS. 19 period by the number of valid or non-missing
hours. If the total number of valid hours is
TABLE 9–3—AVERAGING TIMES FOR SITE-SPE- less than 18 for 24-hour averages, less than 6
CIFIC WIND AND TURBULENCE MEASUREMENTS for 8-hour averages or less than 3 for 3-hour
averages, the total concentration should be
Averaging divided by 18 for the 24-hour average, 6 for
Parameter time
the 8-hour average and 3 for the 3-hour aver-
Surface wind speed (for use in stability de- 1-hr. age. For annual averages, the sum of all
terminations). valid hourly concentrations is divided by the
Transport direction ...................................... 1-hr. number of non-calm hours during the year. A
Dilution wind speed ..................................... 1-hr. post-processor computer program,
Turbulence measurements (σE and σA) for 1-hr.1 CALMPRO 73 has been prepared following
use in stability determinations. these instructions and has been coded in
1 To minimize meander effects in σ RAM and ISC.
A when wind conditions
are light and/or variable, determine the hourly average σ b. The recommendations in paragraph a of
value from four sequential 15-minute σ’s according to the fol-
lowing formula: this section apply to the use of calms for
short term averages and do not apply to the
σ 15 2 + σ 15 2 + σ 15 2 + σ 15 2 determination of long term averages using
σ 1-hr = ‘‘STAR’’ data summaries. Calms should con-
4 tinue to be included in the preparation of
‘‘STAR’’ summaries. A treatment for calms
9.3.4 Treatment of Calms and very light winds is built into the soft-
ware that produces the ‘‘STAR’’ summaries.
9.3.4.1 Discussion c. Stagnant conditions, including extended
a. Treatment of calm or light and variable periods of calms, often produce high con-
wind poses a special problem in model appli- centrations over wide areas for relatively
cations since Gaussian models assume that long averaging periods. The standard short
concentration is inversely proportional to term Gaussian models are often not applica-
wind speed. Furthermore, concentrations be- ble to such situations. When stagnation con-
come unrealistically large when wind speeds ditions are of concern, other modeling tech-
less than 1 m/s are input to the model. A pro- niques should be considered on a case-by-
cedure has been developed for use with NWS case basis (see also section 8.2.10).
data to prevent the occurrence of overly con- d. When used in Gaussian models, meas-
servative concentration estimates during pe- ured on-site wind speeds of less than 1 m/s
riods of calms. This procedure acknowledges but higher than the response threshold of the
that a Gaussian plume model does not apply instrument should be input as 1 m/s; the cor-
during calm conditions and that our knowl- responding wind direction should also be
edge of plume behavior and wind patterns input. Observations below the response
during these conditions does not, at present, threshold of the instrument are also set to 1
permit the development of a better tech- m/s but the wind direction from the previous
nique. Therefore, the procedure disregards hour is used. If the wind speed or direction
hours which are identified as calm. The hour can not be determined, that hour should be
is treated as missing and a convention for treated as missing and short term averages
handling missing hours is recommended. should then be calculated as described in
b. Preprocessed meteorological data input paragraph a of this section.
to most appendix A EPA models substitute a 10.0 ACCURACY AND UNCERTAINTY OF MODELS
1.00 m/s wind speed and the previous direc-
tion for the calm hour. The new treatment of 10.1 Discussion
calms in those models attempts to identify
a. Increasing reliance has been placed on
the original calm cases by checking for a 1.00
concentration estimates from models as the
m/s wind speed coincident with a wind direc-
primary basis for regulatory decisions con-
cerning source permits and emission control

376
Environmental Protection Agency Pt. 51, App. W
requirements. In many situations, such as model is normally determined by an evalua-
review of a proposed source, no practical al- tion procedure which involves the compari-
ternative exists. Therefore, there is an obvi- son of model concentration estimates with
ous need to know how accurate models really measured air quality data. 78 The statement
are and how any uncertainty in the esti- of accuracy is based on statistical tests or
mates affects regulatory decisions. EPA rec- performance measures such as bias, noise,
ognizes the need for incorporating such in- correlation, etc. 11 However, information that
formation and has sponsored workshops 11 74 allows a distinction between contributions of
on model accuracy, the possible ways to the various elements of inherent and reduc-
quantify accuracy, and on considerations in ible uncertainty is only now beginning to
the incorporation of model accuracy and un- emerge. As a result most discussions of the
certainty in the regulatory process. The Sec- accuracy of models make no quantitative
ond (EPA) Conference on Air Quality Model- distinction between (1) Limitations of the
ing, August 1982,75 was devoted to that sub- model versus (2) limitations of the data base
ject. and of knowledge concerning atmospheric
variability. The reader should be aware that
10.1.1 Overview of Model Uncertainty statements on model accuracy and uncer-
a. Dispersion models generally attempt to tainty may imply the need for improvements
estimate concentrations at specific sites in model performance that even the ‘‘per-
that really represent an ensemble average of fect’’ model could not satisfy.
numerous repetitions of the same event. The
10.1.2 Studies of Model Accuracy
event is characterized by measured or
‘‘known’’ conditions that are input to the a. A number of studies 79 80 have been con-
models, e.g., wind speed, mixed layer height, ducted to examine model accuracy, particu-
surface heat flux, emission characteristics, larly with respect to the reliability of short-
etc. However, in addition to the known con- term concentrations required for ambient
ditions, there are unmeasured or unknown standard and increment evaluations. The re-
variations in the conditions of this event, sults of these studies are not surprising. Ba-
e.g., unresolved details of the atmospheric sically, they confirm what leading atmos-
flow such as the turbulent velocity field. pheric scientists have said for some time: (1)
These unknown conditions may vary among Models are more reliable for estimating
repetitions of the event. As a result, devi- longer time-averaged concentrations than
ations in observed concentrations from their for estimating short-term concentrations at
ensemble average, and from the concentra- specific locations; and (2) the models are rea-
tions estimated by the model, are likely to sonably reliable in estimating the magnitude
occur even though the known conditions are of highest concentrations occurring some-
fixed. Even with a perfect model that pre- time, somewhere within an area. For exam-
dicts the correct ensemble average, there are ple, errors in highest estimated concentra-
likely to be deviations from the observed tions of #10 to 40 percent are found to be typ-
concentrations in individual repetitions of ical, 81 i.e., certainly well within the often
the event, due to variations in the unknown quoted factor-of-two accuracy that has long
conditions. The statistics of these concentra- been recognized for these models. However,
tion residuals are termed ‘‘inherent’’ uncer- estimates of concentrations that occur at a
tainty. Available evidence suggests that this specific time and site, are poorly correlated
source of uncertainty alone may be respon- with actually observed concentrations and
sible for a typical range of variation in con- are much less reliable.
centrations of as much as #50 percent. 76 b. As noted in paragraph a of this section,
b. Moreover, there is ‘‘reducible’’ uncer- poor correlations between paired concentra-
tainty 77 associated with the model and its tions at fixed stations may be due to ‘‘reduc-
input conditions; neither models nor data ible’’ uncertainties in knowledge of the pre-
bases are perfect. Reducible uncertainties cise plume location and to unquantified in-
are caused by: (1) Uncertainties in the input herent uncertainties. For example,
values of the known conditions—emission Pasquill 82 estimates that, apart from data
characteristics and meteorological data; (2) input errors, maximum ground-level con-
errors in the measured concentrations which centrations at a given hour for a point
are used to compute the concentration re- source in flat terrain could be in error by 50
siduals; and (3) inadequate model physics and percent due to these uncertainties. Uncer-
formulation. The ‘‘reducible’’ uncertainties tainty of five to 10 degrees in the measured
can be minimized through better (more accu- wind direction, which transports the plume,
rate and more representative) measurements can result in concentration errors of 20 to 70
and better model physics. percent for a particular time and location,
c. To use the terminology correctly, ref- depending on stability and station location.
erence to model accuracy should be limited Such uncertainties do not indicate that an
to that portion of reducible uncertainty estimated concentration does not occur, only
which deals with the physics and the formu- that the precise time and locations are in
lation of the model. The accuracy of the doubt.

377
Pt. 51, App. W 40 CFR Ch. I (7–1–97 Edition)
10.1.3 Use of Uncertainty in Decision-Making mate’’ until sufficient technical progress has
a. The accuracy of model estimates varies been made to meaningfully implement such
with the model used, the type of application, concepts dealing with uncertainty.
and site-specific characteristics. Thus, it is
10.1.4 Evaluation of Models
desirable to quantify the accuracy or uncer-
tainty associated with concentration esti- a. A number of actions are being taken to
mates used in decision-making. Communica- ensure that the best model is used correctly
tions between modelers and decision-makers for each regulatory application and that a
must be fostered and further developed. Com- model is not arbitrarily imposed. First, the
munications concerning concentration esti- Guideline clearly recommends the most ap-
mates currently exist in most cases, but the propriate model be used in each case. Pre-
communications dealing with the accuracy ferred models, based on a number of factors,
of models and its meaning to the decision-
are identified for many uses. General guid-
maker are limited by the lack of a technical
ance on using alternatives to the preferred
basis for quantifying and directly including
uncertainty in decisions. Procedures for models is also provided. Second, all the mod-
quantifying and interpreting uncertainty in els in eight categories (i.e., rural, urban, in-
the practical application of such concepts dustrial complex, reactive pollutants, mobile
are only beginning to evolve; much study is source, complex terrain, visibility and long
still required.74 75 77 range transport) that are candidates for in-
b. In all applications of models an effort is clusion in the Guideline are being subjected
encouraged to identify the reliability of the to a systematic performance evaluation and
model estimates for that particular area and a peer scientific review. 85 The same data
to determine the magnitude and sources of bases are being used to evaluate all models
error associated with the use of the model. within each of eight categories. Statistical
The analyst is responsible for recognizing performance measures, including measures
and quantifying limitations in the accuracy, of difference (or residuals) such as bias, vari-
precision and sensitivity of the procedure. ance of difference and gross variability of
Information that might be useful to the deci- the difference, and measures of correlation
sion-maker in recognizing the seriousness of such as time, space, and time and space com-
potential air quality violations includes such bined as recommended by the AMS Woods
model accuracy estimates as accuracy of Hole Workshop, 11 are being followed. The re-
peak predictions, bias, noise, correlation,
sults of the scientific review are being incor-
frequency distribution, spatial extent of high
porated in the Guideline and will be the basis
concentration, etc. Both space/time pairing
for future revision.12 13 Third, more specific
of estimates and measurements and unpaired
comparisons are recommended. Emphasis information has been provided for justifying
should be on the highest concentrations and the site specific use of alternative models in
the averaging times of the standards or in- the documents ‘‘Interim Procedures for Eval-
crements of concern. Where possible, con- uating Air Quality Models’’, 15 and the ‘‘Pro-
fidence intervals about the statistical values tocol for Determining the Best Performing
should be provided. However, while such in- Model’’. 17 Together these documents provide
formation can be provided by the modeler to methods that allow a judgment to be made
the decision-maker, it is unclear how this in- as to what models are most appropriate for a
formation should be used to make an air pol- specific application. For the present, per-
lution control decision. Given a range of pos- formance and the theoretical evaluation of
sible outcomes, it is easiest and tends to en- models are being used as an indirect means
sure consistency if the decision-maker con- to quantify one element of uncertainty in air
fines his judgment to use of the ‘‘best esti- pollution regulatory decisions.
mate’’ provided by the modeler (i.e., the de- b. In addition to performance evaluation of
sign concentration estimated by a model rec- models, sensitivity analyses are encouraged
ommended in the Guideline or an alternate since they can provide additional informa-
model of known accuracy). This is an indica-
tion on the effect of inaccuracies in the data
tion of the practical limitations imposed by
bases and on the uncertainty in model esti-
current abilities of the technical commu-
mates. Sensitivity analyses can aid in deter-
nity.
c. To improve the basis for decision-mak- mining the effect of inaccuracies of vari-
ing, EPA has developed and is continuing to ations or uncertainties in the data bases on
study procedures for determining the accu- the range of likely concentrations. Such in-
racy of models, quantifying the uncertainty, formation may be used to determine source
and expressing confidence levels in decisions impact and to evaluate control strategies.
that are made concerning emissions con- Where possible, information from such sen-
trols.83 84 However, work in this area involves sitivity analyses should be made available to
‘‘breaking new ground’’ with slow and spo- the decision-maker with an appropriate in-
radic progress likely. As a result, it may be terpretation of the effect on the critical con-
necessary to continue using the ‘‘best esti- centrations.

378
Environmental Protection Agency Pt. 51, App. W
10.2 Recommendations tifies the relationship to emission limits.
The recommendations in section 11.2 apply
a. No specific guidance on the consider-
ation of model uncertainty in decision-mak- to: (1) revisions of State Implementation
ing is being given at this time. There is in- Plans; (2) the review of new sources and the
complete technical information on measures prevention of significant deterioration
of model uncertainty that are most relevant (PSD); and (3) analyses of the emissions
to the decision-maker. It is not clear how a trades (‘‘bubbles’’).
decisionmaker could use such information,
11.2 Recommendations
particularly given limitations of the Clean
Air Act. As procedures for considering uncer- 11.2.1 Analysis Requirements
tainty develop and become implementable,
this guidance will be changed and expanded. a. Every effort should be made by the Re-
For the present, continued use of the ‘‘best gional Office to meet with all parties in-
estimate’’ is acceptable and is consistent volved in either a SIP revision or a PSD per-
with Clean Air Act requirements. mit application prior to the start of any
work on such a project. During this meeting,
11.0 REGULATORY APPLICATION OF MODELS a protocol should be established between the
11.1 Discussion preparing and reviewing parties to define the
procedures to be followed, the data to be col-
a. Procedures with respect to the review lected, the model to be used, and the analy-
and analysis of air quality modeling and sis of the source and concentration data. An
data analyses in support of SIP revisions, example of requirements for such an effort is
PSD permitting or other regulatory require- contained in the Air Quality Analysis Check-
ments need a certain amount of standardiza- list included here as appendix C. This check-
tion to ensure consistency in the depth and list suggests the level of detail required to
comprehensiveness of both the review and assess the air quality resulting from the pro-
the analysis itself. This section recommends posed action. Special cases may require addi-
procedures that permit some degree of stand- tional data collection or analysis and this
ardization while at the same time allowing should be determined and agreed upon at
the flexibility needed to assure the tech- this preapplication meeting. The protocol
nically best analysis for each regulatory ap- should be written and agreed upon by the
plication. parties concerned, although a formal legal
b. Dispersion model estimates, especially document is not intended. Changes in such a
with the support of measured air quality protocol are often required as the data col-
data, are the preferred basis for air quality lection and analysis progresses. However, the
demonstrations. Nevertheless, there are in- protocol establishes a common understand-
stances where the performance of rec-
ing of the requirements.
ommended dispersion modeling techniques,
b. An air quality analysis should begin
by comparison with observed air quality
with a screening model to determine the po-
data, may be shown to be less than accept-
able. Also, there may be no recommended tential of the proposed source or control
modeling procedure suitable for the situa- strategy to violate the PSD increment or
tion. In these instances, emission limitations NAAQS. It is recommended that the screen-
may be established solely on the basis of ob- ing techniques found in ‘‘Screening Proce-
served air quality data as would be applied dures for Estimating the Air Quality Impact
to a modeling analysis. The same care should of Stationary Sources’’ 18 be used for point
be given to the analyses of the air quality source analyses. Screening procedures for
data as would be applied to a modeling anal- area source analysis are discussed in ‘‘Apply-
ysis. ing Atmospheric Simulation Models to Air
c. The current NAAQS for SO2 and CO are Quality Maintenance Areas’’. 87 For mobile
both stated in terms of a concentration not source impact assessments the ‘‘Guideline
to be exceeded more than once a year. There for Modeling Carbon Monoxide from Road-
is only an annual standard for NO2 and a way Intersections’’ 34 is available.
quarterly standard for Pb. The PM–10 and c. If the concentration estimates from
ozone standards permit the exceedance of a screening techniques indicate that the PSD
concentration on an average of not more increment or NAAQS may be approached or
than once a year; the convention is to aver- exceeded, then a more refined modeling anal-
age over a 3-year period.5 86 103 This rep- ysis is appropriate and the model user should
resents a change from a deterministic to a select a model according to recommenda-
more statistical form of the standard and tions in sections 4.0–8.0. In some instances,
permits some consideration to be given to no refined technique may be specified in this
unusual circumstances. The NAAQS are sub- guide for the situation. The model user is
jected to extensive review and possible revi- then encouraged to submit a model devel-
sion every 5 years. oped specifically for the case at hand. If that
d. This section discusses general require- is not possible, a screening technique may
ments for concentration estimates and iden- supply the needed results.

379
Pt. 51, App. W 40 CFR Ch. I (7–1–97 Edition)
d. Regional Offices should require permit vi. Can it be demonstrated through the
applicants to incorporate the pollutant con- comparison of monitored data with model re-
tributions of all sources into their analysis. sults that available models are not applica-
Where necessary this may include emissions ble?
associated with growth in the area of impact c. The number of monitors required is a
of the new or modified source’s impact. PSD function of the problem being considered.
air quality assessments should consider the The source configuration, terrain configura-
amount of the allowable air quality incre- tion, and meteorological variations all have
ment that has already been granted to any an impact on number and placement of mon-
other sources. Therefore, the most recent itors. Decisions can only be made on a case-
source applicant should model the existing by-case basis. The Interim Procedures for
or permitted sources in addition to the one Evaluating Air Quality Models 15 should be
currently under consideration. This would used in establishing criteria for demonstrat-
permit the use of newly acquired data or im- ing that a model is not applicable.
proved modeling techniques if such have be- d. Sources should obtain approval from the
come available since the last source was per- Regional Office or reviewing authority for
mitted. When remodeling, the worst case the monitoring network prior to the start of
used in the previous modeling analysis monitoring. A monitoring protocol agreed to
should be one set of conditions modeled in by all concerned parties is highly desirable.
the new analysis. All sources should be mod- The design of the network, the number, type
eled for each set of meteorological condi- and location of the monitors, the sampling
tions selected and for all receptor sites used period, averaging time as well as the need for
in the previous applications as well as new meteorological monitoring or the use of mo-
sites specific to the new source. bile sampling or plume tracking techniques,
should all be specified in the protocol and
11.2.2 Use of Measured Data in Lieu of Model agreed upon prior to start-up of the network.
Estimates
11.2.3 Emission Limits
a. Modeling is the preferred method for de-
termining emission limitations for both new 11.2.3.1 Design Concentrations
and existing sources. When a preferred model a. Emission limits should be based on con-
is available, model results alone (including centration estimates for the averaging time
background) are sufficient. Monitoring will that results in the most stringent control re-
normally not be accepted as the sole basis quirements. The concentration used in speci-
for emission limitation determination in flat fying emission limits is called the design
terrain areas. In some instances when the value or design concentration and is a sum of
modeling technique available is only a the concentration contributed by the source
screening technique, the addition of air qual- and the background concentration.
ity data to the analysis may lend credence to b. To determine the averaging time for the
model results. design value, the most restrictive National
b. There are circumstances where there is Ambient Air Quality Standard (NAAQS)
no applicable model, and measured data may should be identified by calculating, for each
need to be used. Examples of such situations averaging time, the ratio of the applicable
are: (1) complex terrain locations; (2) land/ NAAQS (S)¥ background (B) to the pre-
water interface areas; and (3) urban locations dicted concentration (P) (i.e., (S¥B)/P). The
with a large fraction of particulate emis- averaging time with the lowest ratio identi-
sions from nontraditional sources. However, fies the most restrictive standard. If the an-
only in the case of an existing source should nual average is the most restrictive, the
monitoring data alone be a basis for emis- highest estimated annual average concentra-
sion limits. In addition, the following items tion from one or a number of years of data is
should be considered prior to the acceptance the design value. When short term standards
of the measured data: are most restrictive, it may be necessary to
i. Does a monitoring network exist for the consider a broader range of concentrations
pollutants and averaging times of concern? than the highest value. For example, for pol-
ii. Has the monitoring network been de- lutants such as SO2, the highest, second-
signed to locate points of maximum con- highest concentration is the design value.
centration? For pollutants with statistically based
iii. Do the monitoring network and the NAAQS, the design value is found by deter-
data reduction and storage procedures meet mining the more restrictive of: (1) the short-
EPA monitoring and quality assurance re- term concentration that is not expected to
quirements? be exceeded more than once per year over
iv. Do the data set and the analysis allow the period specified in the standard, or (2)
impact of the most important individual the long-term concentration that is not ex-
sources to be identified if more than one pected to exceed the long-term NAAQS. De-
source or emission point is involved? termination of design values for PM–10 is
v. Is at least one full year of valid ambient presented in more detail in the ‘‘PM–10 SIP
data available? Development Guideline’’. 108

380
Environmental Protection Agency Pt. 51, App. W
c. When the highest, second-highest con- a model should be less than or equal to the
centration is used in assessing potential vio- permitted increment. The modeled annual
lations of a short term NAAQS, criteria that averages should not exceed the increment.
are identified in ‘‘Guideline for Interpreta- b. Screening techniques defined in sections
tion of Air Quality Standards’’88 should be 4.0 and 5.0 can sometimes be used to estimate
followed. This guidance specifies that a vio- short term incremental concentrations for
lation of a short term standard occurs at a the first new source that triggers the base-
site when the standard is exceeded a second line in a given area. However, when multiple
time. Thus, emission limits that protect increment-consuming sources are involved in
standards for averaging times of 24 hours or the calculation, the use of a refined model
less are appropriately based on the highest, with at least 1 year of on-site or 5 years of
second-highest estimated concentration plus off-site NWS data is normally required. In
a background concentration which can rea- such cases, sequential modeling must dem-
sonably be assumed to occur with the con- onstrate that the allowable increments are
centration. not exceeded temporally and spatially, i.e.,
for all receptors for each time period
11.2.3.2 NAAQS Analyses for New or Modified throughout the year(s) (time period means
Sources the appropriate PSD averaging time, e.g., 3-
a. For new or modified sources predicted to hour, 24-hour, etc.).
have a significant ambient impact 63 and to c. The PSD regulations require an esti-
mation of the SO2, particulate matter, and
be located in areas designated attainment or
NO2 impact on any Class I area. Normally,
unclassifiable for the SO2, Pb, NO2, or CO
Gaussian models should not be applied at
NAAQS, the demonstration as to whether
distances greater than can be accommodated
the source will cause or contribute to an air
by the steady state assumptions inherent in
quality violation should be based on: (1) the
such models. The maximum distance for re-
highest estimated annual average concentra-
fined Gaussian model application for regu-
tion determined from annual averages of in-
latory purposes is generally considered to be
dividual years; or (2) the highest, second-
50km. Beyond the 50km range, screening
highest estimated concentration for averag- techniques may be used to determine if more
ing times of 24-hours or less; and (3) the sig- refined modeling is needed. If refined models
nificance of the spatial and temporal con- are needed, long range transport models
tribution to any modeled violation. For Pb, should be considered in accordance with sec-
the highest estimated concentration based tion 7.2.6. As previously noted in sections 3.0
on an individual calendar quarter averaging and 7.0, the need to involve the Federal Land
period should be used. Background con- Manager in decisions on potential air quality
centrations should be added to the estimated impacts, particularly in relation to PSD
impact of the source. The most restrictive Class I areas, cannot be overemphasized.
standard should be used in all cases to assess
the threat of an air quality violation. For 11.2.3.4 Emissions Trading Policy (Bubbles)
new or modified sources predicted to have a
a. EPA’s final Emissions Trading Policy,
significant ambient impact 63 in areas des-
commonly referred to as the ‘‘bubble pol-
ignated attainment or unclassifiable for the
icy,’’ was published in the FEDERAL REGISTER
PM–10 NAAQS, the demonstration of wheth-
in 1986.89 Principles contained in the policy
er or not the source will cause or contribute
should be used to evaluate ambient impacts
to an air quality violation should be based
of emission trading activities.
on sufficient data to show whether: (1) the
b. Emission increases and decreases within
projected 24-hour average concentrations the bubble should result in ambient air qual-
will exceed the 24-hour NAAQS more than ity equivalence. Two levels of analysis are
once per year, on average; (2) the expected defined for establishing this equivalence. In
(i.e., average) annual mean concentration a Level I analysis the source configuration
will exceed the annual NAAQS; and (3) the and setting must meet certain limitations
source contributes significantly, in a tem- (defined in the policy) that ensure ambient
poral and spatial sense, to any modeled vio- equivalence; no modeling is required. In a
lation. Level II analysis a modeling demonstration
11.2.3.3 PSD Air Quality Increments and of ambient equivalence is required but only
Impacts the sources involved in the emissions trade
are modeled. The resulting ambient esti-
a. The allowable PSD increments for cri- mates of net increases/decreases are com-
teria pollutants are established by regula- pared to a set of significance levels to deter-
tion and cited in § 51.166. These maximum al- mine if the bubble can be approved. A Level
lowable increases in pollutant concentra- II analysis requires the use of a refined
tions may be exceeded once per year at each model and the most recent readily available
site, except for the annual increment that full year of representative meteorological
may not be exceeded. The highest, second- data. Sequential modeling must demonstrate
highest increase in estimated concentrations that the significance levels are met tem-
for the short term averages as determined by porally and spatially, i.e., for all receptors

381
Pt. 51, App. W 40 CFR Ch. I (7–1–97 Edition)
for each time period throughout the year 4. Environmental Protection Agency, 1981.
(time period means the appropriate NAAQS Guideline for Fluid Modeling of Atmospheric
averaging time, e.g., 3-hour, 24-hour, etc.). Diffusion. EPA Publication No. EPA–600/8–
c. For those bubbles that cannot meet the 81–009. U.S. Environmental Protection Agen-
Level I or Level II requirements, the Emis- cy, Research Triangle Park, NC. (NTIS No.
sions Trading Policy allows for a Level III PB 81–201410)
analysis. A Level III analysis, from a model- 5. Code of Federal Regulations (Title 40,
ing standpoint, is generally equivalent to the part 50): Protection of the Environment; Na-
requirements for a standard SIP revision tional Primary and Secondary Ambient Air
where all sources (and background) are con- Quality Standards.
sidered and the estimates are compared to 6. Environmental Protection Agency, 1988.
the NAAQS as in section 11.2.3.2. Model Clearinghouse: Operational Plan (Re-
d. The Emissions Trading Policy allows vised). Staff Report. U.S. Environmental
States to adopt generic regulations for proc- Protection Agency, Research Triangle Park,
essing bubbles. The modeling procedures rec- NC. (Docket No. A–88–04, II–J–1)
ommended in the Guideline apply to such ge- 7. Environmental Protection Agency, 1980.
neric regulations. However, an added re- Guidelines on Air Quality Models. FEDERAL
quirement is that the modeling procedures REGISTER, 45: 20157–20158.
8. Londergan, R.J., D.H. Minott, D.J.
contained in any generic regulation must be
Wackter, T. Kincaid and D. Bonitata, 1982.
replicable such that there is no doubt as to
Evaluation of Rural Air Quality Simulation
how each individual bubble will be modeled.
Models. EPA Publication No. EPA–450/4–82–
In general this means that the models, the
020. U.S. Environmental Protection Agency,
data bases and the procedures for applying
Research Triangle Park, NC. (NTIS No. PB
the model must be defined in the regulation.
83–182758)
The consequences of the replicability re-
9. Londergan, R.J., D.H. Minott, D.J.
quirement are that bubbles for sources lo- Wackter and R.R. Fizz, 1983. Evaluation of
cated in complex terrain and certain indus- Urban Air Quality Simulation Models. EPA
trial sources where judgments must be made Publication No. EPA–450/4–83–020. U.S. Envi-
on source characterization cannot be han- ronmental Protection Agency, Research Tri-
dled generically. angle Park, NC. (NTIS No. PB 84–241173).
10. Londergan, R.J. and D.J. Wackter, 1984.
12.0 REFERENCES g h
Evaluation of Complex Terrain Air Quality
1. Code of Federal Regulations (Title 40, Simulation Models. EPA Publication No.
part 51): Protection of the Environment; Re- EPA–450/4–84–017. U.S. Environmental Pro-
quirements for Preparation, Adoption, and tection Agency, Research Triangle Park, NC.
Submittal of Implementation Plans. (NTIS No. PB 85–119485).
2. Environmental Protection Agency, 1977. 11. Fox, D.G., 1981. Judging Air Quality
Guidelines for the Regional Evaluation of Model Performance. Bulletin of the Amer-
State and Local New Source Review Pro- ican Meteorological Society, 62(5): 599–609.
gram. EPA Publication No. EPA–450/2–77–027. 12. American Meteorological Society, 1983.
U.S. Environmental Protection Agency, Re- Synthesis of the Rural Model Reviews. EPA
search Triangle Park, NC. (NTIS No. PB– Publication No. EPA–600/3–83–108. U.S. Envi-
275053) ronmental Protection Agency, Research Tri-
3. Environmental Protection Agency, 1980. angle Park, NC. (NTIS No. PB 84–121037)
Prevention of Significant Deterioration 13. American Meteorological Society, 1984.
Workshop Manual. EPA Publication No. Review of the Attributes and Performance of
EPA–450/2–80–081. U.S. Environmental Pro- Six Urban Diffusion Models. EPA Publica-
tection Agency, Research Triangle Park, NC. tion No. EPA–600/S3–84–089. U.S. Environ-
(NTIS No. PB 81–136459) mental Protection Agency, Research Tri-
angle Park, NC. (NTIS No.PB 84–236850)
14. White, F.D.(Ed.), J.K.S. Ching, R.L.
g Documents not available in the open lit- Dennis and W.H. Snyder, 1985. Summary of
erature or from the National Technical In- Complex Terrain Model Evaluation. EPA
formation Service (NTIS) have been placed Publication No. EPA–600/3–85–060. U.S. Envi-
in Docket No. A–80–46 or A–88–04. Item Num- ronmental Protection Agency, Research Tri-
bers for documents placed in the Docket are angle Park, NC. (NTIS No. PB 85–236891)
shown at the end of the reference. 15. Environmental Protection Agency, 1984.
h Some EPA references, e.g., model user’s Interim Procedures for Evaluating Air Qual-
guides, etc., are periodically revised. Users ity Models (Revised). EPA Publication No.
are referred to the SCRAM BBS19 to EPA–450/4–84–023. U.S. Environmental Pro-
download updates or addenda; see section A.0 tection Agency, Research Triangle Park, NC.
of this appendix. (NTIS No. PB 85–106060)
i The documents listed here are major 16. Environmental Protection Agency, 1985.
sources of supplemental information on the Interim Procedures for Evaluating Air Qual-
theory and application of mathematical air ity Models: Experience with Implementa-
quality models. tion. EPA Publication No. EPA–450/4–85–006.

382
Environmental Protection Agency Pt. 51, App. W
U.S. Environmental Protection Agency, Re- LONGZ Computer Programs, Volumes I and
search Triangle Park, NC. (NTIS No. PB 85– II. EPA Publication No. EPA–903/9–82–004a
242477) and b. U.S. Environmental Protection Agen-
17. Environmental Protection Agency, 1992. cy Region III, Philadelphia, PA. (NTIS Nos.
Protocol for Determining the Best Perform- PB 83–146092 and PB 83–146100)
ing Model. EPA Publication No. EPA–454/R– 28. Environmental Protection Agency, 1990.
92–025. U.S. Environmental Protection Agen- User’s Guide for the Urban Airshed Model,
cy, Research Triangle Park, NC. Volume I–VIII. EPA Publication Nos. EPA–
18. Environmental Protection Agency, 1992. 450/4–90–007a–c, d(R), e–g; EPA–454/B–93–004,
Screening Procedures for Estimating the Air respectively. U.S. Environmental Protection
Quality Impact of Stationary Sources, Re- Agency, Research Triangle Park, NC. (NTIS
vised. EPA Publication No. EPA–454/R–92– Nos. PB 91–131227, PB 91–131235, PB 91–131243,
019. U.S. Environmental Protection Agency, PB 93–122380, PB 91–131268, PB 92–145382, and
Research Triangle Park, NC.
PB 92–224849, respectively, for Vols. I–VII)
19. Environmental Protection Agency, 1989.
Support Center for Regulatory Air Models 29. Environmental Protection Agency, 1991.
Bulletin Board System (SCRAM BBS). Guideline for Regulatory Application of the
Source Receptor Analysis Branch, Research Urban Airshed Model. EPA Publication No.
Triangle Park, NC. (Docket Nos. A–88–04, II– EPA–450/4–91–013 U.S. Environmental Protec-
J–4a and b) tion Agency, Research Triangle Park, NC.
20. Environmental Protection Agency, 1995. (NTIS No. PB 92–108760/AS)
SCREEN3 User’s Guide. EPA Publication No. 30. Environmental Protection Agency, 1989.
EPA–454/B–95–004. U.S. Environmental Pro- Procedures for Applying City-Specific EKMA
tection Agency, Research Triangle Park, NC. (Empirical Kinetic Modeling Approach). EPA
(NTIS No. PB 95–222766) Publication No. EPA–450/4–89–012. U.S. Envi-
21. Environmental Protection Agency, 1987. ronmental Protection Agency, Research Tri-
EPA Complex Terrain Model Development: angle Park, NC. (NTIS No. PB 90–256777)
Final Report. EPA Publication No. EPA–600/ 31. Environmental Protection Agency, 1989.
3–88–006. U.S. Environmental Protection User’s Manual for OZIPM–4 (Ozone Isopleth
Agency, Research Triangle Park, NC. (NTIS Plotting with Optional Mechanisms), Vol-
No. PB 88–162110) umes 1 and 2. EPA Publication Nos. EPA–450/
22. Perry, S.G., D.J. Burns, L.H. Adams, 4–89–009a and b. U.S. Environmental Protec-
R.J. Paine, M.G. Dennis, M.T. Mills, D.J. tion Agency, Research Triangle Park, NC.
Strimaitis, R.J. Yamartino and E.M. Insley, (NTIS Nos. PB 90–261488 and PB 90–255399)
1989. User’s Guide to the Complex Terrain 32. Environmental Protection Agency, 1988.
Dispersion Model Plus Algorithms for Unsta- A PC Based System for Generating EKMA
ble Situations (CTDMPLUS) Volume 1; Input Files. EPA Publication No. EPA–450/4–
Model Description and User Instructions. 88–016. U.S. Environmental Protection Agen-
EPA Publication No. EPA–600/8–89–041. U.S. cy, Research Triangle Park, NC. (NTIS No.
Environmental Protection Agency, Research PB 90–255407)
Triangle Park, NC. (NTIS No. PB 89–181424)
33. Meyer, Jr., E.L. and K.A. Baugues, 1987.
23. Mills, M.T., R.J. Paine, E.A. Insley and
Consideration of Transported Ozone and Pre-
B.A. Egan, 1987. The Complex Terrain Disper-
cursors and Their Use in EKMA. EPA Publi-
sion Model Terrain Preprocessor System—
User’s Guide and Program Description. EPA cation No. EPA–450/4–89–010. U.S. Environ-
Publication No. EPA–600/8–88–003. U.S. Envi- mental Protection Agency, Research Tri-
ronmental Protection Agency, Research Tri- angle Park, NC. (NTIS No. PB 90–255415)
angle Park, NC. (NTIS No. PB 88–162094) 34. Environmental Protection Agency, 1992.
24. Paine, R.J., 1987. User’s Guide to the Guideline for Modeling Carbon Monoxide
CTDM Meteorological Preprocessor Pro- from Roadway Intersections. EPA Publica-
gram. EPA Publication No. EPA–600/8–88–004. tion No. EPA–454/R–92–005. U.S. Environ-
U.S. Environmental Protection Agency, Re- mental Protection Agency, Research Tri-
search Triangle Park, NC. (NTIS No. PB 88– angle Park, NC. (NTIS No. PB 93–210391)
162102) 35. Environmental Protection Agency, 1992.
25. Perry, S.G., D.J. Burns and A.J. User’s Guide for CAL3QHC Version 2: A Mod-
Cimorelli, 1990. User’s Guide to CTDMPLUS: eling Methodology for Predicting Pollutant
Volume 2. The Screening Mode (CTSCREEN). Concentrations near Roadway Intersections.
EPA Publication No. EPA–600/8–90–087. U.S. EPA Publication No. EPA–454/R–92–006. U.S.
Environmental Protection Agency, Research Environmental Protection Agency, Research
Triangle Park, NC. (NTIS No. PB 91–136564) Triangle Park, NC. (NTIS No. PB 93–210250)
26. Burt, E.W., 1977. Valley Model User’s 36. Chu, S.H. and E.L. Meyer, 1991. Use of
Guide. EPA Publication No. EPA–450/2–77– Ambient Ratios to Estimate Impact of NOX
018. U.S. Environmental Protection Agency, Sources on Annual NO2 Concentrations. Pro-
Research Triangle Park, NC. (NTIS No. PB– ceedings, 84th Annual Meeting & Exhibition
274054) of the Air & Waste Management Association,
27. Bjorklund, J.R. and J.F. Bowers, 1982. Vancouver, B.C.; 16–21 June 1991. (16 pp.)
User’s Instructions for the SHORTZ and (Docket No. A–92–65, II-A–9)

383
Pt. 51, App. W 40 CFR Ch. I (7–1–97 Edition)
37. U.S. Department of Housing and Urban Environmental Protection Agency, Research
Development, 1980. Air Quality Consider- Triangle Park, NC. (NTIS No. PB 83–164723)
ations in Residential Planning. U.S. Super- 45. White, W.H., C. Seigneur, D.W. Heinold,
intendent of Documents, Washington, DC. M.W. Eltgroth, L.W. Richards, P.T. Roberts,
(GPO Order Nos. 023–000–00577–8, 023–000– P.S. Bhardwaja, W.D. Conner and W.E. Wil-
00576–0, 023–000–00575–1) son, Jr., 1985. Predicting the Visibility of
38. Environmental Protection Agency, 1981. Chimney Plumes: An Intercomparison of
Receptor Model Technical Series. Volume I: Four Models with Observations at a Well-
Overview of Receptor Model Application to Controlled Power Plant. Atmospheric Envi-
Particulate Source Apportionment. EPA ronment, 19(3): 515–528.
Publication No. EPA–450/4–81–016a (NTIS No. 46. Environmental Protection Agency, 1985.
PB 82–139429); Volume II: Chemical Mass Bal- Guideline for Determination of Good Engi-
ance. EPA Publication No. EPA–450/4–81–016b neering Practice Stack Height (Technical
(NTIS No. PB 82–187345); Volume III (Re- Support Document for the Stack Height Reg-
vised): CMB User’s Manual (Version 7.0). ulations), Revised. EPA Publication No.
EPA Publication No. EPA–450/4–90–004 (NTIS EPA–450/4–80–023R. U.S. Environmental Pro-
No. PB 90–185067); Volume IV: Technical Con- tection Agency, Research Triangle Park, NC.
siderations In Source Apportionment By (NTIS No. PB 85–225241)
Particle Identification. EPA Publication No. 47. Environmental Protection Agency, 1981.
EPA–450/4–83–018 (NTIS No. PB 84–103340); Guideline for Use of Fluid Modeling to De-
Volume V: Source Apportionment Tech- termine Good Engineering Practice (GEP)
niques and Considerations in Combining Stack Height. EPA Publication No. EPA–450/
their Use. EPA Publication No. EPA–450/4– 4–81–003. U.S. Environmental Protection
84–020 (NTIS No. PB 85–111524); Volume VI: A Agency, Research Triangle Park, NC. (NTIS
Guide To The Use of Factor Analysis and No. PB 82–145327)
Multiple Regression (FA/MR) Techniques in 48. Lawson, Jr., R.E. and W.H. Snyder, 1983.
Source Apportionment. EPA Publication No. Determination of Good Engineering Practice
Stack Height: A Demonstration Study for a
EPA–450/4–85–007 (NTIS No. PB 86–107638).
Power Plant. EPA Publication No. EPA–600/
U.S. Environmental Protection Agency, Re-
3–83–024. U.S. Environmental Protection
search Triangle Park, NC.
Agency, Research Triangle Park, NC. (NTIS
39. Pace, T.G., 1982. The Role of Receptor
No. PB 83–207407)
Models for Revised Particulate Matter 49. Snyder, W.H. and R.E. Lawson, Jr., 1985.
Standards. A Specialty Conference on: Re- Fluid Modeling Demonstration of Good Engi-
ceptor Models Applied to Contemporary Pol- neering-Practice Stack Height in Complex
lution Problems. Air Pollution Control Asso- Terrain. EPA Publication No. EPA–600/3–85–
ciation, Pittsburgh, PA; pp. 18–28. (Docket 022. U.S. Environmental Protection Agency,
No. A–80–46, II–P–10) Research Triangle Park, NC. (NTIS No. PB
40. Environmental Protection Agency, 1978. 85–203107)
Supplementary Guidelines for Lead Imple- 50. Turner, D.B., 1969. Workbook of Atmos-
mentation Plans. EPA Publication No. EPA– pheric Dispersion Estimates. PHS Publica-
450/2–78–038. U.S. Environmental Protection tion No. 999–AP–26. U.S. Department of
Agency, Research Triangle Park, NC. (NTIS Health, Education and Welfare, Public
No. PB 82–232737) Health Service, Cincinnati, OH. (NTIS No.
41. Environmental Protection Agency, 1983. PB–191482)
Updated Information on Approval and Pro- 51. McElroy, J.L. and F. Pooler, Jr., 1968.
mulgation of Lead Implementation Plans St. Louis Dispersion Study, Volume II—
(DRAFT). U.S. Environmental Protection Analysis. National Air Pollution Control Ad-
Agency, Research Triangle Park, NC. (Dock- ministration Publication No. AP–53, U.S. De-
et No. A–80–46, II-B–38) partment of Health, Education and Welfare,
42. Environmental Protection Agency, 1979. Public Health Service, Arlington, VA. (NTIS
Protecting Visibility: An EPA Report to No. PB–190255)
Congress. EPA Publication No. EPA–450/5–79– 52. Irwin, J.S., 1983. Estimating Plume Dis-
008. U.S. Environmental Protection Agency, persion—A Comparison of Several Sigma
Research Triangle Park, NC. (NTIS No. PB Schemes. Journal of Climate and Applied
80–220320) Meteorology, 22: 92–114.
43. Environmental Protection Agency, 1992. 53. Briggs, G.A. and F.S. Binkowski, 1985.
Workbook for Plume Visual Impact Screen- Research on Diffusion in Atmospheric
ing and Analysis (Revised). EPA Publication Boundary Layers: A Position Paper on Sta-
No. EPA–454/R–92–023. U.S. Environmental tus and Needs. EPA Publication No. EPA–600/
Protection Agency, Research Triangle Park, 3–25–072. U.S. Environmental Protection
NC. Agency, Research Triangle Park, NC. (NTIS
44. Seigneur C., A.B. Hudischewskyj and No. PB 86–122587)
R.W. Bergstrom, 1982. Evaluation of the EPA 54. Pasquill, F., 1976. Atmospheric Disper-
PLUVUE Model and the ERT Visibility sion Parameters in Gaussian Plume Model-
Model Based on the 1979 VISTTA Data Base. ing, Part II. Possible Requirements for
EPA Publication No. EPA–450/4–82–008. U.S. Change in the Turner Workbook Values.

384
Environmental Protection Agency Pt. 51, App. W
EPA Publication No. EPA–600/4–76–030b. U.S. Publication No. EPA–450/4–87–013. U.S. Envi-
Environmental Protection Agency, Research ronmental Protection Agency, Research Tri-
Triangle Park, NC. (NTIS No. PB–258036/3BA) angle Park, NC. (NTIS No. PB 87–227542)
55. Turner, D.B., 1964. A Diffusion Model for 67. Environmental Protection Agency, 1989.
an Urban Area. Journal of Applied Meteor- Quality Assurance for Air Pollution Meas-
ology, 3(1): 83–91. urement Systems, Volume IV—Meteorologi-
56. Briggs, G.A., 1975. Plume Rise Pre- cal Measurements. EPA Publication No.
dictions. Chapter 3 in Lectures on Air Pollu- EPA–600/4–82–060R. U.S. Environmental Pro-
tion and Environmental Impact Analyses. tection Agency, Research Triangle Park, NC.
American Meteorological Society, Boston, 68. Irwin, J.S., 1980. Dispersion Estimate
MA; pp. 59–111. Suggestion #8: Estimation of Pasquill Stabil-
57. Hanna, S.R., G.A. Briggs and R.P. ity Categories. U.S. Environmental Protec-
Hosker, Jr., 1982. Plume Rise. Chapter 2 in tion Agency, Research Triangle Park, NC
Handbook on Atmospheric Diffusion. Tech- (Docket No. A–80–46, II–B–10)
nical Information Center, U.S. Department 69. Mitchell, Jr., A.E. and K.O. Timbre,
of Energy, Washington, DC; pp. 11–24. DOE/ 1979. Atmospheric Stability Class from Hori-
TIC–11223 (DE 82002045) zontal Wind Fluctuation. Presented at 72nd
58. Environmental Protection Agency, 1995. Annual Meeting of Air Pollution Control As-
User’s Guide for the Industrial Source Com- sociation, Cincinnati, OH; June 24–29, 1979.
plex (ISC3) Dispersion Models, Volumes 1 and (Docket No. A–80–46, II–P–9)
2. EPA Publication Nos. EPA–454/B–95–003a &
70. Nuclear Regulatory Commission, 1972.
b. U.S. Environmental Protection Agency,
Meteorological Programs in Support of Nu-
Research Triangle Park, NC. (NTIS Nos. PB
clear Power Plants. Regulatory Guide 1.23
95–222741 and PB 95–222758, respectively)
(DRAFT). U.S. Nuclear Regulatory Commis-
59. Irwin, J.S., 1978. Proposed Criteria for
sion, Office of Standards Development,
Selection of Urban Versus Rural Dispersion
Washington, DC. (Docket No. A–80–46, II–P–
Coefficients. (Draft Staff Report). Meteor-
11)
ology and Assessment Division, U.S. Envi-
ronmental Protection Agency, Research Tri- 71. Smedman-Hogstrom, A. and V.
angle Park, NC. (Docket No. A–80–46, II–B–8) Hogstrom, 1978. A Practical Method for De-
60. Auer, Jr., A.H., 1978. Correlation of termining Wind Frequency Distributions for
Land Use and Cover with Meteorological the Lowest 200m from Routine Meteorologi-
Anomalies. Journal of Applied Meteorology, cal Data. Journal of Applied Meteorology,
17(5): 636–643. 17(7): 942–954.
61. Brier, G.W., 1973. Validity of the Air 72. Smith, T.B. and S.M. Howard, 1972.
Quality Display Model Calibration Proce- Methodology for Treating Diffusivity. MRI
dure. EPA Publication No. EPA–R4–73–017. 72 FR–1030. Meteorology Research, Inc., Alta-
U.S. Environmental Protection Agency, Re- dena, CA. (Docket No. A–80–46, II–P–8)
search Triangle Park, NC. (NTIS No. PB– 73. Environmental Protection Agency, 1984.
218716) Calms Processor (CALMPRO) User’s Guide.
62. Environmental Protection Agency, 1985 EPA Publication No. EPA–901/9–84–001. U.S.
and ff. Compilation of Air Pollutant Emis- Environmental Protection Agency, Region I,
sion Factors, Volume I: Stationary Point Boston, MA. (NTIS No. PB 84–229467)
and Area Sources (Fourth Edition; GPO 74. Burton, C.S., 1981. The Role of Atmos-
Stock No. 055–000–00251–7), and Supplements; pheric Models in Regulatory Decision-Mak-
Volume II: Mobile Sources (NTIS PB 87– ing: Summary Report. Systems Applications,
205266) and Supplement(s). EPA Publication Inc., San Rafael, CA. Prepared under con-
No. AP–42. U.S. Environmental Protection tract No. 68–01–5845 for U.S. Environmental
Agency, Research Triangle Park, NC. Protection Agency, Research Triangle Park,
63. Environmental Protection Agency, 1987. NC. (Docket No. A–80–46, II–M–6)
Ambient Air Monitoring Guidelines for Pre- 75. Environmental Protection Agency, 1981.
vention of Significant Deterioration (PSD). Proceedings of the Second Conference on Air
EPA Publication No. EPA–450/4–87–007. U.S. Quality Modeling, Washington, DC. U.S. En-
Environmental Protection Agency, Research vironmental Protection Agency, Research
Triangle Park, NC. (NTIS No. PB 90–168030) Triangle Park, NC. (Docket No. A–80–46, II–
64. Landsberg, H.E. and W.C. Jacobs, 1951. M–16)
Compendium of Meteorology. American Me- 76. Hanna, S.R., 1982. Natural Variability of
teorological Society, Boston, MA; pp. 976–992. Observed Hourly SO2 and CO Concentrations
65. Burton, C.S., T.E. Stoeckenius and J.P. in St. Louis. Atmospheric Environment,
Nordin, 1983. The Temporal Representative- 16(6): 1435–1440.
ness of Short-Term Meteorological Data 77. Fox, D.G., 1983. Uncertainty in Air
Sets: Implications for Air Quality Impact Quality Modeling. Bulletin of the American
Assessments. Systems Applications, Inc., Meteorological Society, 65(1): 27–36.
San Rafael, CA. (Docket No. A–80–46, II–G–11) 78. Bowne, N.E., 1981. Validation and Per-
66. Environmental Protection Agency, 1987. formance Criteria for Air Quality Models.
On-Site Meteorological Program Guidance Appendix F in Air Quality Modeling and the
for Regulatory Modeling Applications. EPA Clean Air Act: Recommendations to EPA on

385
Pt. 51, App. W 40 CFR Ch. I (7–1–97 Edition)
Dispersion Modeling for Regulatory Applica- Use of Emission Reduction Credits. FEDERAL
tions. American Meteorological Society, REGISTER, 51: 43814–43860.
Boston, MA; pp. 159–171. (Docket No. A–80–46, 90. Environmental Research and Tech-
II–A–106) nology, 1987. User’s Guide to the Rough Ter-
79. Bowne, N.E. and R.J. Londergan, 1983. rain Diffusion Model (RTDM), Rev. 3.20. ERT
Overview, Results, and Conclusions for the Document No. P–D535–585. Environmental
EPRI Plume Model Validation and Develop- Research and Technology, Inc., Concord, MA.
ment Project: Plains Site. EPRI EA–3074. (NTIS No. PB 88–171467)
Electric Power Research Institute, Palo 91. Burns, D.J., S.G. Perry and A.J.
Alto, CA. Cimorelli, 1991. An Advanced Screening
80. Moore, G.E., T.E. Stoeckenius and D.A. Model for Complex Terrain Applications.
Stewart, 1982. A Survey of Statistical Meas- Paper presented at the 7th Joint Conference
ures of Model Performance and Accuracy for on Applications of Air Pollution Meteor-
Several Air Quality Models. EPA Publica- ology (cosponsored by the American Mete-
tion No. EPA–450/4–83–001. U.S. Environ- orological Society and the Air & Waste Man-
mental Protection Agency, Research Tri- agement Association), January 13–18, 1991,
angle Park, NC. (NTIS No. PB 83–260810) New Orleans, LA.
81. Rhoads, R.G., 1981. Accuracy of Air 92. Perry, S.G., 1992. CTDMPLUS: A Disper-
Quality Models. Staff Report. U.S. Environ- sion Model for Sources near Complex Topog-
mental Protection Agency, Research Tri- raphy. Part I: Technical Formulations. Jour-
angle Park, NC. (Docket No. A–80–46, II–G–6) nal of Applied Meteorology, 31(7): 633–645.
82. Pasquill, F., 1974. Atmospheric Diffu-
93. Paumier, J.O., S.G. Perry and D.J.
sion, 2nd Edition. John Wiley and Sons, New
Burns, 1992. CTDMPLUS: A Dispersion Model
York, NY; 479 pp.
for Sources near Complex Topography. Part
83. Austin, B.S., T.E. Stoeckenius, M.C.
II: Performance Characteristics. Journal of
Dudik and T.S. Stocking, 1988. User’s Guide
Applied Meteorology, 31(7): 646–660.
to the Expected Exceedances System. Sys-
tems Applications, Inc., San Rafael, CA. Pre- 94. Environmental Protection Agency, 1986.
pared under Contract No. 68–02–4352 Option I Evaluation of Mobile Source Air Quality
for the U.S. Environmental Protection Agen- Simulation Models. EPA Publication No.
cy, Research Triangle Park, NC. (Docket No. EPA–450/4–86–002. U.S. Environmental Pro-
A–88–04, II–I–3) tection Agency, Research Triangle Park, NC.
84. Thrall, A.D., T.E. Stoeckenius and C.S. (NTIS No. PB 86–167293)
Burton, 1985. A Method for Calculating Dis- 95. Shannon, J.D., 1987. Mobile Source Mod-
persion Modeling Uncertainty Applied to the eling Review. A report prepared under a co-
Regulation of an Emission Source. Systems operative agreement with the Environ-
Applications, Inc., San Rafael, CA. Prepared mental Protection Agency. (Docket No. A–
for the U.S. Environmental Protection Agen- 88–04, II–J–2)
cy, Research Triangle Park, NC. (Docket No. 96. Environmental Protection Agency, 1991.
A–80–46, IV–G–1) Emission Inventory Requirements for Car-
85. Environmental Protection Agency, 1981. bon Monoxide State Implementation Plans.
Plan for Evaluating Model Performance. EPA Publication No. EPA–450/4–91–011. U.S.
Staff Report. U.S. Environmental Protection Environmental Protection Agency, Research
Agency, Research Triangle Park, NC. (Dock- Triangle Park, NC. (NTIS No. PB 92–112150)
et No. A–80–46, II–G–6) 97. Environmental Protection Agency, 1992.
86. Environmental Protection Agency, 1979. Guideline for Regulatory Application of the
Guideline for the Interpretation of Ozone Air Urban Airshed Model for Areawide Carbon
Quality Standards. EPA Publication No. Monoxide. EPA Publication No. EPA–450/4–
EPA 450/4–79–003. U.S. Environmental Protec- 92–011a and b. U.S. Environmental Protec-
tion Agency, Research Triangle Park, NC. tion Agency, Research Triangle Park, NC.
(NTIS No. PB–292271) (NTIS Nos. PB 92–213222 and PB 92–213230)
87. Environmental Protection Agency, 1974. 98. Environmental Protection Agency, 1992.
Guidelines for Air Quality Maintenance Technical Support Document to Aid States
Planning and Analysis, Volume 12: Applying with the Development of Carbon Monoxide
Atmospheric Simulation Models to Air Qual- State Implementation Plans. EPA Publica-
ity Maintenance Areas. EPA Publication No. tion No. EPA–452/R–92–003. U.S. Environ-
EPA–450/4–74–013. U.S. Environmental Pro- mental Protection Agency, Research Tri-
tection Agency, Research Triangle Park, NC. angle Park, NC. (NTIS No. PB 92–233055)
(NTIS No. PB–237750) 99. Environmental Protection Agency, 1986.
88. Environmental Protection Agency, 1977. Evaluation of Short-Term Long-Range
Guidelines for Interpretation of Air Quality Transport Models, Volumes I and II. EPA
Standards (Revised). OAQPS No. 1.2–008. U.S. Publication Nos. EPA–450/4–86–016a and b.
Environmental Protection Agency, Research U.S. Environmental Protection Agency, Re-
Triangle Park, NC. (NTIS No. PB 81–196420) search Triangle Park, NC. (NTIS Nos. PB 87–
89. Environmental Protection Agency, 1986. 142337 and PB 87–142345)
Emissions Trading Policy Statement; Gen- 100. McNider, R.T., 1987. Review of Short-
eral Principles for Creation, Banking, and Term Long-Range Models. A report prepared

386
Environmental Protection Agency Pt. 51, App. W
under a cooperative agreement with the En- 110. Sestak, M.L. and A.R. Riebau, 1988.
vironmental Protection Agency. (Docket No. SASEM Simple Approach Smoke Estimation
A–88–04, II–I–10) Model. U.S. Bureau of Land Management,
101. Lamb, R.G., 1983. A Regional Scale Technical Note 382. BLM/YA/PT–88/003 + 7000.
(1,000km) Model of Photochemical Air Pollu- Available from Printed Materials Distribu-
tion, Part I—Theoretical Formulation, Part tion Section, BLM Service Center (SC–658B),
II—Input Processor Network Design, and Denver, CO 80225–0047. (NTIS No. PB 90–
Part III—Tests of Numerical Algorithms. 185653)
EPA Publication Nos. EPA–600/3–83–035, 111. Environmental Protection Agency,
EPA–600/3–84–085, and EPA–600/3–85–037. U.S. 1992. A Modeling Protocol for Applying
Environmental Protection Agency, Research Mesopuff II to Long Range Transport Prob-
Triangle Park, NC. (NTIS Nos. PB 83–207688, lems. EPA Publication No. EPA–454/R–92–021.
PB 84–232651, and PB 85–203818, respectively) U.S. Environmental Protection Agency, Re-
102. Young, J.O., M. Aissa, T.L. Boehm et search Triangle Park, NC.
al., 1989. Development of the Regional Oxi- 112. DiCristofaro, D.C. and S.R. Hanna,
dant Model, Version 2.1. EPA Publication 1989. The Offshore and Coastal Dispersion
No. EPA–600/3–89–044. U.S. Environmental (OCD) Model, Volume I: User’s Guide, Vol-
Protection Agency, Research Triangle Park, ume II: Appendices. Version 4 Prepared for
NC. (NTIS No. PB 89–194252) Minerals Management Services by Sigma Re-
103. Environmental Protection Agency, search Corporation, Westford, MA. (Docket
1991. The Regional Oxidant Model (ROM) No. A–88–04, II–D–A–06)
113. Federal Aviation Administration, 1988.
User’s Guide. Part 1: The ROM
A Microcomputer Pollution Model for Civil-
Preprocessors. EPA Publication No. EPA–
ian Airports and Air Force Bases, Model De-
600/8–90–083a (NTIS No. PB 91–171926); Part 2:
scription, Model Application and Back-
The ROM Processor Network. EPA Publica-
ground, and EDMS User’s Guide (June 1991).
tion No. EPA–600/8–90–083b (NTIS No. PB 91–
Federal Aviation Administration Publica-
171934); Part 3: The Core Model. EPA Publi-
tion Nos. FAA–EE–88–4 and 5; FAA–EE–91–3,
cation No. EPA–600/8–90–083c (NTIS No. PB
respectively. United States Air Force Publi-
91–171942). U.S. Environmental Protection
cation Nos. ESL–TR–88–53 and 55; ESL–TR–
Agency, Research Triangle Park, NC.
91–31, respectively. Federal Aviation Admin-
104. Chang, J.S., R.A. Brost, I.S.A. Isaksen, istration, Office of Environment and Energy,
S. Madronich, P. Middleton, W.R. Stockwell Washington, D.C. (NTIS Nos. ADA 199003,
and C.J. Waleck, 1987. A Three-Dimensional ADA 199794, and ADA 240528, respectively)
Eulerian Acid Deposition Model: Physical 114. Environmental Protection Agency,
Concepts and Formulation. Journal of Geo- 1992. Workbook of Screening Techniques for
physical Research, 92(D12): 14681–14700. Assessing Impacts of Toxic Air Pollutants
105. Environmental Protection Agency, (Revised). EPA Publication No. EPA–454/R–
1987. Protocol for Applying and Validating 92–024. U.S. Environmental Protection Agen-
the CMB. U.S. Environmental Protection cy, Research Triangle Park, NC.
Agency. EPA Publication No. EPA–450/4–87– 115. Environmental Protection Agency,
010. U.S. Environmental Protection Agency, 1990. User’s Guide to TSCREEN: A Model for
Research Triangle Park, NC. (NTIS No. PB Screening Toxic Air Pollutant Concentra-
87–206496) tions. EPA Publication No. EPA–450/4–90–013.
106. Environmental Protection Agency, U.S. Environmental Protection Agency, Re-
1987. Protocol for Reconciling Differences search Triangle Park, NC. (NTIS No. PB 91–
Among Receptor and Dispersion Models. 141820)
EPA Publication No. EPA–450/4–87–008. U.S. 116. Environmental Protection Agency,
Environmental Protection Agency, Research 1989. Hazardous Waste TSDF Fugitive Par-
Triangle Park, NC. (NTIS No. PB 87–206504) ticulate Matter Air Emissions Guidance Doc-
107. Environmental Protection Agency, ument. EPA Publication No. EPA–450/3–89–
1988. Chemical Mass Balance Model 019. U.S. Environmental Protection Agency,
Diagnostics. EPA Publication No. EPA–450/4– Research Triangle Park, NC. (NTIS No. PB
88–005. U.S. Environmental Protection Agen- 90–103250)
cy, Research Triangle Park, NC. (NTIS No. 117. Environmental Protection Agency,
PB 88–208319) 1989. Procedures for Conducting Air Pathway
108. Environmental Protection Agency, Analyses for Superfund Applications, Vol-
1987. PM–10 SIP Development Guideline. EPA ume I Applications of Air Pathway Analyses
Publication No. EPA–450/2–86–001. U.S. Envi- for Superfund Activities and Volume IV Pro-
ronmental Protection Agency, Research Tri- cedures for Dispersion Modeling and Air
angle Park, NC. (NTIS No. PB 87–206488) Monitoring for Superfund Air Pathway Anal-
109. Environmental Protection Agency, ysis, EPA–450/1–89–001 and 004. U.S. Environ-
1987. Example Modeling to Illustrate SIP De- mental Protection Agency, Research Tri-
velopment for The PM–10 NAAQS. EPA Pub- angle Park, NC. (NTIS Nos. PB 89–113374 and
lication No. EPA–450/4–87–012. U.S. Environ- PB 89–113382)
mental Protection Agency, Research Tri- 118. Environmental Protection Agency,
angle Park, NC. (NTIS No. PB 87–205191) 1988. Air Dispersion Modeling as Applied to

387
Pt. 51, App. W 40 CFR Ch. I (7–1–97 Edition)
Hazardous Waste Incinerator Evaluations, 129. Environmental Protection Agency,
An Introduction For the Permit Writer. U.S. 1992. Toxic Modeling System Long-term
Environmental Protection Agency, Research (TOXLT) User’s Guide. EPA Publication No.
Triangle Park, NC. (Docket No. A–88–04, II– EPA–450/4–92–003. Environmental Protection
J–10) Agency, Research Triangle Park, NC.
119. Environmental Protection Agency, 130. Environmental Protection Agency,
1989. U.S. EPA Office of Toxic Substances 1989. User’s Guide for the DEGADIS 2.1 Dense
Graphical Exposure Modeling System Gas Dispersion Model. EPA Publication No.
(GEMS) User’s Guide and GAMS Version 3.0 EPA–450/4–89–019. U.S. Environmental Pro-
User’s Guide (DRAFT). Prepared under Con-
tection Agency, Research Triangle Park, NC.
tract No. 68–02–0481 for the U.S. Environ-
(NTIS No. PB 90–213893)
mental Protection Agency, Washington, D.C.
(Docket No. A–88–04, II–J–5a and II–J–13) 131. Environmental Protection Agency,
120. Federal Emergency Management 1993. Guidance on the Application of Refined
Agency, 1989. Handbook of Chemical Hazard Models for Air Toxics Releases. EPA Publi-
Analysis Procedures. Available on request by cation No. EPA–450/4–91–007. Environmental
writing to: Federal Emergency Management Protection Agency, Research Triangle Park,
Agency, Publications Office, 500 C Street, NC. (NTIS No. PB 91–190983)
S.W., Washington, D.C. 20472. 132. Perry, R.H. and Chilton, C.H., 1973.
121. Environmental Protection Agency, Chemical Engineers’ Handbook, Fifth Edition,
1987. Technical Guidance for Hazards Analy- McGraw-Hill Book Company, New York, NY.
sis: Emergency Planning for Extremely Haz- 133. Environmental Protection Agency,
ardous Substances. Available on request by 1988. User’s Guide to SDM—A Shoreline Dis-
telephone: 1–800–535–0202. persion Model. EPA Publication No. EPA–
122. Environmental Protection Agency, 450/4–88–017. U.S. Environmental Protection
1988. Superfund Exposure Assessment Man- Agency, Research Triangle Park, NC. (NTIS
ual. EPA–540/1–88–001, OSWER Directive No. PB 89–164305)
9285.5–1. Office of Remedial Response, Wash- 134. Environmental Protection Agency,
ington, D.C. 20460. (NTIS No. PB 89–135859) 1987. Analysis and Evaluation of Statistical
123. Environmental Protection Agency, Coastal Fumigation Models. EPA Publica-
1989. Incineration of Sewage Sludge; Tech- tion No. EPA–450/4–87–002. U.S. Environ-
nical Support Document. Office of Water
mental Protection Agency, Research Tri-
Regulations and Standards, Washington,
angle Park, NC. (NTIS No. PB 87–175519)
D.C. 20460. (NTIS No. PB 89–136592)
124. Environmental Protection Agency, 135. Environmental Protection Agency,
1989. Sludge Incineration Modeling (SIM) 1996. Meteorological Processor for Regu-
System User’s Guide (Draft). Office of Pes- latory Models (MPRM) User’s Guide. EPA
ticides and Toxic Substances, Exposure Eval- Publication No. EPA–454/B–96–002. U.S. Envi-
uation Division, Washington, D.C. 20460. ronmental Protection Agency, Research Tri-
(NTIS No. PB 89–138762) angle Park, NC. (NTIS No. PB 96–180518)
125. Environmental Protection Agency, 136. Bowen, B.M., J.M. Dewart and A.I.
1989. Risk Assessment Guidance for Chen, 1983. Stability Class Determination: A
Superfund. Volume I: Human Health Evalua- Comparison for One Site. Proceedings, Sixth
tion Manual Part A. (Interim Final). OSWER Symposium on Turbulence and Diffusion.
Directive 9285.7–01a. Office of Solid Waste American Meteorological Society, Boston,
and Emergency Response, Washington, D.C. MA; pp. 211–214. (Docket No. A–92–65, II–A–7)
20460. 137. Environmental Protection Agency,
126. Environmental Protection Agency, 1993. An Evaluation of a Solar Radiation/
1986. User’s Manual for the Human Exposure Delta-T (SRDT) Method for Estimating
Model (HEM). EPA Publication No. EPA–450/ Pasquill-Gifford (P–G) Stability Categories.
5–86–001. Office of Air Quality Planning and EPA Publication No. EPA–454/R–93–055. U.S.
Standards, Research Triangle Park, NC.
Environmental Protection Agency, Research
27711.
Triangle Park, NC. (NTIS No. PB 94–113958)
127. Environmental Protection Agency,
138. Environmental Protection Agency,
1992. A Tiered Modeling Approach for Assess-
ing the Risks Due to Sources of Hazardous 1993. PCRAMMET User’s Guide. EPA Publi-
Air Pollutants. EPA Publication No. EPA– cation No. EPA–454/B–93–009. U.S. Environ-
450/4–92–001. Environmental Protection Agen- mental Protection Agency, Research Tri-
cy, Research Triangle Park, NC. (NTIS No. angle Park, NC.
PB 92–164748) 139. American Society of Mechanical Engi-
128. Environmental Protection Agency, neers, 1979. Recommended Guide for the Pre-
1992. Toxic Modeling System Short-term diction of Airborne Effluents, Third Edition.
(TOXST) User’s Guide. EPA Publication No. American Society of Mechanical Engineers,
EPA–450/4–92–002. Environmental Protection New York, NY.
Agency, Research Triangle Park, NC.

388
Environmental Protection Agency Pt. 51, App. W
13.0 BIBLIOGRAPHY i Hanna, S.R., G.A. Briggs and R.P. Hosker,
Jr., 1982. Handbook on Atmospheric Diffu-
American Meteorological Society, 1971–
sion. Technical Information Center, U.S. De-
1985. Symposia on Turbulence, Diffusion, and
partment of Energy, Washington, D.C.
Air Pollution (1st–7th), Boston, MA.
Haugen, D.A., Workshop Coordinator, 1975.
American Meteorological Society, 1977–
Lectures on Air Pollution and Environ-
1984. Joint Conferences on Applications of
mental Impact Analyses. Sponsored by the
Air Pollution Meteorology (1st–4th). Spon-
American Meteorological Society, Boston,
sored by the American Meteorological Soci-
MA.
ety and the Air Pollution Control Associa-
Hoffnagle, G.F., M.E. Smith, T.V. Crawford
tion, Boston, MA.
and T.J. Lockhart, 1981. On-site Meteorologi-
American Meteorological Society, 1978. Ac-
cal Instrumentation Requirements to Char-
curacy of Dispersion Models. Bulletin of the
acterize Diffusion from Point Sources—A
American Meteorological Society, 59(8): 1025–
Workshop, 15–17 January 1980, Raleigh, NC.
1026.
Bulletin of the American Meteorological So-
American Meteorological Society, 1981. Air ciety, 62(2): 255–261.
Quality Modeling and the Clean Air Act:
McMahon, R.A. and P.J. Denison, 1979. Em-
Recommendations to EPA on Dispersion
pirical Atmospheric Deposition Param-
Modeling for Regulatory Applications, Bos-
eters—A Survey. Atmospheric Environment,
ton, MA.
13: 571–585.
Briggs, G.A., 1969. Plume Rise. U.S. Atomic McRae, G.J., J.A. Leone and J.H. Seinfeld,
Energy Commission Critical Review Series, 1983. Evaluation of Chemical Reaction Mech-
Oak Ridge National Laboratory, Oak Ridge, anisms for Photochemical Smog. Part I:
TN. Mechanism Descriptions and Documenta-
Dickerson, W.H. and P.H. Gudiksen, 1980. tion. EPA Publication No. EPA–600/3/83–086.
ASCOT FY 79 Program Report. Report U.S. Environmental Protection Agency, Re-
UCRL—52899, ASCOT 80–1. Lawrence Liver- search Triangle Park, NC.
more National Laboratory, Livermore, CA. Pasquill, F. and F.B. Smith, 1983. Atmos-
Drake, R.L. and S.M. Barrager, 1979. Math- pheric Diffusion, 3rd Edition. Ellis Horwood
ematical Models for Atmospheric Pollutants. Limited, Chichester, West Sussex, England,
EPRI EA–1131. Electric Power Research In- 438 pp.
stitute, Palo Alto, CA. Randerson, D., Ed., 1984. Atmospheric
Environmental Protection Agency, 1978. Science and Power Production. DOE/TIC
Workbook for Comparison of Air Quality 2760l. Office of Scientific and Technical In-
Models. EPA Publication No. EPA–450/2–78– formation, U.S. Department of Energy, Oak
028a and b. U.S. Environmental Protection Ridge, TN.
Agency, Research Triangle Park, NC. Roberts, J.J., Ed., 1977. Report to U.S. EPA
Fox, D.G., and J.E. Fairobent, 1981. NCAQ of the Specialists’ Conference on the EPA
Panel Examines Uses and Limitations of Air Modeling Guideline. U.S. Environmental
Quality Models. Bulletin of the American Protection Agency, Research Triangle Park,
Meteorological Society, 62(2): 218–221. NC.
Gifford, F.A., 1976. Turbulent Diffusion Smith, M.E., Ed., 1973. Recommended
Typing Schemes: A Review. Nuclear Safety, Guide for the Prediction of the Dispersion of
17(1): 68–86. Airborne Effluents. The American Society of
Gudiksen, P.H., and M.H. Dickerson, Eds., Mechanical Engineers, New York, NY.
Executive Summary: Atmospheric Studies in Stern, A.C., Ed., 1976. Air Pollution, Third
Complex Terrain Technical Progress Report Edition, Volume I: Air Pollutants, Their
FY–1979 Through FY–1983. Lawrence Liver- Transformation and Transport. Academic
more National Laboratory, Livermore, CA. Press, New York, NY.
(Docket Reference No. II–I–103). Turner, D.B., 1979. Atmospheric Dispersion
Hales, J.M., 1976. Tall Stacks and the At- Modeling: A Critical Review. Journal of the
mospheric Environment. EPA Publication Air Pollution Control Association, 29(5): 502–
No. EPA–450/3–76–007. U.S. Environmental 519.
Protection Agency, Research Triangle Park, Whiteman, C.D. and K.J. Allwine, 1982.
NC. Green River Ambient Model Assessment Pro-
Hanna, S.R., G.A. Briggs, J. Deardorff, B.A. gram FY–1982 Progress Report. PNL–4520.
Egan, G.A. Gifford and F. Pasquill, 1977. AMS Pacific Northwest Laboratory, Richland,
Workshop on Stability Classification WA.
Schemes And Sigma Curves—Summary of
Recommendations. Bulletin of the American 14.0 GLOSSARY OF TERMS
Meteorological Society, 58(12): 1305–1309.
Air quality. Ambient pollutant concentra-
tions and their temporal and spatial dis-
i The documents listed here are major tribution.
sources of supplemental information on the Algorithm. A specific mathematical cal-
theory and application of mathematical air culation procedure. A model may contain
quality models. several algorithms.

389
Pt. 51, App. W 40 CFR Ch. I (7–1–97 Edition)
Background. Ambient pollutant concentra- APPENDIX A TO APPENDIX W OF PART 51—
tions due to: SUMMARIES OF PREFERRED AIR QUALITY
(1) Natural sources; MODELS
(2) Nearby sources other than the one(s)
currently under consideration; and Table of Contents
(3) Unidentified sources. A.0 Introduction and Availability
Calibrate. An objective adjustment using A.1 Buoyant Line and Point Source Disper-
measured air quality data (e.g., an adjust- sion Model (BLP)
ment based on least-squares linear regres- A.2 Caline3
sion). A.3 Climatological Dispersion Model (CDM
Calm. For purposes of air quality modeling, 2.0)
calm is used to define the situation when the A.4 Gaussian-Plume Multiple Source Air
wind is indeterminate with regard to speed Quality Algorithm (RAM)
or direction. A.5 Industrial Source Complex Model (ISC3)
Complex terrain. Terrain exceeding the A.6 Urban Airshed Model (UAM)
height of the stack being modeled. A.7 Offshore and Coastal Dispersion Model
Computer code. A set of statements that (OCD)
comprise a computer program. A.8 Emissions and Dispersion Modeling Sys-
Evaluate. To appraise the performance and
tem (EDMS)
accuracy of a model based on a comparison
A.9 Complex Terrain Dispersion Model Plus
of concentration estimates with observed air
Algorithms For Unstable Situations
quality data.
(CTDMPLUS)
Fluid modeling. Modeling conducted in a
A.REF References
wind tunnel or water channel to quan-
titatively evaluate the influence of buildings A.0 Introduction and Availability
and/or terrain on pollutant concentrations.
Fugitive dust. Dust discharged to the at- This appendix summarizes key features of
mosphere in an unconfined flow stream such refined air quality models preferred for spe-
as that from unpaved roads, storage piles cific regulatory applications. For each
and heavy construction operations. model, information is provided on availabil-
Model. A quantitative or mathematical ity, approximate cost, regulatory use, data
representation or simulation which attempts input, output format and options, simulation
to describe the characteristics or relation- of atmospheric physics, and accuracy. These
ships of physical events. models may be used without a formal dem-
Preferred model. A refined model that is rec- onstration of applicability provided they sat-
ommended for a specific type of regulatory isfy the recommendations for regulatory use;
application. not all options in the models are necessarily
Receptor. A location at which ambient air recommended for regulatory use.
quality is measured or estimated. Many of these models have been subjected
Receptor models. Procedures that examine to a performance evaluation using compari-
an ambient monitor sample of particulate sons with observed air quality data. A sum-
matter and the conditions of its collection to mary of such comparisons for models con-
infer the types or relative mix of sources im- tained in this appendix is included in Moore
pacting on it during collection. et al. (1982). Where possible, several of the
Refined model. An analytical technique models contained herein have been subjected
that provides a detailed treatment of phys- to evaluation exercises, including (1) statis-
ical and chemical atmospheric processes and tical performance tests recommended by the
requires detailed and precise input data. Spe- American Meteorological Society and (2)
cialized estimates are calculated that are peer scientific reviews. The models in this
useful for evaluating source impact relative appendix have been selected on the basis of
to air quality standards and allowable incre- the results of the model evaluations, experi-
ments. The estimates are more accurate ence with previous use, familiarity of the
than those obtained from conservative model to various air quality programs, and
screening techniques. the costs and resource requirements for use.
Rollback. A simple model that assumes All models and user’s documentation in
that if emissions from each source affecting this appendix are available from: Computer
a given receptor are decreased by the same Products, National Technical Information
percentage, ambient air quality concentra- Service (NTIS), U.S. Department of Com-
tions decrease proportionately. merce, Springfield, VA 22161, Phone: (703)
Screening technique. A relatively simple 487–4650. In addition, model codes and se-
analysis technique to determine if a given lected, abridged user’s guides are available
source is likely to pose a threat to air qual- from the Support Center for Regulatory Air
ity. Concentration estimates from screening Models Bulletin Board System 19 (SCRAM
techniques are conservative. BBS), telephone (919) 541–5742. The SCRAM
Simple terrain. An area where terrain fea- BBS is an electronic bulletin board system
tures are all lower in elevation than the top designed to be user friendly and accessible
of the stack of the source. from anywhere in the country. Model users

390
Environmental Protection Agency Pt. 51, App. W
with personal computers are encouraged to ommended model if it can be demonstrated,
use the SCRAM BBS to download current using the criteria in section 3.2, that the
model codes and text files. model is more appropriate for a specific ap-
plication.
A.1 Buoyant Line and Point Source Dispersion
Model (BLP) b. Input Requirements
Reference Source data: point sources require stack
location, elevation of stack base, physical
Schulman, Lloyd L. and Joseph S. Scire, stack height, stack inside diameter, stack
1980. Buoyant Line and Point Source (BLP) gas exit velocity, stack gas exit tempera-
Dispersion Model User’s Guide. Document P– ture, and pollutant emission rate. Line
7304B. Environmental Research and Tech- sources require coordinates of the end points
nology, Inc., Concord, MA. (NTIS No. PB 81– of the line, release height, emission rate, av-
164642) erage line source width, average building
Availability width, average spacing between buildings,
and average line source buoyancy parameter.
The computer code is available on the Sup- Meteorological data: hourly surface weath-
port Center for Regulatory Models Bulletin er data from punched cards or from the
Board System and also on diskette (as PB 90– preprocessor program RAMMET which pro-
500281) from the National Technical Informa- vides hourly stability class, wind direction,
tion Service (see section A.0). wind speed, temperature, and mixing height.
Receptor data: locations and elevations of
Abstract receptors, or location and size of receptor
BLP is a Gaussian plume dispersion model grid or request automatically generated re-
designed to handle unique modeling prob- ceptor grid.
lems associated with aluminum reduction
plants, and other industrial sources where c. Output
plume rise and downwash effects from sta- Printed output (from a separate post-proc-
tionary line sources are important. essor program) includes:
Total concentration or, optionally, source
a. Recommendations for Regulatory Use contribution analysis; monthly and annual
The BLP model is appropriate for the fol- frequency distributions for 1-, 3-, and 24-hour
lowing applications: average concentrations; tables of 1-, 3-, and
Aluminum reduction plants which contain 24-hour average concentrations at each re-
buoyant, elevated line sources; ceptor; table of the annual (or length of run)
Rural areas; average concentrations at each receptor;
Transport distances less than 50 kilo- Five highest 1-, 3-, and 24-hour average
meters; concentrations at each receptor; and
Simple terrain; and Fifty highest 1-, 3-, and 24-hour concentra-
One hour to one year averaging times. tions over the receptor field.
The following options should be selected
for regulatory applications: d. Type of Model
Rural (IRU=1) mixing height option; BLP is a gaussian plume model.
Default (no selection) for plume rise wind
shear (LSHEAR), transitional point source e. Pollutant Types
plume rise (LTRANS), vertical potential
BLP may be used to model primary pollut-
temperature gradient (DTHTA), vertical
ants. This model does not treat settling and
wind speed power law profile exponents
deposition.
(PEXP), maximum variation in number of
stability classes per hour (IDELS), pollutant f. Source-Receptor Relationship
decay (DECFAC), the constant in Briggs’ sta-
ble plume rise equation (CONST2), constant BLP treats up to 50 point sources, 10 par-
in Briggs’ neutral plume rise equation allel line sources, and 100 receptors arbitrar-
(CONST3), convergence criterion for the line ily located.
source calculations (CRIT), and maximum it- User-input topographic elevation is applied
erations allowed for line source calculations for each stack and each receptor.
(MAXIT); and
g. Plume Behavior
Terrain option (TERAN) set equal to 0.0,
0.0, 0.0, 0.0, 0.0, 0.0 BLP uses plume rise formulas of Schulman
For other applications, BLP can be used if and Scire (1980).
it can be demonstrated to give the same esti- Vertical potential temperature gradients
mates as a recommended model for the same of 0.02 Kelvin per meter for E stability and
application, and will subsequently be exe- 0.035 Kelvin per meter are used for stable
cuted in that mode. plume rise calculations. An option for user
BLP can be used on a case-by-case basis input values is included.
with specific options not available in a rec- Transitional rise is used for line sources.

391
Pt. 51, App. W 40 CFR Ch. I (7–1–97 Edition)
Option to suppress the use of transitional A.2 CALINE3
plume rise for point sources is included.
The building downwash algorithm of Reference
Schulman and Scire (1980) is used. Benson, Paul E., 1979. CALINE3—A Versa-
tile Dispersion Model for Predicting Air Pol-
h. Horizontal Winds lutant Levels Near Highways and Arterial
Constant, uniform (steady-state) wind is Streets. Interim Report, Report Number
FHWA/CA/TL–79/23. Federal Highway Admin-
assumed for an hour.
istration, Washington, D.C. (NTIS No. PB 80–
Straight line plume transport is assumed 220841)
to all downwind distances.
Wind speeds profile exponents of 0.10, 0.15, Availability
0.20, 0.25, 0.30, and 0.30 are used for stability
The CALINE3 model is available on
classes A through F, respectively. An option diskette (as PB 95–502712) from NTIS. The
for user-defined values and an option to sup- source code and user’s guide are also avail-
press the use of the wind speed profile fea- able on the Support Center for Regulatory
ture are included. Models Bulletin Board System (see section
A.0).
i. Vertical Wind Speed
Abstract
Vertical wind speed is assumed equal to
zero. CALINE3 can be used to estimate the con-
centrations of nonreactive pollutants from
j. Horizontal Dispersion highway traffic. This steady-state Gaussian
model can be applied to determine air pollu-
Rural dispersion coefficients are from
tion concentrations at receptor locations
Turner (1969), with no adjustment made for
downwind of ‘‘at-grade,’’ ‘‘fill,’’ ‘‘bridge,’’
variations in surface roughness or averaging and ‘‘cut section’’ highways located in rel-
time. atively uncomplicated terrain. The model is
Six stability classes are used. applicable for any wind direction, highway
orientation, and receptor location. The
k. Vertical Dispersion model has adjustments for averaging time
Rural dispersion coefficients are from and surface roughness, and can handle up to
Turner (1969), with no adjustment made for 20 links and 20 receptors. It also contains an
variations in surface roughness. algorithm for deposition and settling veloc-
Six stability classes are used. ity so that particulate concentrations can be
predicted.
Mixing height is accounted for with mul-
tiple reflections until the vertical plume a. Recommendations for Regulatory Use
standard deviation equals 1.6 times the mix-
ing height; uniform mixing is assumed be- CALINE–3 is appropriate for the following
applications:
yond that point.
Highway (line) sources;
Perfect reflection at the ground is as- Urban or rural areas;
sumed. Simple terrain;
Transport distances less than 50 kilo-
l. Chemical Transformation
meters; and
Chemical transformations are treated One-hour to 24-hour averaging times.
using linear decay. Decay rate is input by
the user. b. Input Requirements
Source data: up to 20 highway links classed
m. Physical Removal as ‘‘at-grade,’’ ‘‘fill’’ ‘‘bridge,’’ or ‘‘de-
Physical removal is not explicitly treated. pressed’’; coordinates of link end points;
traffic volume; emission factor; source
n. Evaluation Studies height; and mixing zone width.
Meteorological data: wind speed, wind
Schulman, L.L. and J.S. Scire, 1980. Buoy- angle (measured in degrees clockwise from
ant Line and Point Source (BLP) Dispersion the Y axis), stability class, mixing height,
Model User’s Guide, P–7304B. Environmental ambient (background to the highway) con-
Research and Technology, Inc., Concord, MA. centration of pollutant.
Scire, J.S. and L.L. Schulman, 1981. Eval- Receptor data: coordinates and height
uation of the BLP and ISC Models with SF6 above ground for each receptor. c.
Tracer Data and SO2 Measurements at Alu-
c. Output
minum Reduction Plants. APCA Specialty
Conference on Dispersion Modeling for Com- Printed output includes concentration at
plex Sources, St. Louis, MO. each receptor for the specified meteorologi-
cal condition.

392
Environmental Protection Agency Pt. 51, App. W
d. Type of Model A.3 Climatological Dispersion Model (CDM 2.0)
CALINE–3 is a Gaussian plume model. Reference
e. Pollutant Types Irwin, J.S., T. Chico and J. Catalano, 1985.
CALINE–3 may be used to model primary CDM 2.0—Climatological Dispersion Model—
pollutants. User’s Guide. U.S. Environmental Protection
Agency, Research Triangle Park, NC. (NTIS
f. Source-Receptor Relationship No. PB 86–136546)
Up to 20 highway links are treated. Availability
CALINE–3 applies user input location and
emission rate for each link. User-input re- The source code and user’s guide is avail-
ceptor locations are applied. able on the Support Center for Regulatory
Models Bulletin Board System. The com-
g. Plume Behavior puter code is also available on diskette (as
Plume rise is not treated. PB 90–500406) from the National Technical
Information Service (see section A.0).
h. Horizontal Winds
Abstract
User-input hourly wind speed and direction
are applied. CDM is a climatological steady-state
Constant, uniform (steady-state) wind is Gaussian plume model for determining long-
assumed for an hour. term (seasonal or annual) arithmetic aver-
age pollutant concentrations at any ground-
i. Vertical Wind Speed level receptor in an urban area.
Vertical wind speed is assumed equal to a. Recommendations for Regulatory Use
zero.
CDM is appropriate for the following appli-
j. Horizontal Dispersion cations:
Six stability classes are used. Point and area sources;
Rural dispersion coefficients from Turner Urban areas;
(1969) are used, with adjustment for rough- Flat terrain;
ness length and averaging time. Transport distances less than 50 kilo-
Initial traffic-induced dispersion is handled meters;
implicitly by plume size parameters. Long term averages over one month to one
year or longer.
k. Vertical Dispersion The following option should be selected for
Six stability classes are used. regulatory applications:
Empirical dispersion coefficients from Ben- Set the regulatory ‘‘default option’’
son (1979) are used including an adjustment (NDEF=1) which automatically selects stack
for roughness length. tip downwash, final plume rise, buoyancy-in-
Initial traffic-induced dispersion is handled duced dispersion (BID), and the appropriate
implicitly by plume size parameters. wind profile exponents.
Adjustment for averaging time is included. Enter ‘‘0’’ for pollutant half-life for all pol-
lutants except for SO2 in an urban setting.
l. Chemical Transformation This entry results in no decay (infinite half-
Not treated. life) being calculated. For SO2 in an urban
setting, the pollutant half-life (in hours)
m. Physical Removal should be set to 4.0.
Optional deposition calculations are in- b. Input Requirements
cluded.
Source data: location, average emissions
n. Evaluation Studies rates and heights of emissions for point and
Bemis, G.R. et al., 1977. Air Pollution and area sources. Point source data requirements
Roadway Location, Design, and Operation— also include stack gas temperature, stack
Project Overview. FHWA–CA–TL–7080–77–25, gas exit velocity, and stack inside diameter
Federal Highway Administration, Washing- for plume rise calculations for point sources.
ton, D.C. Meteorological data: stability wind rose
Cadle, S.H. et al., 1976. Results of the Gen- (STAR deck day/night version), average mix-
eral Motors Sulfate Dispersion Experiment, ing height and wind speed in each stability
GMR–2107. General Motors Research Labora- category, and average air temperature.
tories, Warren, MI. Receptor data: cartesian coordinates of
Dabberdt, W.F., 1975. Studies of Air Qual- each receptor.
ity on and Near Highways, Project 2761.
c. Output
Stanford Research Institute, Menlo Park,
CA. Printed output includes:

393
Pt. 51, App. W 40 CFR Ch. I (7–1–97 Edition)
Average concentrations for the period of • Briggs-urban (Gifford, 1976).
the stability wind rose data (arithmetic Mixing height has no effect until disper-
mean only) at each receptor, and sion coefficient equals 0.8 times the mixing
Optional point and area concentration rose height; uniform vertical mixing is assumed
for each receptor. beyond that point.
Buoyancy-induced dispersion (Pasquill,
d. Type of Model 1976) is included as an option. Perfect reflec-
CDM is a climatological Gaussian plume tion is assumed at the ground.
model.
l. Chemical Transformation
e. Pollutant Types Chemical transformations are treated
CDM may be used to model primary pollut- using exponential decay. Half-life is input by
ants. Settling and deposition are not treated. the user.

f. Source-Receptor Relationship m. Physical Removal

CDM applies user-specified locations for all Physical removal is not explicitly treated.
point sources and receptors. n. Evaluation Studies
Area sources are input as multiples of a
user-defined unit area source grid size. Busse, A.D. and J.R. Zimmerman, 1973.
User specified release heights are applied User’s Guide for the Climatological Disper-
for individual point sources and the area sion Model—Appendix E. EPA Publication
source grid. No. EPA/R4–73–024. Office of Research and
Actual separation between each source-re- Development, Research Triangle Park, NC.
ceptor pair is used. Irwin, J.S. and T.M. Brown, 1985. A Sen-
The user may select a single height at or sitivity Analysis of the Treatment of Area
above ground level that applies to all recep- Sources by the Climatological Dispersion
tors. Model. Journal of Air Pollution Control As-
No terrain differences between source and sociation, 35: 359–364.
receptor are treated. Londergan, R., D. Minott, D. Wachter and
R. Fizz, 1983. Evaluation of Urban Air Qual-
g. Plume Behavior ity Simulation Models, EPA Publication No.
CDM uses Briggs (1969, 1971, 1975) plume EPA–450/4–83–020. U.S. Environmental Pro-
rise equations. Optionally a plume rise-wind tection Agency, Research Triangle Park, NC.
speed product may be input for each point Zimmerman, J.R., 1971. Some Preliminary
source. Results of Modeling from the Air Pollution
Stack tip downwash equation from Briggs Study of Ankara, Turkey, Proceedings of the
(1974) is preferred for regulatory use. The Second Meeting of the Expert Panel on Air
Bjorklund and Bowers (1982) equation is also Pollution Modeling, NATO Committee on
included. the Challenges of Modern Society, Paris,
No plume rise is calculated for area France.
sources. Zimmerman, J.R., 1972. The NATO/CCMS
Does not treat fumigation or building Air Pollution Study of St. Louis, Missouri.
downwash. Presented at the Third Meeting of the Expert
Panel on Air Pollution Modeling, NATO
h. Horizontal Winds Committee on the Challenges of Modern So-
ciety, Paris, France.
Wind data are input as a stability wind
rose (joint frequency distribution of 16 wind A.4 Gaussian-Plume Multiple Source Air
directions, 6 wind classes, and 5 stability Quality Algorithm (RAM)
classes).
Wind speed profile exponents for the urban Reference
case (Irwin, 1979; EPA, 1980) are used, assum-
Turner, D.B. and J.H. Novak, 1978. User’s
ing the anemometer height is at 10.0 meters.
Guide for RAM. Publication No. EPA–600/8–
i. Vertical Wind Speed 78–016, Vol. a and b. U.S. Environmental Pro-
tection Agency, Research Triangle Park, NC.
Vertical wind speed is assumed equal to (NTIS Nos. PB 294791 and PB 294792)
zero. Catalano, J.A., D.B. Turner and H. Novak,
1987. User’s Guide for RAM—Second Edition.
j. Horizontal Dispersion
U.S. Environmental Protection Agency, Re-
Pollutants are assumed evenly distributed search Triangle Park, NC.
across a 22.5 or 10.0 degree sector.
Availability
k. Vertical Dispersion
The source code and user’s guide is avail-
There are seven vertical dispersion param- able on the Support Center for Regulatory
eter schemes, but the following is rec- Models Bulletin Board System. The com-
ommended for regulatory applications: puter code is also available on diskette (as

394
Environmental Protection Agency Pt. 51, App. W
PB 90–500315) from the National Technical f. Source-Receptor Relationship
Information Service (see section A.0).
RAM applies user-specified locations for
Abstract all point sources and receptors. Area sources
are input as multiples of a user-defined unit
RAM is a steady-state Gaussian plume area source grid size.
model for estimating concentrations of rel- User specified stack heights are applied for
atively stable pollutants, for averaging individual point sources.
times from an hour to a day, from point and Up to 3 effective release heights may be
area sources in a rural or urban setting. specified for the area sources. Area source
Level terrain is assumed. Calculations are release heights are assumed to be appro-
performed for each hour. priate for a 5 meter per second wind and to
be inversely proportional to wind speed.
a. Recommendations for Regulatory Use
Actual separation between each source-re-
RAM is appropriate for the following appli- ceptor pair is used.
cations: All receptors are assumed to be at the
Point and area sources; same height at or above ground level.
Urban areas; No terrain differences between source and
Flat terrain; receptor are accounted for.
Transport distances less than 50 kilo-
meters; and g. Plume Behavior
One hour to one year averaging times. RAM uses Briggs (1969, 1971, 1975) plume
The following options should be selected rise equations for final rise.
for regulatory applications: Stack tip downwash equation from Briggs
Set the regulatory ‘‘default option’’ to (1974) is used.
automatically select stack tip downwash, A user supplied fraction of the area source
final plume rise, buoyancy-induced disper- height is treated as the physical height. The
sion (BID), the new treatment for calms, the remainder is assumed to be plume rise for a
appropriate wind profile exponents, and the 5 meter per second wind speed, and to be in-
appropriate value for pollutant half-life. versely proportional to wind speed.
Fumigation and building downwash are not
b. Input Requirements
treated.
Source data: point sources require loca-
tion, emission rate, physical stack height, h. Horizontal Winds
stack gas exit velocity, stack inside diame- Constant, uniform (steady state) wind is
ter and stack gas temperature. Area sources assumed for an hour.
require location, size, emission rate, and Straight line plume transport is assumed
height of emissions. to all downwind distances.
Meteorological data: hourly surface weath- Separate wind speed profile exponents
er data from the preprocessor program (Irwin, 1979; EPA, 1980) for urban cases are
RAMMET which provides hourly stability used.
class, wind direction, wind speed, tempera-
ture, and mixing height. Actual anemometer i. Vertical Wind Speed
height (a single value) is also required. Vertical wind speed is assumed equal to
Receptor data: coordinates of each recep- zero.
tor. Options for automatic placement of re-
ceptors near expected concentration maxi- j. Horizontal Dispersion
ma, and a gridded receptor array are in-
cluded. Urban dispersion coefficients from Briggs
(Gifford, 1976) are used.
c. Output Buoyancy-induced dispersion (Pasquill,
1976) is included.
Printed output optionally includes: Six stability classes are used.
One to 24-hour and annual average con-
centrations at each receptor, k. Vertical Dispersion
Limited individual source contribution
Urban dispersion coefficients from Briggs
list, and
(Gifford, 1976) are used.
Highest through fifth highest concentra-
Buoyancy-induced dispersion (Pasquill,
tions at each receptor for period, with the
1976) is included.
highest and high, second-high values flagged.
Six stability classes are used.
d. Type of Model Mixing height is accounted for with mul-
tiple reflections until the vertical plume
RAM is a Gaussian plume model. standard deviation equals 1.6 times the mix-
ing height; uniform vertical mixing is as-
e. Pollutant Types
sumed beyond that point.
RAM may be used to model primary pollut- Perfect reflection is assumed at the
ants. Settling and deposition are not treated. ground.

395
Pt. 51, App. W 40 CFR Ch. I (7–1–97 Edition)
l. Chemical Transformation A.5 Industrial Source Complex Model (ISC3)
Chemical transformations are treated Reference
using exponential decay. Half-life is input by
the user. Environmental Protection Agency, 1995.
User’s Guide for the Industrial Source Com-
m. Physical Removal plex (ISC3) Dispersion Models, Volumes 1 and
2. EPA Publication Nos. EPA–454/B–95–003a &
Physical removal is not explicitly treated. b. Environmental Protection Agency, Re-
search Triangle Park, NC. (NTIS Nos. PB 95–
n. Evaluation Studies 222741 and PB 95–222758, respectively)
Ellis, H., P. Lou, and G. Dalzell, 1980. Com-
Availability
parison Study of Measured and Predicted
Concentrations with the RAM Model at Two The model code is available on the Support
Power Plants Along Lake Erie. Second Joint Center for Regulatory Air Models Bulletin
Conference on Applications of Air Pollution Board System. ISCST3 (as PB 96–502000) and
Meteorology, New Orleans, LA. ISCLT3 (PB 96–502018) are also available on
Environmental Research and Technology, diskette from the National Technical Infor-
mation Service (see section A.0).
1980. SO2 Monitoring and RAM (Urban) Model
Comparison Study in Summit County, Ohio. Abstract
Document P–3618–152, Environmental Re-
search & Technology, Inc., Concord, MA. The ISC3 model is a steady-state Gaussian
Guldberg, P.H. and C.W. Kern, 1978. A Com- plume model which can be used to assess pol-
lutant concentrations from a wide variety of
parison Validation of the RAM and PTMTP
sources associated with an industrial source
Models for Short-Term Concentrations in
complex. This model can account for the fol-
Two Urban Areas. Journal of Air Pollution lowing: settling and dry deposition of par-
Control Association, 28: 907–910. ticles; downwash; area, line and volume
Hodanbosi, R.R. and L.K. Peters, 1981. sources; plume rise as a function of down-
Evaluation of RAM Model for Cleveland, wind distance; separation of point sources;
Ohio. Journal of Air Pollution Control Asso- and limited terrain adjustment. ISC3 oper-
ciation, 31: 253–255. ates in both long-term and short-term
Kennedy, K.H., R.D. Siegel and M.P. Stein- modes.
berg, 1981. Case-Specific Evaluation of the
RAM Atmospheric Dispersion Model in an a. Recommendations for Regulatory Use
Urban Area. 74th Annual Meeting of the ISC3 is appropriate for the following appli-
American Institute of Chemical Engineers, cations:
New Orleans, LA. • Industrial source complexes;
Kummier, R.H., B. Cho, G. Roginski, R. • Rural or urban areas;
Sinha and A. Greenburg, 1979. A Comparative • Flat or rolling terrain;
Validation of the RAM and Modified SAI • Transport distances less than 50 kilo-
Models for Short Term SO2 Concentrations meters;
in Detroit. Journal of Air Pollution Control • 1-hour to annual averaging times; and
Association, 29: 720–723. • Continuous toxic air emissions.
Londergan, R.J., N.E. Bowne, D.R. Murray, The following options should be selected
H. Borenstein and J. Mangano, 1980. An Eval- for regulatory applications: For short term
uation of Short-Term Air Quality Models or long term modeling, set the regulatory
Using Tracer Study Data. Report No. 4333, ‘‘default option’’; i.e., use the keyword
DFAULT, which automatically selects stack
American Petroleum Institute, Washington,
tip downwash, final plume rise, buoyancy in-
D.C.
duced dispersion (BID), the vertical potential
Londergan, R., D. Minott, D. Wackter and temperature gradient, a treatment for calms,
R. Fizz, 1983. Evaluation of Urban Air Qual- the appropriate wind profile exponents, the
ity Simulation Models. EPA Publication No. appropriate value for pollutant half-life, and
EPA–450/4–83–020. U.S. Environmental Pro- a revised building wake effects algorithm;
tection Agency, Research Triangle Park, NC. set the ‘‘rural option’’ (use the keyword
Morgenstern, P., M.J. Geraghty, and A. RURAL) or ‘‘urban option’’ (use the keyword
McKnight, 1979. A Comparative Study of the URBAN); and set the ‘‘concentration option’’
RAM (Urban) and RAMR (Rural) Models for (use the keyword CONC).
Short-term SO2 Concentrations in Metropoli-
tan Indianapolis. 72nd Annual Meeting of the b. Input Requirements
Air Pollution Control Association, Cin- Source data: location, emission rate, phys-
cinnati, OH. ical stack height, stack gas exit velocity,
Ruff, R.E., 1980. Evaluation of the RAM stack inside diameter, and stack gas tem-
Using the RAPS Data Base. Contract 68–02– perature. Optional inputs include source ele-
2770, SRI International, Menlo Park, CA. vation, building dimensions, particle size

396
Environmental Protection Agency Pt. 51, App. W
distribution with corresponding settling ve- g. Plume Behavior
locities, and surface reflection coefficients.
ISC3 uses Briggs (1969, 1971, 1975) plume rise
Meteorological data: ISCST3 requires equations for final rise.
hourly surface weather data from the Stack tip downwash equation from Briggs
preprocessor program RAMMET, which pro- (1974) is used.
vides hourly stability class, wind direction, Revised building wake effects algorithm is
wind speed, temperature, and mixing height. used. For stacks higher than building height
For ISCLT3, input includes stability wind plus one-half the lesser of the building
rose (STAR deck), average afternoon mixing height or building width, the building wake
height, average morning mixing height, and algorithm of Huber and Snyder (1976) is used.
average air temperature. For lower stacks, the building wake algo-
Receptor data: coordinates and optional rithm of Schulman and Scire (Schulman and
ground elevation for each receptor. Hanna, 1986) is used, but stack tip downwash
and BID are not used.
c. Output For rolling terrain (terrain not above
stack height), plume centerline is horizontal
Printed output options include:
at height of final rise above source.
• Program control parameters, source Fumigation is not treated.
data, and receptor data;
• Tables of hourly meteorological data for h. Horizontal Winds
each specified day;
Constant, uniform (steady-state) wind is
• ‘‘N’’-day average concentration or total assumed for each hour.
deposition calculated at each receptor for Straight line plume transport is assumed
any desired source combinations; to all downwind distances.
• Concentration or deposition values cal- Separate wind speed profile exponents
culated for any desired source combinations (Irwin, 1979; EPA, 1980) for both rural and
at all receptors for any specified day or time urban cases are used.
period within the day; An optional treatment for calm winds is
• Tables of highest and second highest con- included for short term modeling.
centration or deposition values calculated at
each receptor for each specified time period i. Vertical Wind Speed
during a(n) ‘‘N’’-day period for any desired Vertical wind speed is assumed equal to
source combinations, and tables of the maxi- zero.
mum 50 concentration or deposition values
calculated for any desired source combina- j. Horizontal Dispersion
tions for each specified time period.
Rural dispersion coefficients from Turner
d. Type of Model (1969) are used, with no adjustments for sur-
face roughness or averaging time.
ISC3 is a Gaussian plume model. It has Urban dispersion coefficients from Briggs
been revised to perform a double integration (Gifford, 1976) are used.
of the Gaussian plume kernel for area Buoyancy induced dispersion (Pasquill,
sources. 1976) is included.
Six stability classes are used.
e. Pollutant Types
k. Vertical Dispersion
ISC3 may be used to model primary pollut-
ants and continuous releases of toxic and Rural dispersion coefficients from Turner
hazardous waste pollutants. Settling and (1969) are used, with no adjustments for sur-
deposition are treated. face roughness.
Urban dispersion coefficients from Briggs
f. Source-Receptor Relationships (Gifford, 1976) are used.
Buoyancy induced dispersion (Pasquill,
ISC3 applies user-specified locations for 1976) is included.
point, line, area and volume sources, and Six stability classes are used.
user-specified receptor locations or receptor Mixing height is accounted for with mul-
rings. tiple reflections until the vertical plume
User input topographic evaluation for each standard deviation equals 1.6 times the mix-
receptor is used. Elevations above stack top ing height; uniform vertical mixing is as-
are reduced to the stack top elevation, i.e., sumed beyond that point.
‘‘terrain chopping’’. Perfect reflection is assumed at the
User input height above ground level may ground.
be used when necessary to simulate impact
at elevated or ‘‘flag pole’’ receptors, e.g., on l. Chemical Transformation
buildings. Chemical transformations are treated
Actual separation between each source-re- using exponential decay. Time constant is
ceptor pair is used. input by the user.

397
Pt. 51, App. W 40 CFR Ch. I (7–1–97 Edition)
m. Physical Removal A.6 Urban Airshed Model (UAM)
Dry deposition effects for particles are Reference
treated using a resistance formulation in
which the deposition velocity is the sum of Environmental Protection Agency, 1990.
User’s Guide for the Urban Airshed Model,
the resistances to pollutant transfer within
Volume I–VIII. EPA Publication Nos. EPA–
the surface layer of the atmosphere, plus a
450/4–90–007a–c, d(R), e-g, and EPA–454/B–93–
gravitational settling term (EPA, 1994),
004, respectively. U.S. Environmental Pro-
based on the modified surface depletion
tection Agency, Research Triangle Park, NC
scheme of Horst (1983). (NTIS Nos. PB 91–131227, PB 91–131235, PB 91–
n. Evaluation Studies 131243, PB 93–122380, PB 91–131268, PB 92–
145382, and PB 92–224849, respectively, for
Bowers, J.F. and A.J. Anderson, 1981. An Vols. I–VII).
Evaluation Study for the Industrial Source
Complex (ISC) Dispersion Model, EPA Publi- Availability
cation No. EPA–450/4–81–002. U.S. Environ- The model code is available on the Support
mental Protection Agency, Research Tri- Center for Regulatory Air Models Bulletin
angle Park, NC. Board System (see section A.0).
Bowers, J.F., A.J. Anderson and W.R.
Hargraves, 1982. Tests of the Industrial Abstract
Source Complex (ISC) Dispersion Model at UAM is an urban scale, three dimensional,
the Armco Middletown, Ohio Steel Mill. EPA grid type numerical simulation model. The
Publication No. EPA–450/4–82–006. U.S. Envi- model incorporates a condensed photo-
ronmental Protection Agency, Research Tri- chemical kinetics mechanism for urban
angle Park, NC. atmospheres. The UAM is designed for com-
Environmental Protection Agency, 1992. puting ozone (O3) concentrations under
Comparison of a Revised Area Source Algo- short-term, episodic conditions lasting one
rithm for the Industrial Source Complex or two days resulting from emissions of ox-
Short Term Model and Wind Tunnel Data. ides of nitrogen (NOX), volatile organic com-
EPA Publication No. EPA–454/R–92–014. U.S. pounds (VOC), and carbon monoxide (CO).
Environmental Protection Agency, Research The model treats urban VOC emissions as
Triangle Park, NC. (NTIS No. PB 93–226751) their carbon-bond surrogates.
Environmental Protection Agency, 1992. a. Recommendations for Regulatory Use
Sensitivity Analysis of a Revised Area
Source Algorithm for the Industrial Source UAM is appropriate for the following appli-
Complex Short Term Model. EPA Publica- cations: urban areas having significant ozone
tion No. EPA–454/R–92–015. U.S. Environ- attainment problems and one hour averaging
mental Protection Agency, Research Tri- times.
angle Park, NC. (NTIS No. PB 93–226769) UAM has many options but no specific rec-
Environmental Protection Agency, 1992. ommendations can be made at this time on
all options. The reviewing agency should be
Development and Evaluation of a Revised
consulted on selection of options to be used
Area Source Algorithm for the Industrial
in regulatory applications.
source complex Long Term Model. EPA Pub-
lication No. EPA–454/R–92–016. U.S. Environ- b. Input Requirements
mental Protection Agency, Research Tri-
angle Park, NC. (NTIS No. PB 93–226777) Source data: gridded, hourly emissions of
Environmental Protection Agency, 1994. PAR, OLE, ETH, XYL, TOL, ALD2, FORM,
ISOR, ETOTH, MEOH, CO, NO, and NO2 for
Development and Testing of a Dry Deposi-
low-level sources. For major elevated point
tion Algorithm (Revised). EPA Publication
sources, hourly emissions, stack height,
No. EPA–454/R–94–015. U.S. Environmental
stack diameter, exit velocity, and exit tem-
Protection Agency, Research Triangle Park,
perature.
NC. (NTIS No. PB 94–183100)
Meteorological data: hourly, gridded, di-
Scire, J.S. and L.L. Schulman, 1981. Eval- vergence free, u and v wind components for
uation of the BLP and ISC Models with SF6 each vertical level; hourly gridded mixing
Tracer Data and SO2 Measurements at Alu- heights and surface temperatures; hourly ex-
minum Reduction Plants. Air Pollution Con- posure class; hourly vertical potential tem-
trol Association Specialty Conference on perature gradient above and below the mix-
Dispersion Modeling for Complex Sources, ing height; hourly surface atmospheric pres-
St. Louis, MO. sure; hourly water mixing ratio; and gridded
Schulman, L.L. and S.R. Hanna, 1986. Eval- surface roughness lengths.
uation of Downwash Modification to the In- Air quality data: concentration of all car-
dustrial Source Complex Model. Journal of bon bond 4 species at the beginning of the
the Air Pollution Control Association, 36: simulation for each grid cell; and hourly con-
258–264. centrations of each pollutant at each level

398
Environmental Protection Agency Pt. 51, App. W
along the inflow boundaries and top bound- l. Chemical Transformation
ary of the modeling region. UAM employs a simplified version of the
Other data requirements are: hourly mixed Carbon-Bond IV Mechanism (CBM–IV) devel-
layer average, NO2 photolysis rates; and oped by Gery et al. (1988) employing various
ozone surface uptake resistance along with steady state approximations. The CBM–IV
associated gridded vegetation (scaling) fac- mechanism incorporated in UAM utilizes an
tors. updated simulation of PAN chemistry that
includes a peroxy-peroxy radical termination
c. Output reaction, significant when the atmosphere is
Printed output includes: NOX-limited (Gery et al., 1989). The current
CBM–IV mechanism accommodates 34 spe-
• Gridded instantaneous concentration
cies and 82 reactions.
fields at user-specified time intervals for
user-specified pollutants and grid levels; m. Physical Removal
• Gridded time-average concentration
Dry deposition of ozone and other pollut-
fields for user-specified time intervals, pol-
ant species are calculated. Vegetation (scal-
lutants, and grid levels.
ing) factors are applied to the reference sur-
d. Type of Model face uptake resistance of each species de-
pending on land use type.
UAM is a three dimensional, numerical,
photochemical grid model. n. Evaluation Studies
Builtjes, P.J.H., K.D. van der Hurt and S.D.
e. Pollutant Types Reynolds, 1982. Evaluation of the Perform-
UAM may be used to model ozone (O3) for- ance of a Photochemical Dispersion Model in
mation from oxides of nitrogen (NOX) and Practical Applications. 13th International
Technical Meeting on Air Pollution Model-
volatile organic compound (VOC) emissions.
ing and Its Application, Ile des Embiez,
f. Source-Receptor Relationship France.
Cole, H.S., D.E. Layland, G.K. Moss and
Low-level area and point source emissions C.F. Newberry, 1983. The St. Louis Ozone
are specified within each surface grid cell. Modeling Project. EPA Publication No.
Emissions from major point sources are EPA–450/4–83–019. U.S. Environmental Pro-
placed within cells aloft in accordance with tection Agency, Research Triangle Park, NC.
calculated effective plume heights. Dennis, R.L., M.W. Downton and R.S. Keil,
Hourly average concentrations of each pol- 1983. Evaluation of Performance Measures
lutant are calculated for all grid cells at for an Urban Photochemical Model. EPA
each vertical level. Publication No. EPA–450/4–83–021. U.S. Envi-
ronmental Protection Agency, Research Tri-
g. Plume Behavior angle Park, NC.
Haney, J.L. and T.N. Braverman, 1985.
Plume rise is calculated for major point Evaluation and Application of the Urban
sources using relationships recommended by Airshed Model in the Philadelphia Air Qual-
Briggs (1971). ity Control Region. EPA Publication No.
EPA–450/4–85–003. U.S. Environmental Pro-
h. Horizontal Winds tection Agency, Research Triangle Park, NC.
See Input Requirements. Layland, D.E. and H.S. Cole, 1983. A Review
of Recent Applications of the SAI Urban
i. Vertical Wind Speed Airshed Model. EPA Publication No. EPA–
450/4–84–004. U.S. Environmental Protection
Calculated at each vertical grid cell inter- Agency, Research Triangle Park, NC.
face from the mass continuity relationship Layland, D.E., S.D. Reynolds, H. Hogo and
using the input gridded horizontal wind field. W.R. Oliver, 1983. Demonstration of Photo-
chemical Grid Model Usage for Ozone Con-
j. Horizontal Dispersion trol Assessment. 76th Annual Meeting of the
Horizontal eddy diffusivity is set to a user Air Pollution Control Association, Atlanta,
specified constant value (nominally 50 m2/s). GA.
Morris, R.E. et al., 1990. Urban Airshed
k. Vertical Dispersion Model Study of Five Cities. EPA Publication
No. EPA–450/4–90–006a-g. U.S. Environmental
Vertical eddy diffusivities for unstable and Protection Agency, Research Triangle Park,
neutral conditions calculated using relation- NC.
ships of Lamb et al. (1977); for stable condi- Reynolds, S.D., H. Hogo, W.R. Oliver and
tions, the relationship of Businger and Arya L.E. Reid, 1982. Application of the SAI
(1974) is employed. Stability class, friction Airshed Model to the Tulsa Metropolitan
velocity, and Monin-Obukhov length deter- Area, SAI No. 82004. Systems Applications,
mined using procedure of Liu et al. (1976). Inc., San Rafael, CA.

399
Pt. 51, App. W 40 CFR Ch. I (7–1–97 Edition)
Schere, K.L. and J.H. Shreffler, 1982. Final letin Board System and also on diskette (as
Evaluation of Urban-Scale Photochemical PB 91–505230) from the National Technical
Air Quality Simulation Models. EPA Publi- Information Service (see section A.0).
cation No. EPA–600/3–82–094. U.S. Environ-
mental Protection Agency, Research Tri- Technical Contact
angle Park, NC. Minerals Management Service, Attn: Mr.
Seigneur C., T.W. Tesche, C.E. Reid, P.M. Dirk Herkhof, Parkway Atrium Building, 381
Roth, W.R. Oliver and J.C. Cassmassi, 1981. Elden Street, Herndon, VA 22070–4817, Phone:
The Sensitivity of Complex Photochemical
(703) 787–1735.
Model Estimates to Detail In Input Informa-
tion, Appendix A—A Compilation of Simula- Abstract
tion Results. EPA Publication No. EPA–450/
4–81–031b. U.S. Environmental Protection OCD is a straight-line Gaussian model de-
Agency, Research Triangle Park, NC. veloped to determine the impact of offshore
South Coast Air Quality Management Dis- emissions from point, area or line sources on
trict, 1989. Air Quality Management Plan— the air quality of coastal regions. OCD incor-
Appendix V–R (Urban Airshed Model Per- porates overwater plume transport and dis-
formance Evaluation). El Monte, CA. persion as well as changes that occur as the
Stern, R. and B. Scherer, 1982. Simulation plume crosses the shoreline. Hourly meteoro-
of a Photochemical Smog Episode in the logical data are needed from both offshore
Rhine-Ruhr Area with a Three Dimensional and onshore locations. These include water
Grid Model. 13th International Technical surface temperature, overwater air tempera-
Meeting on Air Pollution Modeling and Its ture, mixing height, and relative humidity.
Application, Ile des Embiez, France. Some of the key features include platform
Tesche, T.W., C. Seigneur, L.E. Reid, P.M. building downwash, partial plume penetra-
Roth, W.R. Oliver and J.C. Cassmassi, 1981. tion into elevated inversions, direct use of
The Sensitivity of Complex Photochemical turbulence intensities for plume dispersion,
Model Estimates to Detail in Input Informa- interaction with the overland internal
tion. EPA Publication No. EPA–450/4–81–031a. boundary layer, and continuous shoreline fu-
U.S. Environmental Protection Agency, Re- migation.
search Triangle Park, NC.
Tesche, T.W., W.R. Oliver, H. Hogo, P. a. Recommendations for Regulatory Use
Saxeena and J.L. Haney, 1983. Volume IV—
OCD has been recommended for use by the
Assessment of NOX Emission Control Re-
Minerals Management Service for emissions
quirements in the South Coast Air Basin—
located on the Outer Continental Shelf (50
Appendix A. Performance Evaluation of the
FR 12248; 28 March 1985). OCD is applicable
Systems Applications Airshed Model for the
for overwater sources where onshore recep-
26–27 June 1974 O3 Episode in the South Coast
tors are below the lowest source height.
Air Basin, SYSAPP 83/037. Systems Applica-
Where onshore receptors are above the low-
tions, Inc., San Rafael, CA.
Tesche, T.W., W.R. Oliver, H. Hogo, P. est source height, offshore plume transport
Saxeena and J.L. Haney, 1983. Volume IV— and dispersion may be modeled on a case-by-
Assessment of NOX Emission Control Re- case basis in consultation with the EPA Re-
quirements in the South Coast Air Basin— gional Office.
Appendix B. Performance Evaluation of the b. Input Requirements
Systems Applications Airshed Model for the
7–8 November 1978 NO2 Episode in the South Source data: point, area or line source lo-
Coast Air Basin, SYSAPP 83/038. Systems cation, pollutant emission rate, building
Applications, Inc., San Rafael, CA. height, stack height, stack gas temperature,
Tesche, T.W., 1988. Accuracy of Ozone Air stack inside diameter, stack gas exit veloc-
Quality Models. Journal of Environmental ity, stack angle from vertical, elevation of
Engineering, 114(4): 739–752. stack base above water surface and gridded
specification of the land/water surfaces. As
A.7 Offshore and Coastal Dispersion Model an option, emission rate, stack gas exit ve-
(OCD) locity and temperature can be varied hourly.
Reference Meteorological data (over water): wind di-
rection, wind speed, mixing height, relative
DiCristofaro, D.C. and S.R. Hanna, 1989. humidity, air temperature, water surface
OCD: The Offshore and Coastal Dispersion temperature, vertical wind direction shear
Model, Version 4. Volume I: User’s Guide, (optional), vertical temperature gradient
and Volume II: Appendices. Sigma Research (optional), turbulence intensities (optional).
Corporation, Westford, MA. (NTIS Nos. PB Meteorological data (over land): wind di-
93–144384 and PB 93–144392) rection, wind speed, temperature, stability
class, mixing height.
Availability
Receptor data: location, height above local
This model code is available on the Sup- ground-level, ground-level elevation above
port Center for Regulatory Air Models Bul- the water surface.

400
Environmental Protection Agency Pt. 51, App. W
c. Output Overwater wind speed can be estimated
from overland wind speed using relationship
All input options, specification of sources, of Hsu (1981).
receptors and land/Water map including lo-
Wind speed profiles are estimated using
cations of sources and receptors.
similarity theory (Businger, 1973). Surface
Summary tables of five highest concentra- layer fluxes for these formulas are cal-
tions at each receptor for each averaging pe- culated from bulk aerodynamic methods.
riod, and average concentration for entire
run period at each receptor. i. Vertical Wind Speed
Optional case study printout with hourly
plume and receptor characteristics. Optional Vertical wind speed is assumed equal to
table of annual impact assessment from non- zero.
permanent activities. j. Horizontal Dispersion
Concentration files written to disk or tape
can be used by ANALYSIS postprocessor to Lateral turbulence intensity is rec-
produce the highest concentrations for each ommended as a direct estimate of horizontal
receptor, the cumulative frequency distribu- dispersion. If lateral turbulence intensity is
tions for each receptor, the tabulation of all not available, it is estimated from boundary
concentrations exceeding a given threshold, layer theory. For wind speeds less than 8 m/
and the manipulation of hourly concentra- s, lateral turbulence intensity is assumed in-
tion files. versely proportional to wind speed.
Horizontal dispersion may be enhanced be-
d. Type of Model cause of obstructions near the source. A vir-
tual source technique is used to simulate the
OCD is a Gaussian plume model con- initial plume dilution due to downwash.
structed on the framework of the MPTER Formulas recommended by Pasquill (1976)
model. are used to calculate buoyant plume en-
e. Pollutant Types hancement and wind direction shear en-
hancement.
OCD may be used to model primary pollut- At the water/land interface, the change to
ants. Settling and deposition are not treated. overland dispersion rates is modeled using a
virtual source. The overland dispersion rates
f. Source-Receptor Relationship can be calculated from either lateral turbu-
Up to 250 point sources, 5 area sources, or lence intensity or Pasquill-Gifford curves.
1 line source and 180 receptors may be used. The change is implemented where the plume
intercepts the rising internal boundary
Receptors and sources are allowed at any
layer.
location.
The coastal configuration is determined by k. Vertical Dispersion
a grid of up to 3600 rectangles. Each element
of the grid is designated as either land or Observed vertical turbulence intensity is
water to identify the coastline. not recommended as a direct estimate of ver-
tical dispersion. Turbulence intensity should
g. Plume Behavior be estimated from boundary layer theory as
default in the model. For very stable condi-
As in MPTER, the basic plume rise algo- tions, vertical dispersion is also a function of
rithms are based on Briggs’ recommenda- lapse rate.
tions. Vertical dispersion may be enhanced be-
Momentum rise includes consideration of cause of obstructions near the source. A vir-
the stack angle from the vertical. tual source technique is used to simulate the
The effect of drilling platforms, ships, or initial plume dilution due to downwash.
any overwater obstructions near the source Formulas recommended by Pasquill (1976)
are used to decrease plume rise using a re- are used to calculate buoyant plume en-
vised platform downwash algorithm based on hancement.
laboratory experiments. At the water/land interface, the change to
Partial plume penetration of elevated in- overland dispersion rates is modeled using a
versions is included using the suggestions of virtual source. The overland dispersion rates
Briggs (1975) and Weil and Brower (1984). can be calculated from either vertical turbu-
Continuous shoreline fumigation is lence intensity or the Pasquill-Gifford coeffi-
parametrized using the Turner method where cients. The change is implemented where the
complete vertical mixing through the ther- plume intercepts the rising internal bound-
mal internal boundary layer (TIBL) occurs ary layer.
as soon as the plume intercepts the TIBL.
l. Chemical Transformation
h. Horizontal Winds
Chemical transformations are treated
Constant, uniform wind is assumed for using exponential decay. Different rates can
each hour. be specified by month and by day or night.

401
Pt. 51, App. W 40 CFR Ch. I (7–1–97 Edition)
m. Physical Removal United States Air Force (USAF), produces an
emission inventory of all airport sources and
Physical removal is also treated using ex-
calculates concentrations produced by these
ponential decay.
sources at specified receptors. The system
n. Evaluation Studies stores emission factors for fixed sources such
as fuel storage tanks and incinerators and
DiCristofaro, D.C. and S.R. Hanna, 1989. also for mobile sources such as automobiles
OCD: The Offshore and Coastal Dispersion or aircraft. EDMS incorporates an emissions
Model. Volume I: User’s Guide. Sigma Re- model to calculate an emission inventory for
search Corporation, Westford, MA. each airport source and a dispersion model,
Hanna, S.R., L.L. Schulman, R.J. Paine the Graphical Input Microcomputer Model
and J.E. Pleim, 1984. The Offshore and Coast- (GIMM) (Segal, 1983) to calculate pollutant
al Dispersion (OCD) Model User’s Guide, Re- concentrations produced by these sources at
vised. OCS Study, MMS 84–0069. Environ- specified receptors. The GIMM, which proc-
mental Research & Technology, Inc., Con- esses point, area, and line sources, also in-
cord, MA. (NTIS No. PB 86–159803) corporates a special meteorological
Hanna, S.R., L.L. Schulman, R.J. Paine, preprocessor for processing up to one year of
J.E. Pleim and M. Baer, 1985. Development National Climatic Data Center (NCDC) hour-
and Evaluation of the Offshore and Coastal ly data. The model operates in both a screen-
Dispersion (OCD) Model. Journal of the Air ing and refined mode, accepting up to 170
Pollution Control Association, 35: 1039–1047. sources and 10 receptors.
Hanna, S.R. and D.C. DiCristofaro, 1988.
Development and Evaluation of the OCD/API a. Recommendations for Regulatory Use
Model. Final Report, API Pub. 4461, Amer-
EDMS is appropriate for the following ap-
ican Petroleum Institute, Washington, D.C.
plications:
A.8 Emissions and Dispersion Modeling System • Cumulative effect of changes in aircraft
(EDMS) operations, point source and mobile source
emissions at airports or air bases;
Reference • Simple terrain;
Segal, H.M., 1991. ‘‘EDMS—Microcomputer • Transport distances less than 50 kilo-
Pollution Model for Civilian Airports and Air meters; and
Force Bases: User’s Guide.’’ FAA Report No. • 1-hour to annual averaging times.
FAA–EE–91–3; USAF Report No. ESL–TR–91– b. Input Requirements
31, Federal Aviation Administration, 800
Independence Avenue, S.W., Washington, All data are entered through a ‘‘runtime’’
D.C. 20591. (NTIS No. ADA 240528) version of the Condor data base which is an
Segal, H.M. and Hamilton, P.L., 1988. ‘‘A integral part of EDMS. Typical entry items
Microcomputer Pollution Model for Civilian are source and receptor coordinates, percent
Airports and Air Force Bases—Model De- cold starts, vehicles per hour, etc. Some
scription.’’ FAA Report No. FAA–EE–88–4; point sources, such as heating plants, require
USAF Report No. ESL–TR–88–53, Federal stack height, stack diameter, and effluent
Aviation Administration, 800 Independence temperature inputs.
Avenue, S.W., Washington, D.C. 20591. (NTIS Wind speed, wind direction, hourly tem-
No. ADA 199003) perature, and Pasquill-Gifford stability cat-
Segal, H.M., 1988. ‘‘A Microcomputer Pollu- egory (P–G) are the meteorological inputs.
tion Model for Civilian Airports and Air They can be entered manually through the
Force Bases—Model Application and Back- EDMS data entry screens or automatically
ground.’’ FAA Report No. FAA–EE–88–5; through the processing of previously loaded
USAF Report No. ESL–TR–88–55, Federal NCDC hourly data.
Aviation Administration, 800 Independence
Avenue, S.W., Washington, D.C. 20591. (NTIS c. Output
No. ADA 199794) Printed outputs consist of:
• A monthly and yearly emission inven-
Availability
tory report for each source entered; and
EDMS is available for $40 from: Federal • A concentration summing report for up
Aviation Administration, Attn: Ms. Diana to 8760 hours (one year) of data.
Liang, AEE–120, 800 Independence Avenue,
S.W., Washington, D.C. 20591, Phone: (202) d. Type of Model
267–3494. For its emissions inventory calculations,
EDMS uses algorithms consistent with the
Abstract
EPA Compilation of Air Pollutant Emission
EDMS is a combined emissions/dispersion Factors, AP–42. For its dispersion calcula-
model for assessing pollution at civilian air- tions, EDMS uses the GIMM model which is
ports and military air bases. This model, described in reports FAA–EE–88–4 and FAA–
which was jointly developed by the Federal EE–88–5, referenced above. GIMM uses a
Aviation Administration (FAA) and the Gaussian plume algorithm.

402
Environmental Protection Agency Pt. 51, App. W
e. Pollutant Types l. Chemical Transformation
EDMS inventories and calculates the dis- Chemical transformations are not ac-
persion of carbon monoxide, nitrogen oxides, counted for.
sulphur oxides, hydrocarbons, and suspended
particles. m. Physical Removal

f. Source-Receptor Relationship Deposition is not treated.


Up to 170 sources and 10 receptors can be n. Evaluation Studies
treated simultaneously. Area sources are
treated as a series of lines that are posi- Segal, H.M. and P.L. Hamilton, 1988. A
tioned perpendicular to the wind. Microcomputer Pollution Model for Civilian
Line sources (roadways, runways) are mod- Airports and Air Force Bases—Model De-
eled as a series of points. Terrain elevation scription. FAA Report No. FAA–EE–88–4;
differences between sources and receptors USAF Report No. ESL–TR–88–53, Federal
are neglected. Aviation Administration, 800 Independence
Receptors are assumed to be at ground Avenue, S.W., Washington, D.C. 20591.
level. Segal, H.M., 1988. A Microcomputer Pollu-
tion Model for Civilian Airports and Air
g. Plume Behavior Force Bases—Model Application and Back-
Plume rise is calculated for all point ground. FAA Report No. FAA–EE–88–5;
sources (heating plants, incinerators, etc.) USAF Report No. ESL–TR–88–55, Federal
using Briggs plume rise equations (Catalano, Aviation Administration, 800 Independence
1986; Briggs, 1969; Briggs, 1971; Briggs, 1972). Avenue, S.W., Washington, D.C. 20591.
Building and stack tip downwash effects
A.9 Complex Terrain Dispersion Model Plus Al-
are not treated.
gorithms for Unstable Situations
Roadway dispersion employs a modifica-
(CTDMPLUS)
tion to the Gaussian plume algorithms as
suggested by Rao and Keenan (1980) to ac- Reference
count for close-in vehicle-induced turbu-
lence. Perry, S.G., D.J. Burns, L.H. Adams, R.J.
Paine, M.G. Dennis, M.T. Mills, D.G.
h. Horizontal Winds Strimaitis, R.J. Yamartino and E.M. Insley,
Steady state winds are assumed for each 1989. User’s Guide to the Complex Terrain
hour. Winds are assumed to be constant with Dispersion Model Plus Algorithms for Unsta-
altitude. ble Situations (CTDMPLUS). Volume 1:
Winds are entered manually by the user or Model Descriptions and User Instructions.
automatically by reading previously loaded EPA Publication No. EPA–600/8–89–041. Envi-
NCC annual data files. ronmental Protection Agency, Research Tri-
angle Park, NC. (NTIS No. PB 89–181–424)
i. Vertical Wind Speed Paine, R.J., D.G. Strimaitis, M.G. Dennis,
R.J. Yamartino, M.T. Mills and E.M. Insley,
Vertical wind speed is assumed to be zero.
1987. User’s Guide to the Complex Terrain
j. Horizontal Dispersion Dispersion Model, Volume 1. EPA Publica-
tion No. EPA–600/8–87–058a. U.S. Environ-
Four stability classes are used (P–G classes mental Protection Agency, Research Tri-
B through E). angle Park, NC. (NTIS No. PB 88–162169)
Horizontal dispersion coefficients are com-
puted using a table look-up and linear inter- Availability
polation scheme. Coefficients are based on
Pasquill (1976) as adapted by Petersen (1980). This model code is available on the Sup-
A modified coefficient table is used to ac- port Center for Regulatory Air Models Bul-
count for traffic-enhanced turbulence near letin Board System and also on diskette (as
roadways. Coefficients are based upon data PB 90–504119) from the National Technical
included in Rao and Keenan (1980). Information Service (see section A.0).

k. Vertical Dispersion Abstract


Four stability classes are used (P–G classes CTDMPLUS is a refined point source
B through E). Gaussian air quality model for use in all sta-
Vertical dispersion coefficients are com- bility conditions for complex terrain applica-
puted using a table look-up and linear inter- tions. The model contains, in its entirety,
polation scheme. Coefficients are based on the technology of CTDM for stable and neu-
Pasquill (1976) as adapted by Petersen (1980). tral conditions. However, CTDMPLUS can
A modified coefficient table is used to ac- also simulate daytime, unstable conditions,
count for traffic-enhanced turbulence near and has a number of additional capabilities
roadways. Coefficients are based upon data for improved user friendliness. Its use of me-
from Roa and Keenan (1980). teorological data and terrain information is

403
Pt. 51, App. W 40 CFR Ch. I (7–1–97 Edition)
different from other EPA models; consider- • Plume characteristics at each receptor,
able detail for both types of input data is re- i.e.,
quired and is supplied by preprocessors spe- ¥> distance in along-flow and cross flow
cifically designed for CTDMPLUS. direction
CTDMPLUS requires the parameterization of ¥> effective plume-receptor height dif-
individual hill shapes using the terrain ference
preprocessor and the association of each ¥> effective σy & σz values, both flat ter-
model receptor with a particular hill. rain and hill induced (the difference shows
the effect of the hill)
a. Recommendation for Regulatory Use ¥> concentration components due to
WRAP, LIFT and FLAT.
CTDMPLUS is appropriate for the follow-
If the user selects the TOPN option, a sum-
ing applications:
mary table of the top 4 concentrations at
• Elevated point sources; each receptor is given. If the ISOR option is
• Terrain elevations above stack top; selected, a source contribution table for
• Rural or urban areas; every hour will be printed.
• Transport distances less than 50 kilo- A separate disk file of predicted (1-hour
meters; and only) concentrations (‘‘CONC’’) is written if
• One hour to annual averaging times the user chooses this option. Three forms of
when used with a post-processor program output are possible:
such as CHAVG. (1) A binary file of concentrations, one
value for each receptor in the hourly se-
b. Input Requirements quence as run;
Source data: For each source, user supplies (2) A text file of concentrations, one value
source location, height, stack diameter, for each receptor in the hourly sequence as
stack exit velocity, stack exit temperature, run; or
and emission rate; if variable emissions are (3) A text file as described above, but with
appropriate, the user supplies hourly values a listing of receptor information (names, po-
for emission rate, stack exit velocity, and sitions, hill number) at the beginning of the
stack exit temperature. file.
Meteorological data: the user must supply Hourly information provided to these files
hourly averaged values of wind, temperature besides the concentrations themselves in-
and turbulence data for creation of the basic cludes the year, month, day, and hour infor-
meteorological data file (‘‘PROFILE’’). Me- mation as well as the receptor number with
teorological preprocessors then create a the highest concentration.
SURFACE data file (hourly values of mixed
d. Type of Model
layer heights, surface friction velocity,
Monin-Obukhov length and surface rough- CTDMPLUS is a refined steady-state, point
ness length) and a RAWINsonde data file source plume model for use in all stability
(upper air measurements of pressure, tem- conditions for complex terrain applications.
perature, wind direction, and wind speed).
Receptor data: receptor names (up to 400) e. Pollutant Types
and coordinates, and hill number (each re- CTDMPLUS may be used to model non-re-
ceptor must have a hill number assigned). active, primary pollutants.
Terrain data: user inputs digitized contour
information to the terrain preprocessor f. Source-Receptor Relationship
which creates the TERRAIN data file (for up Up to 40 point sources, 400 receptors and 25
to 25 hills). hills may be used. Receptors and sources are
allowed at any location. Hill slopes are as-
c. Output
sumed not to exceed 15°, so that the linear-
When CTDMPLUS is run, it produces a ized equation of motion for Boussinesq flow
concentration file, in either binary or text are applicable. Receptors upwind of the im-
format (user’s choice), and a list file contain- pingement point, or those associated with
ing a verification of model inputs, i.e., any of the hills in the modeling domain, re-
• Input meteorological data from ‘‘SUR- quire separate treatment.
FACE’’ and ‘‘PROFILE’’
• Stack data for each source g. Plume Behavior
• Terrain information As in CTDM, the basic plume rise algo-
• Receptor information rithms are based on Briggs’ (1975) rec-
• Source-receptor location (line printer ommendations.
map). A central feature of CTDMPLUS for neu-
In addition, if the case-study option is se- tral/stable conditions is its use of a critical
lected, the listing includes: dividing-streamline height (Hc) to separate
• Meteorological variables at plume height the flow in the vicinity of a hill into two sep-
• Geometrical relationships between the arate layers. The plume component in the
source and the hill upper layer has sufficient kinetic energy to

404
Environmental Protection Agency Pt. 51, App. W
pass over the top of the hill while stream- m. Physical Removal
lines in the lower portion are constrained to
Physical removal is not treated by
flow in a horizontal plane around the hill.
CTDMPLUS (complete reflection at the
Two separate components of CTDMPLUS
ground/hill surface is assumed).
compute ground-level concentrations result-
ing from plume material in each of these n. Evaluation Studies
flows.
The model calculates on an hourly (or ap- Burns, D.J., L.H. Adams and S.G. Perry,
1990. Testing and Evaluation of the
propriate steady averaging period) basis how
CTDMPLUS Dispersion Model: Daytime Con-
the plume trajectory (and, in stable/neutral
vective Conditions. Environmental Protec-
conditions, the shape) is deformed by each
tion Agency, Research Triangle Park, NC.
hill. Hourly profiles of wind and temperature
Paumier, J.O., S.G. Perry and D.J. Burns,
measurements are used by CTDMPLUS to
1990. An Analysis of CTDMPLUS Model Pre-
compute plume rise, plume penetration (a
dictions with the Lovett Power Plant Data
formulation is included to handle penetra- Base. Environmental Protection Agency, Re-
tion into elevated stable layers, based on search Triangle Park, NC.
Briggs (1984)), convective scaling parameters, Paumier, J.O., S.G. Perry and D.J. Burns,
the value of Hc, and the Froude number 1992. CTDMPLUS: A Dispersion Model for
above Hc. Sources near Complex Topography. Part II:
Performance Characteristics. Journal of Ap-
h. Horizontal Winds
plied Meteorology, 31(7): 646–660.
CTDMPLUS does not simulate calm mete-
orological conditions. Both scalar and vector A. REF References
wind speed observations can be read by the Benson, P.E., 1979. CALINE3—A Versatile
model. If vector wind speed is unavailable, it Dispersion Model for Predicting Air Pollu-
is calculated from the scalar wind speed. The tion Levels Near Highways and Arterial
assignment of wind speed (either vector or Streets. Interim Report, Report Number
scalar) at plume height is done by either: FHWA/CA/TL–79/23. Federal Highway Admin-
• Interpolating between observations istration, Washington, D.C.
above and below the plume height, or Briggs, G.A., 1969. Plume Rise. U.S. Atomic
• Extrapolating (within the surface layer) Energy Commission Critical Review Series,
from the nearest measurement height to the Oak Ridge National Laboratory, Oak Ridge,
plume height. TN. (NTIS No. TID–25075)
Briggs, G.A., 1971. Some Recent Analyses
i. Vertical Wind Speed of Plume Rise Observations. Proceedings of
the Second International Clean Air Congress,
Vertical flow is treated for the plume com- edited by H.M. Englund and W.T. Berry. Aca-
ponent above the critical dividing streamline demic Press, New York, NY.
height (Hc); see ‘‘Plume Behavior’’. Briggs, G.A., 1974. Diffusion Estimation for
j. Horizontal Dispersion Small Emissions. USAEC Report ATDL–106.
U.S. Atomic Energy Commission, Oak Ridge,
Horizontal dispersion for stable/neutral TN.
conditions is related to the turbulence veloc- Briggs, G.A., 1975. Plume Rise Predictions.
ity scale for lateral fluctuations, σv, for Lectures on Air Pollution and Environ-
which a minimum value of 0.2 m/s is used. mental Impact Analyses. American Meteoro-
Convective scaling formulations are used to logical Society, Boston, MA, pp. 59–111.
estimate horizontal dispersion for unstable Bjorklund, J.R. and J.F. Bowers, 1982.
conditions. User’s Instructions for the SHORTZ and
LONGZ Computer Programs. EPA Publica-
k. Vertical Dispersion tion No. EPA–903/9–82–004a and b. U.S. Envi-
ronmental Protection Agency, Region III,
Direct estimates of vertical dispersion for
Philadelphia, PA.
stable/neutral conditions are based on ob-
Businger, J.A., 1973. Turbulence Transfer
served vertical turbulence intensity, e.g., σw
in the Atmospheric Surface Layer. Workshop
(standard deviation of the vertical velocity
in Micrometeorology. American Meteoro-
fluctuation). In simulating unstable (convec-
logical Society, Boston, MA, pp. 67–100.
tive) conditions, CTDMPLUS relies on a
Businger, J.A. and S.P. Arya, 1974. Height
skewed, bi-Gaussian probability density
of the Mixed Layer in the Stably Stratified
function (PDF) description of the vertical
Planetary Boundary Layer. Advances in Geo-
velocities to estimate the vertical distribu-
physics, Vol. 18A, F.N. Frankiel and R.E.
tion of pollutant concentration.
Munn (Eds.), Academic Press, New York, NY.
l. Chemical Transformation Catalano, J.A., 1986. Addendum to the
User’s Manual for the Single Source
Chemical transformation is not treated by (CRSTER) Model. EPA Publication No. EPA–
CTDMPLUS. 600/8–86–041. U.S. Environmental Protection

405
Pt. 51, App. W 40 CFR Ch. I (7–1–97 Edition)
Agency, Research Triangle Park, NC. (NTIS mental Protection Agency, Research Tri-
No. PB 87–145843) angle Park, NC.
Environmental Protection Agency, 1980. Petersen, W.B., 1980. User’s Guide for
Recommendations on Modeling (October 1980 HIWAY–2 A Highway Air Pollution Model.
Meetings). Appendix G to: Summary of Com- EPA Publication No. EPA–600/8–80–018. U.S.
ments and Responses on the October 1980 Environmental Protection Agency, Research
Proposed Revisions to the Guideline on Air Triangle Park, NC. (NTIS PB 80–227556)
Quality Models. Meteorology and Assess- Rao, T.R. and M.T. Keenan, 1980. Sugges-
ment Division, Office of Research and Devel- tions for Improvement of the EPA–HIWAY
opment, Research Triangle Park, NC. Model. Journal of the Air Pollution Control
Gery, M.W., G.Z. Whitten and J.P. Killus, Association, 30: 247–256 (and reprinted as Ap-
1988. Development and Testing of CBM–IV for pendix C in Petersen, 1980).
Urban and Regional Modeling. EPA Publica- Segal, H.M., 1983. Microcomputer Graphics
tion No. EPA–600/3–88–012. U.S. Environ- in Atmospheric Dispersion Modeling. Jour-
mental Protection Agency, Research Tri- nal of the Air Pollution Control Association,
angle Park, NC. (NTIS No. PB 88–180039) 23: 598–600.
Gery, M.W., G.Z. Whitten, J.P. Killus and Turner, D.B., 1969. Workbook of Atmos-
M.C. Dodge, 1989. A Photochemical Kinetics pheric Dispersion Estimates. PHS Publica-
Mechanism for Urban and Regional Scale tion No. 999–26. U.S. Environmental Protec-
Computer Modeling. Journal of Geophysical tion Agency, Research Triangle, Park, NC.
Research, 94: 12,925–12,956. Weil, J.C. and R.P. Brower, 1984. An Up-
Gifford, F.A., Jr. 1976. Turbulent Diffusion dated Gaussian Plume Model for Tall Stacks.
Typing Schemes—A Review. Nuclear Safety, Journal of the Air Pollution Control Asso-
17: 68–86. ciation, 34: 818–827.
Horst, T.W., 1983. A Correction to the
Gaussian Source-depletion Model. In Precipi- APPENDIX B TO APPENDIX W OF PART
tation Scavenging, Dry Deposition and Re- 51—SUMMARIES OF ALTERNATIVE AIR
suspension. H. R. Pruppacher, R.G. Semonin QUALITY MODELS
and W.G.N. Slinn, eds., Elsevier, NY.
Hsu, S.A., 1981. Models for Estimating Off- Table of Contents
shore Winds from Onshore Meteorological B.0 Introduction and Availability
Measurements. Boundary Layer Meteor- B.1 AVACTA II Model
ology, 20: 341–352. B.2 Dense Gas Dispersion Model (DEGADIS)
Huber, A.H. and W.H. Snyder, 1976. Build- B.3 ERT Visibility Model
ing Wake Effects on Short Stack Effluents. B.4 HGSYSTEM
Third Symposium on Atmospheric Turbu- B.5 HOTMAC/RAPTAD
lence, Diffusion and Air Quality, American B.6 LONGZ
Meteorological Society, Boston, MA. B.7 Maryland Power Plant Siting Program
Irwin, J.S., 1979. A Theoretical Variation (PPSP) Model
of the Wind Profile Power-Law Exponent as B.8 Mesoscale Puff Model (MESOPUFF II)
a Function of Surface Roughness and Stabil- B.9 Mesoscale Transport Diffusion and Depo-
ity. Atmospheric Environment, 13: 191–194. sition Model For Industrial Sources
Lamb, R.G. et al., 1977. Continued Research (MTDDIS)
in Mesoscale Air Pollution Simulation Mod- B.10 Multi-Source (SCSTER) Model
eling—Vol. VI: Further Studies in the Model- B.11 PANACHE
ing of Microscale Phenomena, Report Num- B.12 PLUME Visibility Model (PLUVUE II)
ber EF77–143. Systems Applications, Inc., B.13 Point, Area, Line Source Algorithm
San Rafael, CA. (PAL–DS)
Liu, M.K. et al., 1976. The Chemistry, Dis- B.14 Reactive Plume Model (RPM–IV)
persion, and Transport of Air Pollutants B.15 Shoreline Dispersion Model (SDM)
Emitted from Fossil Fuel Power Plants in B.16 SHORTZ
California: Data Analysis and Emission Im- B.17 Simple Line-Source Model
pact Model. Systems Applications, Inc., San B.18 SLAB
Rafael, CA. B.19 WYNDvalley Model
Moore, G.E., T.E. Stoeckenius and D.A. B.REF References
Stewart, 1982. A Survey of Statistical Meas-
B.0 Introduction and Availability
ures of Model Performance and Accuracy for
Several Air Quality Model. EPA Publication This appendix summarizes key features of
No. EPA–450/4–83–001. U.S. Environmental refined air quality models that may be con-
Protection Agency, Research Triangle Park, sidered on a case-by-case basis for individual
NC. regulatory applications. For each model, in-
Pasquill, F., 1976. Atmospheric Dispersion formation is provided on availability, ap-
Parameters in Gaussian Plume Modeling proximate cost, regulatory use, data input,
Part II. Possible Requirements for Change in output format and options, simulation of at-
the Turner Workbook Values. EPA Publica- mospheric physics and accuracy. The models
tion No. EPA–600/4–76–030b. U.S. Environ- are listed by name in alphabetical order.

406
Environmental Protection Agency Pt. 51, App. W
There are three separate conditions under Abstract
which these models will normally be ap-
The AVACTA II model is a Gaussian model
proved for use:
in which atmospheric dispersion phenomena
1. A demonstration can be made that the
are described by the evolution of plume ele-
model produces concentration estimates
ments, either segments or puffs. The model
equivalent to the estimates obtained using a
can be applied for short time (e.g., one day)
preferred model (e.g., the maximum or high,
simulations in both transport and calm con-
second-high concentration is within 2% of
ditions.
the estimate using the comparable preferred
The user is given flexibility in defining the
model);
computational domain, the three-dimen-
2. A statistical performance evaluation has
sional meteorological and emission input,
been conducted using measured air quality
the receptor locations, the plume rise for-
data and the results of that evaluation indi-
mulas, the sigma formulas, etc. Without ex-
cate the model in appendix B performs better
plicit user’s specifications, standard default
for the application than a comparable model
values are assumed.
in appendix A; and
AVACTA II provides both concentration
3. There is no preferred model for the spe-
fields on the user specified receptor points,
cific application but a refined model is need-
and dry/wet deposition patterns throughout
ed to satisfy regulatory requirements.
the domain. The model is particularly ori-
Any one of these three separate conditions
ented to the simulation of the dynamics and
may warrant use of these models. See sec-
transformation of sulfur species (SO2 and
tion 3.2, Use of Alternative Models, for addi-
SO4=), but can handle virtually any pair of
tional details.
primary-secondary pollutants.
Many of these models have been subject to
a performance evaluation by comparison a. Recommendations for Regulatory Use
with observed air quality data. A summary
of such comparisons for models contained in AVACTA II can be used if it can be dem-
this appendix is included in Moore et al. onstrated to estimate concentrations equiva-
(1982). Where possible, several of the models lent to those provided by the preferred model
contained herein have been subjected to rig- for a given application. AVACTA II must be
orous evaluation exercises, including (1) sta- executed in the equivalent mode.
tistical performance measures recommended AVACTA II can be used on a case-by-case
by the American Meteorological Society and basis in lieu of a preferred model if it can be
(2) peer scientific reviews. demonstrated, using the criteria in section
A source for some of these models and 3.2, that AVACTA II is more appropriate for
user’s documentation is: Computer Products, the specific application. In this case the
National Technical Information Service model options/modes which are most appro-
(NTIS), U.S. Department of Commerce, priate for the application should be used.
Springfield, VA 22161, Phone: (703) 487–4650. A
b. Input Requirements
number of the model codes and selected,
abridged user’s guides are also available A time-varying input is required at each
from the Support Center for Regulatory Air computational step. Only those data which
Models Bulletin Board System19 (SCRAM have changed need to be input by the user.
BBS), Telephone (919) 541–5742. The SCRAM Source data requirements are: Coordinates,
BBS is an electronic bulletin board system emission rates of primary and secondary pol-
designed to be user friendly and accessible lutants, initial plume sigmas (for non-point
from anywhere in the country. Model users sources), exit temperature, exit velocity,
with personal computers are encouraged to stack inside diameter.
use the SCRAM BBS to download current Meteorological data requirements are: sur-
model codes and text files. face wind measurements, wind profiles (if
available), atmospheric stability profiles,
B.1 AVACTA II Model mixing heights.
Receptor data requirements are: receptor
Reference
coordinates.
Zannetti, P., G. Carboni and R. Lewis, 1985. Other data requirements: coordinates of
AVACTA II User’s Guide (Release 3). the computational domain, grid cell speci-
AeroVironment, Inc., Technical Report AV– fication, terrain elevations, user’s computa-
OM–85/520. tional and printing options.
Availability c. Output
A 31⁄2’’ diskette of the FORTRAN coding The model’s output is provided according
and the user’s guide are available at a cost of to user’s printing flags. Hourly, 3-hour and
$3,500 (non-profit organization) or $5,000 24-hour concentration averages are com-
(other organizations) from: AeroVironment, puted, together with highest and highest-sec-
Inc., 222 Huntington Drive, Monrovia, CA ond-highest concentration values. Both par-
91016, Phone: (818) 357–9983. tial and total concentrations are provided.

407
Pt. 51, App. W 40 CFR Ch. I (7–1–97 Edition)
d. Type of Model User-specified function, with a user’s sub-
routine
AVACTA II is Gaussian segment/puff
The virtual distance/age concept is used for
model.
incrementing the sigmas at each time step.
e. Pollutant Types
k. Vertical Dispersion
AVACTA II can handle any couple of pri-
During each step, the sigmas of each ele-
mary-secondary pollutants (e.g., SO2 and
ment are increased. The user can select the
SO4=).
following sigma functions:
f. Source Receptor Relationship Pasquill-Gifford-Turner (in the functional
form specified by Green et al., 1980)
The AVACTA II approach maintains the Brookhaven (Gifford, 1975)
basic Gaussian formulation, but allows a nu- Briggs, open country (Gifford, 1975)
merical simulation of both nonstationary Briggs, urban, i.e., McElroy-Pooler (Gif-
and nonhomogeneous meteorological condi- ford, 1975)
tions. The emitted pollutant material is di- LO–LOCAT (MacCready et al., 1974)
vided into a sequence of ‘‘elements,’’ either User-specified function, with a user’s sub-
segments or puffs, which are connected to- routine
gether but whose dynamics are a function of The virtual distance/age concept is used for
the local meteorological conditions. Since incrementing the sigmas at each time step.
the meteorological parameters vary with
time and space, each element evolves accord- l. Chemical Transformation
ing to the different meteorological condi-
First order chemical reactions (primary-
tions encountered along its trajectory.
to-secondary pollutant)
AVACTA II calculates the partial con-
tribution of each source in each receptor m. Physical Removal
during each interval. The partial concentra-
tion is the sum of the contribution of all ex- First order dry and wet deposition schemes
isting puffs, plus that of the closest segment.
n. Evaluation Studies
g. Plume Behavior Zannetti P., G. Carboni and A. Ceriani,
The user can select the following plume 1985. AVACTA II Model Simulations of
rise formulas: Worst-Case Air Pollution Scenarios in
Briggs (1969, 1971, 1972) Northern Italy. 15th International Technical
CONCAWE (Briggs, 1975) Meeting on Air Pollution Modeling and Its
Lucas-Moore (Briggs, 1975) Application, St. Louis, Missouri, April 15–19.
User’s function, i.e., a subroutine supplied
by the user B.2 Dense Gas Dispersion Model (DEGADIS)
With cold plumes, the program uses a spe-
Reference
cial routine for the computation of the jet
plume rise. The user can also select several Environmental Protection Agency, 1989.
computational options that control plume User’s Guide for the DEGADIS 2.1—Dense
behavior in complex terrain and its total/ Gas Dispersion Model. EPA Publication No.
partial reflections. EPA–450/4–89–019. U.S. Environmental Pro-
tection Agency, Research Triangle Park, NC
h. Horizontal Winds 27711. (NTIS No. PB 90–213893)
A 3D mass-consistent wind field is option-
Availability
ally generated.
The model code is only available on the
i. Vertical Wind Speed Support Center for Regulatory Air Models
A 3D mass-consistent wind field is option- Bulletin Board System (see section B.0).
ally generated.
Abstract
j. Horizontal Dispersion DEGADIS 2.1 is a mathematical dispersion
During each step, the sigmas of each ele- model that can be used to model the trans-
ment are increased. The user can select the port of toxic chemical releases into the at-
following sigma functions: mosphere. Its range of applicability includes
Pasquill-Gifford-Turner (in the functional continuous, instantaneous, finite duration,
form specified by Green et al., 1980) and time-variant releases; negatively-buoy-
Brookhaven (Gifford, 1975) ant and neutrally-buoyant releases; ground-
Briggs, open country (Gifford, 1975) level, low-momentum area releases; ground-
Briggs, urban, i.e., McElroy-Pooler (Gif- level or elevated upwardly-directed stack re-
ford, 1975) leases of gases or aerosols. The model simu-
Irwin (1979a) lates only one set of meteorological condi-
LO–LOCAT (MacCready et al., 1974) tions, and therefore should not be considered
User-specified function, by points applicable over time periods much longer

408
Environmental Protection Agency Pt. 51, App. W
than 1 or 2 hours. The simulations are car- • Off-centerline distances to 2 specified
ried out over flat, level, unobstructed terrain concentration values at a specified receptor
for which the characteristic surface rough- height at each downwind distance (these val-
ness is not a significant fraction of the depth ues can be used to draw concentration
of the dispersion layer. The model does not isopleths after model execution);
characterize the density of aerosol-type re- • Concentration vs. time histories for fi-
leases; rather, the user must assess that nite-duration releases (if specified by user).
independently prior to the simulation. The output print file is automatically
saved and must be sent to the appropriate
a. Recommendations for Regulatory Use printer by the user after program execution.
DEGADIS can be used as a refined model- No graphical output is generated by the
ing approach to estimate short-term ambient current version of this program.
concentrations (1-hour or less averaging d. Type of Model
times) and the expected area of exposure to
concentrations above specified threshold val- DEGADIS estimates plume rise and disper-
ues for toxic chemical releases. The model is sion for vertically-upward jet releases using
especially useful in situations where density mass and momentum balances with air en-
effects are suspected to be important and trainment based on laboratory and field-
where screening estimates of ambient con- scale data. These balances assume Gaussian
centrations are above levels of concern. similarity profiles for velocity, density, and
concentration within the jet. Ground-level
b. Input Requirements denser-than-air phenomena is treated using a
Data may be input directly from an exter- power law concentration distribution profile
nal input file or via keyboard using an inter- in the vertical and a hybrid top hat-Gaussian
active program module. The model is not set concentration distribution profile in the hor-
up to accept real-time meteorological data izontal. A power law specification is used for
or convert units of input values. Chemical the vertical wind profile. Ground-level cloud
property data must be input by the user. slumping phenomena and air entrainment
Such data for a few selected species are are based on laboratory measurements and
available within the model. Additional data field-scale observations.
may be added to this data base by the user. e. Pollutant Types
Source data requirements are: emission
rate and release duration; emission chemical Neutrally- or negatively-buoyant gases and
and physical properties (molecular weight, aerosols. Pollutants are assumed to be non-
density vs. concentration profile in the case reactive and non-depositing.
of aerosol releases, and contaminant heat ca-
pacity in the case of a nonisothermal gas re- f. Source-Receptor Relationships
lease; stack parameters (i.e., diameter, ele- Only one source can be modeled at a time.
vation above ground level, temperature at There is no limitation to the number of re-
release point). ceptors; the downwind receptor distances are
Meteorological data requirements are: internally-calculated by the model. The
wind speed at designated height above DEGADIS calculation is carried out until
ground, ambient temperature and pressure, the plume centerline concentration is 50%
surface roughness, relative humidity, and below the lowest concentration level speci-
ground surface temperature (which in most fied by the user.
cases can be adequately approximated by the The model contains no modules for source
ambient temperature). calculations or release characterization.
Receptor data requirements are: averaging
time of interest, above-ground height of re- g. Plume Behavior
ceptors, and maximum distance between re- Jet/plume trajectory is estimated from
ceptors (since the model computes downwind mass and momentum balance equations. Sur-
receptor distances to optimize model per- rounding terrain is assumed to be flat, and
formance, this parameter is used only for stack tip downwash, building wake effects,
nominal control of the output listing, and is and fumigation are not treated.
of secondary importance). No indoor con-
centrations are calculated by the model. h. Horizontal Winds

c. Output Constant logarithmic velocity profile


which accounts for stability and surface
Printed output includes in tabular form: roughness is used.
• Listing of model input data; The wind speed profile exponent is deter-
• Plume centerline elevation, mole frac- mined from a least squares fit of the loga-
tion, concentration, density, and tempera- rithmic profile from ground level to the wind
ture at each downwind distance; speed reference height. Calm winds can be
• σy and σz values at each downwind dis- simulated for ground-level low-momentum
tance; releases.

409
Pt. 51, App. W 40 CFR Ch. I (7–1–97 Edition)
Along-wind dispersion of transient releases M2020–003. ENSR Consulting and Engineer-
is treated using the methods of Colenbrander ing, 35 Nagog Park, Acton, MA 01720.
(1980) and Beals (1971).
Availability
i. Vertical Wind Speed
The user’s guide and model code on
Not treated. diskette are available as a package (as PB
96–501978) from the National Technical Infor-
j. Horizontal Dispersion mation Service (see section B.0).
When the plume centerline is above ground
Abstract
level, horizontal dispersion coefficients are
based upon Turner (1969) and Slade (1968) The ERT Visibility Model is a Gaussian
with adjustments made for averaging time dispersion model designed to estimate visi-
and plume density. bility impairment for arbitrary lines of sight
When the plume centerline is at ground due to isolated point source emissions by
level, horizontal dispersion also accounts for simulating gas-to-particle conversion, dry
entrainment due to gravity currents as deposition, NO to NO2 conversion and linear
parameterized from laboratory experiments. radiative transfer.

k. Vertical Dispersion a. Recommendations for Regulatory Use


When the plume centerline is above ground There is no specific recommendation at the
level, vertical dispersion coefficients are present time. The ERT Visibility Model may
based upon Turner (1969) and Slade (1968). be used on a case-by-case basis.
Perfect ground reflection is applied.
In the ground-level dense-gas regime, ver- b. Input Requirements
tical dispersion is also based upon results Source data requirements are: stack
from laboratory experiments in density- height, stack temperature, emissions of SO2,
stratified fluids. NOX, TSP, fraction of NOX as NO2, fraction of
TSP which is carbonaceous, exit velocity,
l. Chemical Transformation and exit radius.
Not specifically treated. Meteorological data requirements are:
hourly ambient temperature, mixing depth,
m. Physical Removal wind speed at stack height, stability class,
potential temperature gradient, and wind di-
Not treated.
rection.
n. Evaluation Studies Receptor data requirements are: observer
coordinates with respect to source, latitude,
Spicer, T.O. and J.A. Havens, 1986. Devel- longitude, time zone, date, time of day, ele-
opment of Vapor Dispersion Models for Non- vation, relative humidity, background visual
neutrally Buoyant Gas Mixtures—Analysis range, line-of-sight azimuth and elevation
of USAF/N2O4 Test Data. USAF Engineering angle, inclination angle of the observed ob-
and Services Laboratory, Final Report ESL– ject, distance from observer to object, object
TR–86–24. and surface reflectivity, number and spacing
Spicer, T.O. and J.A. Havens, 1988. Devel- of integral receptor points along line of
opment of Vapor Dispersion Models for Non- sight.
neutrally Buoyant Gas Mixtures—Analysis Other data requirements are: ambient con-
of TFI/NH3 Test Data. USAF Engineering centrations of O3 and NOX, deposition veloc-
and Services Laboratory, Final Report. ity of TSP, sulfate, nitrate, SO2 and NOX,
first-order transformation rate for sulfate
o. Operating Information and nitrate.
The model requires either a VAX computer
or an IBM—compatible PC for its execu- c. Output
tion. The model currently does not require Printed output includes both summary and
supporting software. A FORTRAN compiler detailed results as follows: Summary output:
is required to generate program executables Page 1—site, observer and object parameters;
in the VAX computing environment. PC Page 2—optical pollutants and associated ex-
executables are provided within the source tinction coefficients; Page 3—plume model
code; however, a PC FORTRAN compiler input parameters; Page 4—total calculated
may be used to tailor a PC executable to the visual range reduction, and each pollutant’s
user’s PC environment. contribution; Page 5—calculated plume con-
trast, object contrast and object contrast
B.3 ERT Visibility Model degradation at the 550nm wavelength; Page
6—calculated blue/red ratio and ΛE (U*V*W*)
Reference
values for both sky and object discoloration.
ENSR Consulting and Engineering, 1990. Detailed output: phase functions for each
ERT Visibility Model: Version 4; Technical pollutant in four wavelengths (400, 450, 550,
Description and User’s Guide. Document 650nm), concentrations for each pollutant

410
Environmental Protection Agency Pt. 51, App. W
along sight path, solar geometry contrast Chimney Plumes: An Inter-comparison of
parameters at all wavelengths, intensities, Four Models with Observations at a Well-
tristimulus values and chromaticity coordi- Controlled Power Plant. Atmospheric Envi-
nates for views of the object, sun, back- ronment, 19: 515–528.
ground sky and plume.
B.4 HGSYSTEM
d. Type of Model
(Dispersion Models for Ideal Gases and Hy-
ERT Visibility model is a Gaussian plume drogen Fluoride)
model for estimating visibility impairment.
Reference
e. Pollutant Types Post, L. (ed.), 1994. HGSYSTEM 3.0 Tech-
Optical activity of sulfate, nitrate (derived nical Reference Manual. Shell Research Lim-
from SO2 and NOX emissions), primary TSP ited, Thornton Research Centre, Chester,
and NO2 is simulated. United Kingdom. (TNER 94.059)
Post, L., 1994. HGSYSTEM 3.0 User’s Man-
f. Source Receptor Relationship ual. Shell Research Limited, Thornton Re-
Single source and hour is simulated. Un- search Centre, Chester, United Kingdom.
limited number of lines-of-sight (receptors) (TNER 94.059)
is permitted per model run. Availability
g. Plume Behavior The PC–DOS version of the HGSYSTEM
Briggs (1971) plume rise equations for final software (HGSYSTEM: Version 3.0, Programs
rise are used. for modeling the dispersion of ideal gas and
hydrogen fluoride releases, executable pro-
h. Horizontal Wind Field grams and source code can be installed from
diskettes. These diskettes and all docu-
A single wind speed and direction is speci- mentation are available as a package from
fied for each case study. The wind is assumed API [(202) 682–8340] or from NTIS as PB 96–
to be spatially uniform. 501960 (see section B.0).
i. Vertical Wind Speed Technical Contacts
Vertical wind speed is assumed equal to Doug N. Blewitt, AMOCO Corporation, 1670
zero. Broadway/MC 2018, Denver, CO, 80201, (303)
830–5312.
j. Horizontal Dispersion
Howard J. Feldman, American Petroleum
Rural dispersion coefficients from Turner Institute, 1220 L Street Northwest, Washing-
(1969) are used. ton, DC 20005, (202) 682–8340.
k. Vertical Dispersion Abstract
Rural dispersion coefficients from Turner HGSYSTEM is a PC-based software pack-
(1969) are used. Mixing height is accounted age consisting of mathematical models for
for with multiple reflection handled by sum- estimating of one or more consecutive
mation of series near the source, and Fourier phases between spillage and near-field and
representation farther downwind. far-field dispersion of a pollutant. The pol-
lutant can be either a two-phase, multi-
l. Chemical Transformation compound mixture of non-reactive com-
First order transformations of sulfates and pounds or hydrogen fluoride (HF) with chem-
nitrates are used. ical reactions. The individual models are:
Database program:
m. Physical Removal DATAPROP Generates physical properties
used in other HGSYSTEM models
Dry deposition is treated by the source de- Source term models:
pletion method. SPILL Transient liquid release from a
n. Evaluation Studies pressurized vessel
HFSPILL SPILL version specifically for
Seigneur, C., R.W. Bergstrom and A.B. HF
Hudischewskyj, 1982. Evaluation of the EPA LPOOL Evaporating multi-compound liq-
PLUVUE Model and the ERT Visibility uid pool model
Model Based on the 1979 VISTTA Data Base. Near-field dispersion models:
EPA Publication No. EPA–450/4–82–008. U.S. AEROPLUME High-momentum jet disper-
Environmental Protection Agency, Research sion model
Triangle Park, NC. HFPLUME AEROPLUME version specifi-
White, W.H., C. Seigneur, D.W. Heinold, cally for HF
M.W. Eltgroth, L.W. Richards, P.T. Roberts, HEGABOX Dispersion of instantaneous
P.S. Bhardwaja, W.D. Conner and W.E. Wil- heavy gas releases
son, Jr., 1985. Predicting the Visibility of Far-field dispersion models:

411
Pt. 51, App. W 40 CFR Ch. I (7–1–97 Edition)
HEGADAS(S,T) Heavy gas dispersion function of time. For HF liquid, HFSPILL
(steady-state and transient version) generates link data to HFPLUME for the ini-
PGPLUME Passive Gaussian dispersion tial phase of choked liquid flow (flashing
Utility programs: jet), and link data to EVAP for the subse-
HFFLASH Flashing of HF from pressurized quent phase of unchoked liquid flow
vessel (evaporating liquid pool).
POSTHS/POSTHT Post-processing of EVAP output data: pool dimensions, pool
HEGADAS(S,T) results evaporation rate, pool mass and other pool
PROFILE Post-processor for concentration variables for steady state conditions or as a
contours of airborne plumes function of time. EVAP generates link data
GET2COL Utility for data retrieval to the dispersion model HEGADAS (pool di-
The models assume flat, unobstructed ter- mensions and pool evaporation rate).
rain. HGSYSTEM can be used to model HFPLUME and PLUME output data:
steady-state, finite-duration, instantaneous plume variables (concentration, width, cen-
and time dependent releases, depending on troid height, temperature, velocity, etc.) as a
the individual model used. The models can function of downwind distance.
be run consecutively, with relevant data HEGADAS output data: concentration
being passed on from one model to the next variables and temperature as a function of
using link files. The models can be run in downwind distance and (for transient case)
batch mode or using an iterative utility pro- time.
gram. PGPLUME output data: concentration as a
function of downwind distance, cross-wind
a. Recommendations for Regulatory Use distance and height.
HGSYSTEM can be used as a refined model d. Type of Model
to estimate short-term ambient concentra-
tions. For toxic chemical releases (non-reac- HGSYSTEM is made up of four types of
tive chemicals or hydrogen fluoride; 1-hour dispersion models. HFPLUME and PLUME
or less averaging times) the expected area of simulate the near-field dispersion and
exposure to concentrations above specified PGPLUME simulates the passive-gas disper-
threshold values can be determined. For sion downwind of a transition point.
flammable non-reactive gases it can be used HEGADAS simulates the ground-level heavy-
to determine the area in which the cloud gas dispersion.
may ignite.
e. Pollutant Types
b. Input Requirements HGSYSTEM may be used to model non-re-
HFSPILL input data: reservoir data (tem- active chemicals or hydrogen fluoride.
perature, pressure, volume, HF mass, mass-
f. Source-Receptor Relationships
fraction water), pipe-exit diameter and ambi-
ent pressure. HGSYSTEM estimates the expected area of
EVAP input data: spill rate, liquid prop- exposure to concentrations above user-speci-
erties, and evaporation rate (boiling pool) or fied threshold values. By imposing conserva-
ambient data (non-boiling pool). tion of mass, momentum and energy the con-
HFPLUME and PLUME input data: res- centration, density, speed and temperature
ervoir characteristics, pollutant parameters, are evaluated as a function of downwind dis-
pipe/release data, ambient conditions, sur- tance.
face roughness and stability class.
HEGADAS input data: ambient conditions, g. Plume Behavior
pollutant parameters, pool data or data at HFPLUME and PLUME: (1) are steady-
transition point, surface roughness, stability state models assuming a top-hat profile with
class and averaging time. cross-section averaged plume variables; and
PGPLUME input data: link data provided (2) the momentum equation is taken into ac-
by HFPLUME and the averaging time. count for horizontal ambient shear, gravity,
ground collision, gravity-slumping pressure
c. Output forces and ground-surface drag.
The HGSYSTEM models contain three HEGADAS: assumes the heavy cloud to
post-processor programs which can be used move with the ambient wind speed, and
to extract modeling results for graphical dis- adopts a power-law fit of the ambient wind
play by external software packages. speed for the velocity profile.
GET2COL can be used to extract data from PGPLUME: simulates the passive-gas dis-
the model output files. HSPOST can be used persion downwind of a transition point from
to develop isopleths, extract any 2 param- HFPLUME or PLUME for steady-state and
eters for plotting and correct for finite re- finite duration releases.
lease duration. HTPOST can be used to
h. Horizontal Winds
produce time history plots.
HFSPILL output data: reservoir mass, A power law fit of the ambient wind speed
spill rate, and other reservoir variables as a is used.

412
Environmental Protection Agency Pt. 51, App. W
i. Vertical Wind Speed McFarlane, K., Prothero, A., Puttock, J.S.,
Roberts, P.T. and H.W.M. Witlox, 1990. Devel-
Not treated.
opment and validation of atmospheric dis-
j. Horizontal Dispersion persion models for ideal gases and hydrogen
fluoride, Part I: Technical Reference Man-
HFPLUME and PLUME: Plume dilution is ual. Report TNER.90.015. Thornton Research
caused by air entrainment resulting from Centre, Shell Research, Chester, England.
high plume speeds, trailing vortices in wake [EGG 1067–1151] (NTIS No. DE 93–000953)
of falling plume (before touchdown), ambient Witlox, H.W.M., McFarlane, K., Rees, F.J.
turbulence and density stratification. Plume and J.S. Puttock, 1990. Development and val-
dispersion is assumed to be steady and mo- idation of atmospheric dispersion models for
mentum-dominated, and effects of downwind ideal gases and hydrogen fluoride, Part II:
diffusion and wind meander (averaging time) HGSYSTEM Program User’s Manual. Report
are not taken into account. TNER.90.016. Thornton Research Centre,
HEGADAS: This model adopts a concentra- Shell Research, Chester, England. [EGG 1067–
tion similarity profile expressed in terms of 1152] (NTIS No. DE 93–000954)
an unknown center-line ground-level con-
centration and unknown vertical/cross-wind B.5 HOTMAC/RAPTAD
dispersion parameters. These quantities are
determined from a number of basic equations Reference
describing gas-mass conservation, air en- Mellor, G.L. and T. Yamada, 1974. A Hier-
trainment (empirical law describing vertical archy of Turbulence Closure Models for Plan-
top-entrainment in terms of global Richard- etary Boundary Layers. Journal of Atmos-
son number), cross-wind gravity spreading pheric Sciences, 31: 1791–1806.
(initial gravity spreading followed by grav- Mellor, G.L. and T. Yamada, 1982. Develop-
ity-current collapse) and cross-wind diffu- ment of a Turbulence Closure Model for Geo-
sion (Briggs formula). physical Fluid Problems. Rev. Geophys.
PGPLUME: This model assumes a Space Phys., 20: 851–875.
Gaussian concentration profile in which the Yamada, T. and S. Bunker, 1988. Develop-
cross-wind and vertical dispersion coeffi- ment of a Nested Grid, Second Moment Tur-
cients are determined by empirical expres- bulence Closure Model and Application to
sions. All unknown parameters in this pro- the 1982 ASCOT Brush Creek Data Simula-
file are determined by imposing appropriate tion. Journal of Applied Meteorology, 27: 562–
matching criteria at the transition point. 578.
k. Vertical Dispersion Availability
See description above. For a cost to be negotiated with the model
l. Chemical Transformation developer, a 1⁄4–inch data cartridge or a 4mm
DAT tape containing the HOTMAC/RAPTAD
Not treated. computer codes including pre- and post-proc-
essors and hard copies of user manuals
m. Physical Removal (User’s Manual, Maintenance Manual, Oper-
Not treated. ations Manual, Maintenance Interface Man-
ual, Topo Manual, and 3–Dimensional Plume
n. Evaluation Studies Manual) are available from YSA Corpora-
PLUME has been validated against field tion, Rt. 4 Box 81–A, Santa Fe, NM 87501;
data for releases of liquified propane, and Phone: (505) 989–7351; Fax: (505) 989–7965; e-
wind tunnel data for buoyant and vertically- mail: ysa@RT66.com
released dense plumes. HFPLUME and
Abstract
PLUME have been validated against field
data for releases of HF (Goldfish experi- YSA Corporation offers a comprehensive
ments) and propane releases. In addition, the modeling system for environmental studies.
plume rise algorithms have been tested The system includes a mesoscale meteoro-
against Hoot, Meroney, and Peterka, Ooms logical code, a transport and diffusion code,
and Petersen databases. HEGADAS has been and extensive Graphical User Interfaces
validated against steady and transient re- (GUIs). This system is unique because the
leases of liquid propane and LNG over water diffusion code uses time dependent, three-di-
(Maplin Sands field data), steady and finite- mensional winds and turbulence distribu-
duration pressurized releases of HF (Goldfish tions that are forecasted by a mesoscale
experiments; linked with HFPLUME), in- weather prediction model. Consequently the
stantaneous release of Freon (Thorney Island predicted concentration distributions are
field data; linked with the box model more accurate than those predicted by tradi-
HEGABOX) and wind tunnel data for steady, tional models when surface conditions are
isothermal dispersion. heterogeneous. In general, the modeled con-
Validation studies are contained in the fol- centration distributions are not Gaussian be-
lowing references. cause winds and turbulence distributions

413
Pt. 51, App. W 40 CFR Ch. I (7–1–97 Edition)
change considerably in time and space over a. Recommendation for Regulatory Use
complex terrain.
There are no specific recommendations at
The models were originally developed by the present time. The HOTMAC/RAPTAD
using super computers. However, recent ad- modeling system may be used on a case-by-
vancement of computer hardware has made case basis.
it possible to run complex three-dimensional
meteorological models on desktop b. Input Requirements
workstations. The present versions of the
Meteorological Data: The modeling system
programs are running on super computers
is significantly different from the majority
and workstations. GUIs are available on Sun
of regulatory models in terms of how mete-
Microsystems and Silicon Graphics
orological data are provided and used in con-
workstations. The modeling system can also
centration simulations. Regulatory models
run on a laptop workstation which makes it
use the wind data which are obtained di-
possible to run the programs in the field or
rectly from measurements or analyzed by
away from the office. As technology contin-
using a simple constraint such as a mass
ues to advance, a version of HOTMAC/
conservation equation. Thus, the accuracy of
RAPTAD suitable for PC-based platforms the computation will depend significantly on
will be considered for release by YSA. the quantity and quality of the wind data.
HOTMAC, Higher Order Turbulence Model This approach is acceptable as long as the
for Atmospheric Circulation, is a mesoscale study area is flat and the simulation period
weather prediction model that forecasts is short. As the regulations become more
wind, temperature, humidity, and atmos- stringent and more realistic surface condi-
pheric turbulence distributions over complex tions are required, a significantly large vol-
surface conditions. HOTMAC has options to ume of meteorological data is required which
include non-hydrostatic pressure computa- could become very expensive.
tion, nested grids, land-use distributions, An alternative approach is to augment the
cloud, fog, and precipitation physics. measurements with predicted values from a
HOTMAC can interface with tower, rawin- mesoscale meteorological model. This is the
sonde, and large-scale weather data using a approach we have taken here. This approach
four-dimensional data assimilation method. has several advantages over the conventional
RAPTAD, Random Puff Transport and Diffu- method. First, concentration computations
sion, is a Lagrangian random puff model that use the model forecast wind while the con-
is used to forecast transport and diffusion of ventional method extrapolates the observed
airborne materials over complex terrain. winds. Extrapolation of wind data over com-
Concentrations are computed by summing plex terrain and for an extended period of
the concentration of each puff at the recep- time quickly loses its accuracy. Secondly,
tor location. The random puff method is the number of stations for upper air sound-
equivalent to the random particle method ings is typically limited from none to at
with a Gaussian kernel for particle distribu- most a few stations in the study area. The
tion. The advantage of the puff method is the corresponding number in a mesoscale model
accuracy and speed of computation. The par- is the number of grid points in the horizontal
ticle method requires the release of a large plane which is typically 50 X 50. Con-
number of particles which could be sequently, concentration distributions using
computationally expensive. The puff method model forecasted winds would be much more
requires the release of a much less number of accurate than those obtained by using winds
puffs, typically 1⁄10 to 1⁄100 of the number of which were extrapolated from the limited
particles required by the particle method. number of measurements.
The averaging time for concentration esti- HOTMAC requires meteorological data for
mates is variable from 5 minutes to 15 min- initialization and to provide boundary condi-
utes for each receptor. In addition to the tions if the boundary conditions change sig-
concentration computation at the receptor nificantly with time. The minimum amount
sites, RAPTAD computes and graphically of data required to run HOTMAC is wind and
displays hourly concentration contours at potential temperature profiles at a single
the ground level. RAPTAD is applicable to station. HOTMAC forecasts wind and turbu-
point and area sources. lence distributions in the boundary layer
The meteorological data produced from through a set of model equations for solar
HOTMAC are used as input to RAPTAD. radiation, heat energy balance at the
RAPTAD can forecast concentration dis- ground, conservation of momentum, con-
tributions for neutrally buoyant gas, buoy- servation of internal energy, and conserva-
ant gas and denser-than-air gas. The models tion of mass.
are significantly advanced in both their Terrain Data: HOTMAC and RAPTAD use
model physics and in their operational pro- the digitized terrain data from the U.S. Geo-
cedures. GUIs are provided to help the user logical Survey and the Defense Mapping
prepare input files, run programs, and dis- Agency. Extraction of terrain data is greatly
play the modeled results graphically in three simplified by using YSA’s GUI software
dimensions. called Topo. The user specifies the latitudes

414
Environmental Protection Agency Pt. 51, App. W
and longitudes of the southwest and north- plume equations are based on Van Dop (1992).
east corner points of the study area. Then, In general, plumes are non-Gaussian.
Topo extracts the digitized elevation data
within the area specified and converts from h. Horizontal Winds
the latitudes and longitudes to the UTM RAPTAD uses wind speed, wind direction,
(Universal Transverse Mercator) coordinates and turbulence on a gridded array that is
for up to three nested grids. supplied hourly by HOTMAC. Stability effect
Emission Data: Emission data require- and mixed layer height are incorporated
ments are emission rate, stack height, stack through the intensity of turbulence which is
diameter, stack location, stack gas exit ve- a function of stability. HOTMAC predicts
locity, and stack buoyancy. turbulence intensity by solving a turbulence
Receptor Data: Receptor data require- kinetic energy equation and a length scale
ments are names, location coordinates, and equation. RAPTAD interpolates winds and
desired averaging time for concentration es- turbulence at puff center locations every 10
timates, which is variable from 5 to 15 min- seconds from the values on a gridded array.
utes. RAPTAD can also use the winds observed at
towers and by rawinsondes.
c. Output
HOTMAC outputs include hourly winds, i. Vertical Wind Speed
temperatures, and turbulence variables at RAPTAD uses vertical winds on a gridded
every grid point. Ancillary codes graphically array that are supplied hourly by HOTMAC.
display vertical profiles of wind, tempera- HOTMAC computes vertical wind either by
ture, and turbulence variables at selected lo- solving an equation of motion for the verti-
cations and wind vector distributions at cal wind or a mass conservation equation.
specified heights above the ground. These RAPTAD interpolates vertical winds at puff
codes also produce graphic files of wind di- center locations every 10 seconds from the
rection projected on vertical cross sections. values on a gridded array.
RAPTAD outputs include hourly values of
surface concentration, time variations of j. Horizontal Dispersion
mean and standard deviation of concentra- Horizontal dispersion is based on the
tions at selected locations, and coordinates standard deviations of horizontal winds that
of puff center locations. Ancillary codes are computed by HOTMAC.
produce color contour plots of surface con-
centration, time variations of mean con- k. Vertical Dispersion
centrations and ratios of standard deviation
Vertical dispersion is based on the stand-
to mean value at selected locations, and con-
ard deviations of vertical wind that are com-
centration distributions in the vertical cross
puted by HOTMAC.
sections. The averaging time of concentra-
tion at a receptor location is variable from 5 l. Chemical Transformation
to 15 minutes. Color contour plots of surface
concentration can be animated on the mon- HOTMAC can provide meteorological in-
itor to review time variations of high con- puts to other models that handle chemical
centration areas. reactions, e.g., UAM.

d. Type of Model m. Physical Removal

HOTMAC is a 3-dimensional Eulerian Not treated.


model for weather forecasting, and RAPTAD n. Evaluation Studies
is a 3-dimensional Lagrangian random puff
model for pollutant transport and diffusion. Yamada, T., S. Bunker and M. Moss, 1992.
A Numerical Simulation of Atmospheric
e. Pollutant types Transport and Diffusion over Coastal Com-
RAPTAD may be used to model any inert plex Terrain. Journal of Applied Meteor-
pollutants, including dense and buoyant ology, 31: 565–578.
Yamada, T. and T. Henmi, 1994. HOTMAC:
gases.
Model Performance Evaluation by Using
f. Source-Receptor Relationship Project WIND Phase I and II Data. Mesoscale
Modeling of the Atmosphere, American Me-
Up to six point or area sources are speci- teorological Society, Monograph 47, pp. 123–
fied and up to 50 sampling locations are se- 135.
lected. Source and receptor heights are spec-
ified by the user. B.6 LONGZ
g. Plume Behavior Reference
Neutrally buoyant plumes are transported Bjorklund, J.R. and J.F. Bowers, 1982.
by mean and turbulence winds that are mod- User’s Instructions for the SHORTZ and
eled by HOTMAC. Non-neutrally buoyant LONGZ Computer Programs, Volumes I and

415
Pt. 51, App. W 40 CFR Ch. I (7–1–97 Edition)
II, EPA Publication No. EPA–903/9–82–004. c. Output
U.S. Environmental Protection Agency, Re-
Printed output includes total concentra-
gion III, Philadelphia, PA.
tion due to emissions from user-specified
Availability source groups, including the combined emis-
sions from all sources (with optional allow-
The computer code is available on the Sup- ance for depletion by deposition).
port Center for Regulatory Air Models Bul-
letin Board System and on diskette (as PB d. Type of Model
96–501994) from the National Technical Infor-
mation Service (see section B.0). LONGZ is a climatological Gaussian plume
model.
Abstract
e. Pollutant Types
LONGZ utilizes the steady-state univariate
Gaussian plume formulation for both urban LONGZ may be used to model primary pol-
and rural areas in flat or complex terrain to lutants. Settling and deposition are treated.
calculate long-term (seasonal and/or annual) f. Source-Receptor Relationships
ground-level ambient air concentrations at-
tributable to emissions from up to 14,000 ar- LONGZ applies user specified locations for
bitrarily placed sources (stacks, buildings sources and receptors. Receptors are as-
and area sources). The output consists of the sumed to be at ground level.
total concentration at each receptor due to
emissions from each user-specified source or g. Plume Behavior
group of sources, including all sources. An Plume rise equations of Bjorklund and
option which considers losses due to deposi- Bowers (1982) are used.
tion (see the description of SHORTZ) is
Stack tip downwash (Bjorklund and Bow-
deemed inappropriate by the authors for
ers, 1982) is included.
complex terrain, and is not discussed here.
All plumes move horizontally and will
a. Recommendations for Regulatory Use fully intercept elevated terrain.
Plumes above mixing height are ignored.
LONGZ can be used if it can be dem- Perfect reflection at mixing height is as-
onstrated to estimate concentrations equiva- sumed for plumes below the mixing height.
lent to those provided by the preferred model
Plume rise is limited when the mean wind
for a given application. LONGZ must be exe-
at stack height approaches or exceeds stack
cuted in the equivalent mode.
exit velocity.
LONGZ can be used on a case-by-case basis
Perfect reflection at ground is assumed for
in lieu of a preferred model if it can be dem-
pollutants with no settling velocity.
onstrated, using the criteria in section 3.2 of
appendix W, that LONGZ is more appropriate Zero reflection at ground is assumed for
for the specific application. In this case the pollutants with finite settling velocity.
model options/modes which are most appro- LONGZ does not simulate fumigation.
priate for the application should be used. Tilted plume is used for pollutants with
settling velocity specified.
b. Input Requirements Buoyancy-induced dispersion is treated
(Briggs, 1972).
Source data requirements are: for point,
building or area sources, location, elevation, h. Horizontal Winds
total emission rate (optionally classified by
gravitational settling velocity) and decay Wind field is homogeneous and steady-
coefficient; for stack sources, stack height, state.
effluent temperature, effluent exit velocity, Wind speed profile exponents are functions
stack radius (inner), emission rate, and of both stability class and wind speed. De-
ground elevation (optional); for building fault values are specified in Bjorklund and
sources, height, length and width, and ori- Bowers (1982).
entation; for area sources, characteristic
vertical dimension, and length, width and i. Vertical Wind Speed
orientation.
Vertical wind speed is assumed equal to
Meteorological data requirements are: zero.
wind speed and measurement height, wind
profile exponents, wind direction standard j. Horizontal Dispersion
deviations (turbulent intensities), mixing
height, air temperature, vertical potential Pollutants are initially uniformly distrib-
temperature gradient. uted within each wind direction sector. A
Receptor data requirements are: coordi- smoothing function is then used to remove
nates, ground elevation. discontinuities at sector boundaries.

416
Environmental Protection Agency Pt. 51, App. W
k. Vertical Dispersion vective scaling instead of the Turner cri-
teria; (2) Briggs’ dispersion curves for ele-
Vertical dispersion is derived from input
vated sources are used; (3) Briggs plume rise
vertical turbulent intensities using adjust-
formulas for convective conditions are in-
ments to plume height and rate of plume
cluded; and (4) plume penetration of elevated
growth with downwind distance specified in
stable layers is given by Briggs’ (1984) model.
Bjorklund and Bowers (1982).
a. Recommendations for Regulatory Use
l. Chemical Transformation
PPSP can be used if it can be dem-
Chemical transformations are treated
onstrated to estimate concentrations equiva-
using exponential decay. Time constant is
lent to those provided by the preferred model
input by the user.
for a given application. PPSP must be exe-
m. Physical Removal cuted in the equivalent mode.
PPSP can be used on a case-by-case basis
Gravitational settling and dry deposition in lieu of a preferred model if it can be dem-
of particulates are treated. onstrated, using the criteria in section 3.2 of
n. Evaluation Studies appendix W, that PPSP is more appropriate
for the specific application. In this case the
Bjorklund, J.R. and J.F. Bowers, 1982. model options/modes which are most appro-
User’s Instructions for the SHORTZ and priate for the application should be used.
LONGZ Computer Programs, Volume I and
II. EPA Publication No. EPA–903/9–82–004. b. Input Requirements
U.S. Environmental Protection Agency, Re- Source data requirements are: emission
gion III, Philadelphia, PA. rate (monthly rates optional), physical stack
B.7 Maryland Power Plant Siting Program height, stack gas exit velocity, stack inside
(PPSP) Model diameter, stack gas temperature.
Meteorological data requirements are:
Reference hourly surface weather data from the EPA
meteorological preprocessor program.
Brower, R., 1982. The Maryland Power Preprocessor output includes hourly stabil-
Plant Siting Program (PPSP) Air Quality ity class, wind direction, wind speed, tem-
Model User’s Guide. Ref. No. PPSP–MP–38. perature, and mixing height. Actual ane-
Prepared for Maryland Department of Natu- mometer height (a single value) is also re-
ral Resources by Environmental Center, quired. Wind speed profile exponents (one for
Martin Marietta Corporation, Baltimore, each stability class) are required if on-site
MD. (NTIS No. PB 82–238387) data are input.
Weil, J.C. and R.P. Brower, 1982. The Mary-
Receptor data requirements are: distance
land PPSP Dispersion Model for Tall Stacks.
of each of the five receptor rings.
Ref. No. PPSP–MP–36. Prepared for Mary-
land Department of Natural Resources by c. Output
Environmental Center, Martin Marietta Cor-
poration, Baltimore, MD. (NTIS No. PB 82– Printed output includes:
219155) Highest and second highest concentrations
for the year at each receptor for averaging
Availability times of 1, 3, and 24-hours, plus a user-se-
lected averaging time which may be 2, 4, 6, 8,
The model code and test data are available
or 12 hours;
on diskette for a nominal cost to defray ship-
Annual arithmetic average at each recep-
ping and handling charges from: Mr. Roger
tor; and
Brower, Versar, Inc., 9200 Rumsey Road, Co-
For each day, the highest 1-hour and 24-
lumbia, MD 21045; Phone: (410) 964–9299.
hour concentrations over the receptor field.
Abstract
d. Type of Model
PPSP is a Gaussian dispersion model appli-
PPSP is a Gaussian plume model.
cable to tall stacks in either rural or urban
areas, but in terrain that is essentially flat e. Pollutant Types
(on a scale large compared to the ground
roughness elements). The PPSP model fol- PPSP may be used to model primary pol-
lows the same general formulation and com- lutants. Settling and deposition are not
puter coding as CRSTER, also a Gaussian treated.
model, but it differs in four major ways. The
f. Source-Receptor Relationship
differences are in the scientific formulation
of specific ingredients or ‘‘sub-models’’ to Up to 19 point sources are treated.
the Gaussian model, and are based on recent All point sources are assumed at the same
theoretical improvements as well as support- location.
ing experimental data. The differences are: Unique stack height and stack exit condi-
(1) stability during daytime is based on con- tions are applied for each source.

417
Pt. 51, App. W 40 CFR Ch. I (7–1–97 Edition)
Receptor locations are restricted to 36 azi- Weil, J.C. and R.P. Brower, 1982. The Mary-
muths (every 10 degrees) and five user-speci- land PPSP dispersion model for tall stacks.
fied radial distances. Ref. No. PPSP MP–36. Prepared for Maryland
Department of Natural Resources. Prepared
g. Plume Behavior by Environmental Center, Martin Marietta
Briggs (1975) final rise formulas for buoy- Corporation, Baltimore, Maryland. (NTIS
ant plumes are used. Momentum rise is not No. PB 82–219155)
considered.
B.8 Mesoscale Puff Model (MESOPUFF II)
Transitional or distance-dependent plume
rise is not modeled. Reference
Penetration (complete, partial, or zero) of
elevated inversions is treated with Briggs Scire, J.S., F.W. Lurmann, A. Bass and
(1984) model; ground-level concentrations are S.R. Hanna, 1984. User’s Guide to the
dependent on degree of plume penetration. Mesopuff II Model and Related Processor
Programs. EPA Publication No. EPA–600/8–
h. Horizontal Winds 84–013. U.S. Environmental Protection Agen-
cy, Research Triangle Park, NC. (NTIS No.
Wind speeds are corrected for release
PB 84–181775)
height based on power law variation, with
A Modeling Protocol for Applying
different exponents for different stability
MESOPUFF II to Long Range Transport
classes and variable reference height (7 me-
Problems, 1992. EPA Publication No. EPA–
ters is default). Wind speed power law expo-
454/R–92–021. U.S. Environmental Protection
nents are 0.10, 0.15, 0.20, 0.25, 0.30, and 0.30 for
Agency, Research Triangle Park, NC.
stability classes A through F, respectively.
Constant, uniform (steady-state) wind as- Availability
sumed within each hour.
This model code is available on the Sup-
i. Vertical Wind Speed port Center for Regulatory Air Models Bul-
letin Board System and also on diskette (as
Vertical wind speed is assumed equal to
PB 93–500247) from the National Technical
zero.
Information Service (see section B.0).
j. Horizontal Dispersion
Abstract
Rural dispersion parameters are Briggs
MESOPUFF II is a short term, regional
(Gifford, 1975), with stability class defined by
scale puff model designed to calculate con-
u/w* during daytime, and by the method of
centrations of up to 5 pollutant species (SO2,
Turner (1964) at night.
SO4, NOX, HNO3, NO3). Transport, puff
Urban dispersion is treated by changing all
growth, chemical transformation, and wet
stable cases to stability class D.
and dry deposition are accounted for in the
Buoyancy-induced dispersion (Pasquill,
model.
1976) is included (using ∆Η/3.5).
a. Recommendations for Regulatory Use
k. Vertical Dispersion
There is no specific recommendation at the
Rural dispersion parameters are Briggs
present time. The model may be used on a
(Gifford, 1975), with stability class defined by
case-by-case basis.
u/w* during daytime, and by the method of
Turner (1964). b. Input Requirements
Urban dispersion is treated by changing all
stable cases to stability class D. Required input data include four types: (1)
Buoyancy-induced dispersion (Pasquill, input control parameters and selected tech-
1976) is included (using ∆Η/3.5). nical options, (2) hourly surface meteorologi-
cal data and twice daily upper air measure-
l. Chemical Transformation ments, hourly precipitation data are op-
tional, (3) surface land use classification in-
Not treated.
formation, (4) source and emissions data.
m. Physical Removal Data from up to 25 surface National Weath-
er Service stations and up to 10 upper air
Not treated. stations may be considered. Spatially vari-
able fields at hour intervals of winds, mixing
n. Evaluation Studies
height, stability class, and relevant turbu-
Londergan, R., D. Minott, D. Wackter, T. lence parameters are derived by MESOPAC
Kincaid and D. Bonitata, 1983. Evaluation of II, the meteorological preprocessor program
Rural Air Quality Simulation Models, Ap- described in the User Guide.
pendix G: Statistical Tables for PPSP. EPA Source and emission data for up to 25 point
Publication No. EPA–450/4–83–003. Environ- sources and/or up to 5 area sources can be in-
mental Protection Agency, Research Tri- cluded. Required information are: location in
angle Park, NC. grid coordinates, stack height, exit velocity

418
Environmental Protection Agency Pt. 51, App. W
and temperature, and emission rates for the k. Vertical Dispersion
pollutant to be modeled.
For puffs emitted at an effective stack
Receptor data requirements: up to a 40×40 height which is less than the mixing height,
grid may be used and non-gridded receptor uniform mixing of the pollutant within the
locations may be considered. mixed layer is performed. For puffs centered
above the mixing height, no effect at the
c. Output ground occurs.
Line printer output includes: all input pa-
l. Chemical Transformation
rameters, optionally selected arrays of
ground-level concentrations of pollutant spe- Hourly chemical rate constants are com-
cies at specified time intervals. puted from empirical expressions derived
Line printer contour plots output from from photochemical model simulations.
MESOFILE II post-processor program. Com-
puter readable output of concentration array m. Physical Removal
to disk/tape for each hour. Dry deposition is treated with a resistance
method.
d. Type of Model Wet removal may be considered if hourly
precipitation data are input.
MESOPUFF II is a Gaussian puff super-
position model. n. Evaluation Studies
e. Pollutant Types Results of tests for some model parameters
are discussed in:
Up to five pollutant species may be mod- Scire, J.S., F.W. Lurmann, A. Bass and
eled simultaneously and include: SO2, SO4, S.R. Hanna, 1984. Development of the
NOX, HNO3, NO3. MESOPUFF II Dispersion Model. EPA Publi-
cation No. EPA–600/3–84–057. U.S. Environ-
f. Source-Receptor Relationship mental Protection Agency, Research Tri-
Up to 25 point sources and/or up to 5 area angle Park, NC.
sources are permitted. B.9 Mesoscale Transport Diffusion and Deposi-
tion Model for Industrial Sources (MTDDIS)
g. Plume Behavior
Briggs (1975) plume rise equations are used, Reference
including plume penetration with buoyancy Wang, I.T. and T.L. Waldron, 1980. User’s
flux computed in the model. Guide for MTDDIS Mesoscale Transport, Dif-
Fumigation of puffs is considered and may fusion, and Deposition Model for Industrial
produce immediate mixing or multiple re- Sources. EMSC6062.1UR(R2). Combustion En-
flection calculations at user option. gineering, Newbury Park, CA.

h. Horizontal Winds Availability

Gridded wind fields are computed for 2 lay- A diskette copy of the FORTRAN coding
ers; boundary layer and above the mixed and the user’s guide are available for a cost
layer. Upper air rawinsonde data and hourly of $100 from: Dr. I. T. Wang, Environmental
surface winds are used to obtain spatially Modeling & Analysis, 2219 E. Thousand Oaks
variable u,v component fields at hourly in- Blvd., Suite 435, Thousand Oaks, CA 91362.
tervals. The gridded fields are computed by Abstract
interpolation between stations in the
MESOPAC II preprocessor. MTDDIS is a variable-trajectory Gaussian
puff model applicable to long-range trans-
i. Vertical Wind Speed port of point source emissions over level or
rolling terrain. The model can be used to de-
Vertical winds are assumed to be zero. termine 3-hour maximum and 24-hour aver-
age concentrations of relatively nonreactive
j. Horizontal Dispersion pollutants from up to 10 separate stacks.
Incremental puff growth is computed over
a. Recommendations for Regulatory Use
discrete time steps with horizontal growth
parameters determined from power law equa- There is no specific recommendation at the
tions fit to sigma y curves of Turner out to present time. The MTDDIS Model may be
100km. At distances greater than 100km, puff used on a case-by-case basis.
growth is determined by the rate given by
Heffter (1965). b. Input Requirements
Puff growth is a function of stability class Source data requirements are: emission
and changes in stability are treated. Option- rate, physical stack height, stack gas exit
ally, user input plume growth coefficients velocity, stack inside diameter, stack gas
may be considered. temperature, and location.

419
Pt. 51, App. W 40 CFR Ch. I (7–1–97 Edition)
Meteorological data requirements are: rithms of Heffter (1980) to calculate the ef-
hourly surface weather data, from up to 10 fective transport wind speed and direction.
stations, including cloud ceiling, wind direc-
tion, wind speed, temperature, opaque cloud i. Vertical Wind Field
cover and precipitation. For long-range ap- Vertical wind speed is assumed equal to
plications, user-analyzed daily mixing zero.
heights are recommended. If these are not
available, the NWS daily mixing heights will j. Horizontal Dispersion
be used by the program. A single upper air
sounding station for the region is assumed. Transport-time-dependent dispersion coef-
For each model run, air trajectories are gen- ficients from Heffter (1980) are used.
erated for a 48-hour period, and therefore,
the afternoon mixing height of the day be- k. Vertical Dispersion
fore and the mixing heights of the day after Transport-time-dependent dispersion coef-
are also required by the model as input, in ficients from Heffter (1980) are used.
order to generate hourly mixing heights for
the modeled period. l. Chemical Transformation
Receptor data requirements are: up to
three user-specified rectangular grids. Chemical transformations are treated
using exponential decay. Half-life is input by
c. Output the user.
Printed output includes: m. Physical Removal
Tabulations of hourly meteorological pa-
rameters include both input surface observa- Dry deposition is treated. User input depo-
tions and calculated hourly stability classes sition velocity is required.
and mixing heights for each station; Wet deposition is treated. User input hour-
Printed air trajectories for the two con- ly precipitation rate and precipitation layer
secutive 24-hour periods for air parcels gen- depth or cloud ceiling height are required.
erated 4 hours apart starting at 0000 LST;
and n. Evaluation Studies
3-hour maximum and 24-hour average grid Carhart, R.A., A.J. Policastro, M. Wastag
concentrations over user-specified rectangu- and L. Coke, 1989. Evaluation of Eight Short-
lar grids are output for the second 24-hour Term Long-Range Transport Models Using
period. Field Data. Atmospheric Environment, 23:
85–105.
d. Type of Model
MTDDIS is a Gaussian puff model. B.10 Multi-Source (SCSTER) Model

e. Pollutant Types Reference


MTDDIS can be used to model primary pol- Malik, M.H. and B. Baldwin, 1980. Program
lutants. Dry deposition is treated. Expo- Documentation for Multi-Source (SCSTER)
nential decay can account for some reac- Model. Program Documentation EN7408SS.
tions. Southern Company Services, Inc., Technical
Engineering Systems, 64 Perimeter Center
f. Source-Receptor Relationship East, Atlanta, GA.
MTDDIS treats up to 10 point sources.
Up to three rectangular receptor grids may Availability
be specified by the user. The SCSTER model and user’s manual are
available at no charge on a limited basis
g. Plume Behavior through Southern Company Services. The
Briggs (1971, 1972) plume rise formulas are computer code may be provided on a
used. diskette. Requests should be directed to: Mr.
If plume height exceeds mixing height, Stanley S. Vasa, Senior Environmental Spe-
ground level concentration is assumed zero. cialist, Southern Company Services, P.O.
Fumigation and downwash are not treated. Box 2625, Birmingham, AL 35202.

h. Horizontal Winds Abstract


Wind speeds and wind directions at each SCSTER is a modified version of the EPA
station are first corrected for release height. CRSTER model. The primary distinctions of
Speed conversions are based on power law SCSTER are its capability to consider mul-
variation and direction conversions are tiple sources that are not necessarily collo-
based on linear height dependence as rec- cated, its enhanced receptor specifications,
ommended by Irwin (1979b). its variable plume height terrain adjustment
Converted wind speeds and wind directions procedures and plume distortion from direc-
are then weighted according to the algo- tional wind shear.

420
Environmental Protection Agency Pt. 51, App. W
a. Recommendations for Regulatory Use d. Type of Model
SCSTER can be used if it can be dem- SCSTER is a Gaussian plume model.
onstrated to estimate concentrations equiva-
lent to those provided by the preferred model e. Pollutant Types
for a given application. SCSTER must be ex- SCSTER may be used to model primary
ecuted in the equivalent mode. pollutants. Settling and deposition are not
SCSTER can be used on a case-by-case treated.
basis in lieu of a preferred model if it can be
demonstrated, using the criteria in section f. Source-Receptor Relationship
3.2 of appendix W, that SCSTER is more ap- SCSTER can handle up to 60 separate
propriate for the specific application. In this stacks at varying locations and up to 600 re-
case the model options/modes which are ceptors, including up to 15 receptor rings.
most appropriate for the application should User input topographic elevation for each
be used. receptor is used.
b. Input Requirements g. Plume Behavior
Source data requirements are: emission SCSTER uses Briggs (1969, 1971, 1972) final
rate, stack gas exit velocity, stack gas tem- plume rise formulas.
perature, stack exit diameter, physical stack Transitional plume rise is optional.
height, elevation of stack base, and coordi- SCSTER contains options to incorporate
nates of stack location. The variable emis- wind directional shear with a plume distor-
sion data can be monthly or annual aver- tion method described in appendix A of the
ages. User’s Guide.
Meteorological data requirements are: SCSTER provides four terrain adjustments
hourly surface weather data from the EPA including the CRSTER full terrain height ad-
meteorological preprocessor program. justment and a user-input, stability-depend-
Preprocessor output includes hourly stabil- ent plume path coefficient adjustment for re-
ity class wind direction, wind speed, tem- ceptors above stack height.
perature, and mixing height. Actual ane-
mometer height (a single value) is optional. h. Horizontal Winds
Wind speed profile exponents (one for each Wind speeds are corrected for release
stability class) are optional. height based on power law exponents from
Receptor data requirements are: cartesian DeMarrais (1959), different exponents for dif-
coordinates and elevations of individual re- ferent stability classes; default reference
ceptors; distances of receptor rings, with ele- height of 7m. Default exponents are 0.10, 0.15,
vation of each receptor; receptor grid net- 0.20, 0.25, 0.30, and 0.30 for stability classes A
works, with elevation of each receptor. through F, respectively.
Any combination of the three receptor Steady-state wind is assumed within a
input types may be used to consider up to 600 given hour.
receptor locations. Optional consideration of plume distortion
due to user-input, stability-dependent wind-
c. Output direction shear gradients.
Printed output includes: i. Vertical Wind Speed
Highest and second highest concentrations
for the year at each receptor for averaging Vertical wind speed is assumed equal to
times of 1-, 3-, and 24-hours, a user-selected zero.
averaging time which may be 2–12 hours, and
j. Horizontal Dispersion
a 50 high table for 1-, 3-, and 24-hours;
Annual arithmetic average at each recep- Rural dispersion coefficients from Turner
tor; and the highest 1-hour and 24-hour con- (1969) are used.
centrations over the receptor field for each Six stability classes are used.
day considered.
Optional tables of source contributions of k. Vertical Dispersion
individual point sources at up to 20 receptor Rural dispersion coefficients from Turner
locations for each averaging period; (1969) are used.
Optional magnetic tape output in either bi- Six stability classes are used.
nary or fixed block format includes: An optional test for plume height above
All 1-hour concentrations. mixing height before terrain adjustment is
Optional card/disk output includes for each included.
receptor:
l. Chemical Transformation
Receptor coordinates; receptor elevation;
highest and highest, second-highest, 1-, 3-, Chemical transformations are treated
and 24-hour concentrations; and annual aver- using exponential decay. Half-life is input by
age concentration. the user.

421
Pt. 51, App. W 40 CFR Ch. I (7–1–97 Edition)
m. Physical Removal with hourly, monthly or annual averaging
times, chemically reactive or non-reactive
Physical removal is treated using expo-
gases or particulate emissions for stationary
nential decay. Half-life is input by the user.
or roadway sources.
n. Evaluation Studies
b. Input Requirements
Londergan, R., D. Minott, D. Wackter, T.
Data may be input directly from an exter-
Kincaid and D. Bonitata, 1983. Evaluation of
nal source (e.g., GIS file) or interactively.
Rural Air Quality Simulation Models. EPA
The model provides the option to use default
Publication No. EPA–450/4–83–003. U.S. Envi-
values when input parameters are unavail-
ronmental Protection Agency, Research Tri-
able.
angle Park, NC.
PANACHE user environment integrates
B.11 PANACHE the pre- and post-processor with the solver.
The calculations can be done interactively
Reference or in batch mode. An inverse scheme is pro-
vided to estimate missing data from a few
Transoft Group, 1994. User’s Guide of
measured values of the wind.
Fluidyn-PANACHE, a Three-Dimensional
Terrain data requirements:
Deterministic Simulation of Pollutants Dis-
• Location, surface roughness estimates,
persion Model for Complex Terrain; Cary,
and altitude contours.
North Carolina.
• Location and dimensions of obstacles,
Availability forests, fields, and water bodies.
Source data requirements:
For a cost to be negotiated with the model For all types of sources, the exit tempera-
developer, the computer code is available ture and plume mass flow rates and con-
from: Transoft US, Inc., 818 Reedy Creek centration of each of the pollutants are re-
Road, Cary, NC 27513–3307; Phone: (919) 380– quired. External sources require mass flow
7500, Fax: (919) 380–7592. rate. For roadways, estimated traffic volume
and vehicular emissions are required.
Abstract
Meteorological data requirements:
PANACHE is an Eulerian (and Lagrangian Hourly stability class, wind direction, wind
for particulate matter), 3-dimensional finite speed, temperature, cloud cover, humidity,
volume fluid mechanics code designed to and mixing height data with lapse rate below
simulate continuous and short-term pollu- and above it.
tion dispersion in the atmosphere, in simple Primary meteorological variables avail-
or complex terrain. For single or multiple able from the National Weather Service can
sources, pollutant emissions from stack, be processed using PCRAMMET (see section
point, area, volume, general sources and dis- 9.3.3.2 of appendix W) to an input file.
tant sources are treated. The model auto- Data required at the domain boundary:
matically treats obstacles, effects of vegeta- Wind profile (uniform, log or power law),
tion and water bodies, the effects of vertical depending on the terrain conditions (e.g.,
temperature stratification on the wind and residential area, forest, sea, etc.).
diffusion fields, and turbulent shear flows Chemical source data requirements:
caused by atmospheric boundary layer or A database of selected species with specific
terrain effects. The code solves Navier heats and molecular weights can be extended
Stokes equations in a curvilinear mesh es- by the user. For heavy gases the database in-
pousing the terrain and obstacles. A 2nd cludes a compressibility coefficients table.
order resolution helps keep the number of Solar reflection:
cells limited in case of shearing flow. An ini- For natural convection simulation with
tial wind field is computed by using a low wind on a sunny day, approximate values
Lagrangian multiplier to interpolate wind of temperature for fields, forests, water bod-
data collected on site. The mesh generator, ies, shadows and their variations with the
the solver and the numerical schemes have time of the day are determined automati-
been adopted for atmospheric flows with or cally.
without chemical reactions. The model code
operates on any workstation or IBM—com- c. Output
patible PC (486 or higher). Gaussian and puff Printed output option: pollutant con-
modes are available in PANACHE for fast, centration at receptor points, and listing of
preliminary simulation. input data (terrain, chemical, weather, and
source data) with turbulence and precision
a. Recommendations for Regulatory Use
control data.
On a case-by-case basis, PANACHE may be Graphical output includes: In 3-dimen-
appropriate for the following types of situa- sional perspective or in any crosswind, down-
tions: industrial or urban zone on a flat or wind or horizontal plane: wind velocity, pol-
complex terrain, transport distance from a lutant concentration, 3-dimensional
few meters to 50km, continuous releases isosurface. The profile of concentration can

422
Environmental Protection Agency Pt. 51, App. W
be obtained along any line on the terrain. Emissions are based on traffic volume and
The concentration contours can be either in- emission factors.
stantaneous or time integrated for the emis- • Odor Dispersion: Identifies odor sources
sion from a source or a source combination. for waste water plants.
A special utility is included to help prepare • Radon Dispersion: Simulates natural
a report or a video animation. The user can radon accumulation in valleys and mine en-
select images, put in annotations, or do ani- vironments.
mation. PANACHE may also be used in emergency
planning and management for episodic emis-
d. Type of Model sions, and fire and soot spread in forested
and urban areas or from combustible pools.
The model uses an Eulerian (and
Lagrangian for particulate matter) 3-dimen- f. Source-Receptor Relationship
sional finite volume model solving full
Navier-Stokes equations. The numerical dif- Simultaneous use of multiple kinds of
fusion is low with appropriate turbulence sources at user defined locations. Any num-
models for building wakes. A second order ber of user defined receptors can identify
resolution may be sought to limit the diffu- pollutants from each source individually.
sion. Gaussian and puff modes are available. g. Plume Behavior
The numerical scheme is self adaptive for
the following situations: The options influencing the behavior are
• A curvilinear mesh or a chopped Carte- full gravity, Boussinesq approximation or no
sian mesh is generated automatically or gravity.
manually; h. Horizontal Winds
• Thermal and gravity effects are simu-
lated by full gravity (heavy gases), no grav- Horizontal wind speed approximations are
ity (well mixed light gases at ambient tem- made only at the boundaries based on Na-
perature), and Boussinesq approximation tional Weather Service data. Inside the do-
methods; main of interest, full Navier-Stokes resolu-
• K-diff, K-e or a boundary layer turbu- tion with natural viscosity is used for 3-di-
lence models are used for turbulence calcula- mensional terrain and temperature depend-
tions. The flow behind obstacles such as ent wind field calculation.
buildings, is calculated by using a modified i. Vertical Wind Speed
K-e.
• For heavy gases, a 3-dimensional heat Vertical wind speed approximations are
conduction from the ground and a stratifica- made only at the boundaries based on Na-
tion model for heat exchange from the at- tional Weather Service data. The domain of
mosphere are used (with anisotropic turbu- interest is treated as for horizontal winds.
lence).
j. Horizontal Dispersion
• If local wind data are available, an ini-
tial wind field with terrain effects can be Diffusion is calculated using appropriate
computed using a Lagrangian multiplier, turbulence models. A 2nd order solution for
which substantially reduces computation shearing flow can be sought when the num-
time. ber of meshes is limited between obstacles.

e. Pollutant Types k. Vertical Dispersion

• Scavenging, Acid Rain: A module for Dispersion by full gravity unless


water droplets traveling through a plume Boussinesq approximation or no gravity re-
considers the absorption and de-absorption quested. Vertical dispersion is treated as
effects of the pollutants by the droplet. above for horizontal dispersion.
Evaporation and chemical reactions with l. Chemical Transformation
gases are also taken into account.
• Visibility: Predicts plume visibility and PANCHEM, an atmospheric chemistry
surface deposition of aerosol. module for chemical reactions, is available.
• Particulate matter: Calculates settling Photochemical reactions are used for tropo-
and dry deposition of particles based on a spheric ozone calculations.
Probability Density Function (PDF) of their
m. Physical Removal
diameters. The exchange of mass, momen-
tum and heat between particles and gas is Physical removal is treated using dry dep-
treated with implicit coupling procedures. osition coefficients
• Ozone formation and dispersion: The pho-
tochemical model computes ozone formation n. Evaluation Studies
and dispersion at street level in the presence Goldwire, H.C. Jr, T.G. McRae, G.W. John-
of sunlight. son, D.L. Hipple, R.P. Koopman, J.W.
• Roadway Pollutants: Accounts for heat McClure, L.K. Morris and R.T. Cederhall,
and turbulence due to vehicular movement. 1985. Desert Tortoise Series Data Report: 1983

423
Pt. 51, App. W 40 CFR Ch. I (7–1–97 Edition)
Pressurized Ammonia Spills. UCID 20562, ground; horizontal views through the plume
Lawrence Livermore National Laboratory; with white, gray and black viewing back-
Livermore, California. grounds; and horizontal views along the axis
Green, S.R., 1992. Modeling Turbulent Air of the plume with a sky viewing background.
Flow in a Stand of Widely Spaced Trees, The
PHOENICS Journal of Computational Fluid a. Recommendations for Regulatory Use
Dynamics and Its Applications, 5: 294–312. The Plume Visibility Model (PLUVUE II)
Gryning, S.E. and E. Lyck, 1984. Atmos- may be used on a case-by-case basis as a
pheric Dispersion from Elevated Sources in third level screening model. When applying
an Urban Area: Comparison Between Tracer PLUVUE II, the following precautions
Experiments and Model Calculations. Jour- should be taken:
nal of Climate and Applied Meteorology, 23: 1. Treat the optical effects of NO2 and par-
651–660. ticles separately as well as together to avoid
Havens, J., T. Spicer, H. Walker and T. cancellation of NO2 absorption with particle
Williams, 1995. Validation of Mathematical scattering.
Models Using Wind-Tunnel Data Sets for 2. Examine the visual impact of the plume
Dense Gas Dispersion in the Presence of Ob- in 0.1 (or 0), 0.5, and 1.0 times the expected
stacles. University of Arkansas, 8th Inter- level of particulate matter in the back-
national Symposium-Loss Prevention and ground air.
Safety Promotion in the Process Industries; 3. Examine the visual impact of the plume
Antwerp, Belgium. over the full range of observer-plume sun an-
McQuaid, J. (ed), 1985. Heavy Gas Disper- gles.
sion Trials at Thorney Island. Proc. of a
4. The user should consult the appropriate
Symposium held at the University of Shef-
Federal Land Manager when using PLUVUE
field, Great Britain.
II to assess visibility impacts in a Class I
Pavitskiy, N.Y., A.A. Yakuskin and S.V.
area.
Zhubrin, 1993. Vehicular Exhaust Dispersion
Around Group of Buildings. The PHOENICS b. Input Requirements
Journal of Computational Fluid Dynamics
and Its Applications, 6: 270–285. Source data requirements are: location and
Tripathi, S., 1994. Evaluation of Fluidyn- elevation; emission rates of SO2, NOX, and
PANACHE on Heavy Gas Dispersion Test particulates; flue gas flow rate, exit velocity,
Case. Seminar on Evaluation of Models of and exit temperature; flue gas oxygen con-
Heavy Gas Dispersion Organized by Euro- tent; properties (including density, mass me-
pean Commission; Mol, Belgium. dian and standard geometric deviation of ra-
dius) of the emitted aerosols in the accumu-
B.12 Plume Visibility Model (PLUVUE II) lation (0.1–1.0µm) and coarse (1.0–10.µm) size
modes; and deposition velocities for SO2,
Reference NOX, coarse mode aerosol, and accumula-
Environmental Protection Agency, 1992. tions mode aerosol.
User’s Manual for the Plume Visibility Meteorological data requirements are: sta-
Model, PLUVUE II (Revised). EPA Publica- bility class, wind direction (for an observer-
tion No. EPA–454/B–92–008, (NTIS PB93– based run), wind speed, lapse rate, air tem-
188233). U.S. Environmental Protection perature, relative humidity, and mixing
Agency, Research Triangle Park, NC. height.
Other data requirements are: ambient
Availability background concentrations of NOX, NO2, O3,
This model code is available on the Sup- and SO2, and background visual range of sul-
port Center for Regulatory Air Models Bul- fate and nitrate concentrations.
letin Board System and also on diskette (as Receptor (observer) data requirements are:
PB 90–500778) from the National Technical location, terrain elevation at points along
Information Service (see section B.0). plume trajectory, white, gray, and black
viewing backgrounds, the distance from the
Abstract observer to the terrain observed behind the
plume.
The Plume Visibility Model (PLUVUE II)
is used for estimating visual range reduction c. Output
and atmospheric discoloration caused by
plumes consisting of primary particles, ni- Printed output includes plume concentra-
trogen oxides and sulfur oxides emitted from tions and visual effects at specified down-
a single emission source. PLUVUE II uses wind distances for calculated or specified
Gaussian formulations to predict transport lines of sight.
and dispersion. The model includes chemical
d. Type of Model
reactions, optical effects and surface deposi-
tion. Four types of optics calculations are PLUVUE II is a Gaussian plume model.
made: horizontal and non-horizontal views Visibility impairment is quantified once the
through the plume with a sky viewing back- spectral light intensity has been calculated

424
Environmental Protection Agency Pt. 51, App. W
for the specific lines of sight. Visibility im- Bergstrom, R.W., Seigneur, C.D. Johnson
pairment includes visual range reduction, and L.W. Richards, 1984. Measurements and
plume contrast, relative coloration of a Simulations of the Visual Effects of Particu-
plume to its viewing background, and plume late Plumes. Atmospheric Environment,
perceptibility due to its contrast and color 18(10): 2231–2244.
with respect to a viewing background. Seigneur, C., R.W. Bergstrom and A.B.
Hudischewskyj, 1982. Evaluation of the EPA
e. Pollutant Types PLUVUE Model and the ERT Visibility
PLUVUE II treats NO, NO2, SO2, H2SO4, Model Based on the 1979 VISTTA Data Base.
HNO3, O3, primary and secondary particles to EPA Publication No. EPA–450/4–82–008. U.S.
calculate effects on visibility. Environmental Protection Agency, Research
Triangle Park, NC.
f. Source Receptor Relationship White, W.H., C. Seigneur, D.W. Heinold,
M.W. Eltgroth, L.W. Richards, P.T. Roberts,
For performing the optics calculations at P.S. Bhardwaja, W.D. Conner and W.E. Wil-
selected points along the plume trajectory, son, Jr, 1985. Predicting the Visibility of
PLUVUE II has two modes: plume based and Chimney Plumes: An Inter-comparison of
observer based calculations. The major dif- Four Models with Observations at a Well-
ference is the orientation of the viewer to Controlled Power Plant. Atmospheric Envi-
the source and the plume. ronment, 19: 515–528.
g. Plume Behavior B.13 Point, Area, Line Source Algorithm (PAL–
Briggs (1969, 1971, 1972) final plume rise DS)
equations are used. Reference
h. Horizontal Winds Petersen, W.B, 1978. User’s Guide for PAL—
A Gaussian-Plume Algorithm for Point,
User-specified wind speed (and direction
Area, and Line Sources. EPA Publication No.
for an observer-based run) are assumed con-
EPA–600/4–78–013. Office of Research and De-
stant for the calculation.
velopment, Research Triangle Park, NC.
i. Vertical Wind Speed (NTIS No. PB 281306)
Rao, K.S. and H.F. Snodgrass, 1982. PAL–
Vertical wind speed is assumed equal to DS Model: The PAL Model Including Deposi-
zero. tion and Sedimentation. EPA Publication
No. EPA–600/8–82–023. Office of Research and
j. Horizontal Dispersion Development, Research Triangle Park, NC.
Constant, uniform (steady-state) wind is (NTIS No. PB 83–117739)
assumed for each hour. Straight line plume
transport is assumed to all downwind dis- Availability
tances. The computer code is available on diskette
(as PB 90–500802) from the National Tech-
k. Vertical Dispersion nical Information Service (see section B.0).
Rural dispersion coefficients from Turner
Abstract
(1969) are used, with no adjustment for sur-
face roughness. Six stability classes are PAL–DS is an acronym for this point, area,
used. and line source algorithm and is a method of
estimating short-term dispersion using
l. Chemical Transformation Gaussian-plume steady-state assumptions.
The chemistry of NO, NO2, O3, OH, O(1D), The algorithm can be used for estimating
SO2, HNO3, and H2SO4 is treated by means of concentrations of non-reactive pollutants at
nine reactions. Steady state approximations 99 receptors for averaging times of 1 to 24
are used for radicals and for the NO/NO2/O3 hours, and for a limited number of point,
reactions. area, and line sources (99 of each type). This
algorithm is not intended for application to
m. Physical Removal entire urban areas but is intended, rather, to
assess the impact on air quality, on scales of
Dry deposition of gaseous and particulate tens to hundreds of meters, of portions of
pollutants is treated using deposition veloci- urban areas such as shopping centers, large
ties. parking areas, and airports. Level terrain is
assumed. The Gaussian point source equa-
n. Evaluation Studies
tion estimates concentrations from point
Bergstrom, R.W., C. Seigneur, B.L. Babson, sources after determining the effective
H.Y. Holman and M.A. Wojcik, 1981. Com- height of emission and the upwind and cross-
parison of the Observed and Predicted Visual wind distance of the source from the recep-
Effects Caused by Power Plant Plumes. At- tor. Numerical integration of the Gaussian
mospheric Environment, 15: 2135–2150. point source equation is used to determine

425
Pt. 51, App. W 40 CFR Ch. I (7–1–97 Edition)
concentrations from the four types of line c. Output
sources. Subroutines are included that esti-
Printed output includes:
mate concentrations for multiple lane line
Hourly concentration and deposition flux
and curved path sources, special line sources
for each source type at each receptor; and
(line sources with endpoints at different
Average concentration for up to 24 hours
heights above ground), and special curved
for each source type at each receptor.
path sources. Integration over the area
source, which includes edge effects from the d. Type of Model
source region, is done by considering finite
line sources perpendicular to the wind at in- PAL–DS is a Gaussian plume model.
tervals upwind from the receptor. The cross- e. Pollutant Types
wind integration is done analytically; inte-
gration upwind is done numerically by suc- PAL–DS may be used to model non-reac-
cessive approximations. tive pollutants.
The PAL–DS model utilizes Gaussian
f. Source-Receptor Relationships
plume-type diffusion-deposition algorithms
based on analytical solutions of a gradient- Up to 99 sources of each of 6 source types:
transfer model. The PAL–DS model can treat point, area, and 4 types of line sources.
deposition of both gaseous and suspended Source and receptor coordinates are
particulate pollutants in the plume since uniquely defined.
gravitational settling and dry deposition of Unique stack height for each source.
the particles are explicitly accounted for. Coordinates of receptor locations are user
The analytical diffusion-deposition expres- defined.
sions listed in this report in the limit when
pollutant settling and deposition velocities g. Plume Behavior
are zero, they reduce to the usual Gaussian Briggs final plume rise equations are used.
plume diffusion algorithms in the PAL Fumigation and downwash are not treated.
model. If plume height exceeds mixing height,
concentrations are assumed equal to zero.
a. Recommendations for Regulatory Use Surface concentrations are set to zero
PAL–DS can be used if it can be dem- when the plume centerline exceeds mixing
onstrated to estimate concentrations equiva- height.
lent to those provided by the preferred model
h. Horizontal Winds
for a given application. PAL–DS must be ex-
ecuted in the equivalent mode. User-supplied hourly wind data are used.
PAL–DS can be used on a case-by-case Constant, uniform (steady-state) wind is
basis in lieu of a preferred model if it can be assumed within each hour. Wind is assumed
demonstrated, using the criteria in section to increase with height.
3.2, that PAL–DS is more appropriate for the
i. Vertical Wind Speeds
specific application. In this case the model
options/modes which are most appropriate Assumed equal to zero.
for the application should be used.
j. Horizontal Dispersion
b. Input Requirements Rural dispersion coefficients from Turner
Source data: point-sources—emission rate, (1969) are used with no adjustments made for
physical stack height, stack gas tempera- surface roughness.
ture, stack gas velocity, stack diameter, Six stability classes are used.
stack gas volume flow, coordinates of stack, Dispersion coefficients (Pasquill-Gifford)
initial σy and σz; area sources—source are assumed based on a 3cm roughness
strength, size of area source, coordinates of height.
S.W. corner, and height of area source; and
k. Vertical Dispersion
line sources—source strength, number of
lanes, height of source, coordinates of end Six stability classes are used.
points, initial σy and σz, width of line source, Rural dispersion coefficients from Turner
and width of median. Diurnal variations in (1969) are used; no further adjustments are
emissions are permitted. When applicable, made for variation in surface roughness,
the settling velocity and deposition velocity transport or averaging time.
are also permitted. Multiple reflection is handled by summa-
Meteorological data: wind profile expo- tion of series until the vertical standard de-
nents, anemometer height, wind direction viation equals 1.6 times mixing height. Uni-
and speed, stability class, mixing height, air form vertical mixing is assumed thereafter.
temperature, and hourly variations in emis-
sion rate. l. Chemical Transformation
Receptor data: receptor coordinates. Not treated.

426
Environmental Protection Agency Pt. 51, App. W
m. Physical Removal es, photolytic rate constants, and plume
depths or stability classes.
PAL–DS can treat deposition of both gase-
ous and suspended particulates in the plume Receptor data requirements are: downwind
since gravitational settling and dry deposi- distances or travel times at which calcula-
tion of the particles are explicitly accounted tions are to be made.
for. Initial concentration of all species is re-
quired, and the specification of downwind
n. Evaluation Studies ambient concentrations to be entrained by
the plume is optional.
None Cited.
c. Output
B.14 Reactive Plume Model (RPM–IV)
Short-term concentrations of primary and
Reference secondary pollutants at either user specified
Environmental Protection Agency, 1993. time increments, or user specified downwind
Reactive Plume Model IV (RPM–IV) User’s distances.
Guide. EPA Publication No. EPA–454/B–93–
012. U.S. Environmental Protection Agency d. Type of Model
(ESRL), Research Triangle Park, NC. (NTIS Reactive Gaussian plume model.
No. PB 93–217412)
e. Pollutant Types
Availability
Currently, using the Carbon Bond Mecha-
The above report and model computer code nism (CBM–IV), 34 species are simulated (82
are available on the Support Center for Reg- reactions), including NO, NO2, O3, SO2, SO4,
ulatory Air Models Bulletin Board System. five categories of reactive hydrocarbons, sec-
The model code is also available on diskette ondary nitrogen compounds, organic
(as PB 96–502026) from the National Tech- aerosols, and radical species.
nical Information Service (see section B.0).
f. Source-Receptor Relationships
Abstract
Single point source.
The Reactive Plume Model, RPM–IV, is a Single area or volume source.
computerized model used for estimating
Multiple sources can be simulated if they
short-term concentrations of primary and
are lined up along the wind trajectory.
secondary reactive pollutants resulting from
Predicted concentrations are obtained at a
single or, in some special cases, multiple
user specified time increment, or at user
sources if they are aligned with the mean
specified downwind distances.
wind direction. The model is capable of sim-
ulating the complex interaction of plume g. Plume Behavior
dispersion and non-linear photochemistry. If
Carbon Mechanism IV (CBM–IV) is used, Briggs (1971) plume rise equations are used.
emissions must be disaggregated into carbon
bond classes prior to model application. The h. Horizontal Winds
model can be run on a mainframe computer, User specifies wind speeds as a function of
workstation, or IBM-compatible PC with at time.
least 2 megabytes of memory. A major fea-
ture of RPM–IV is its ability to interface i. Vertical Wind Speed
with input and output files from EPA’s Re-
Not treated.
gional Oxidant Model (ROM) and Urban
Airshed Model (UAM) to provide an inter- j. Horizontal Dispersion
nally consistent set of modeled ambient con-
centrations for various pollutant species. User specified plume widths, or user may
specify stability and widths will be computed
a. Recommendations for Regulatory Use using Turner (1969).
There is no specific recommendation at the k. Vertical Dispersion
present time. RPM–IV may be used on a
case-by-case basis. User specified plume depths, or user may
specify stability in which case depths will be
b. Input Requirements calculated using Turner (1969). Note that ver-
Source data requirements are: emission tical uniformity in plume concentration is
rates, name, and molecular weight of each assumed.
species of pollutant emitted; ambient pres-
l. Chemical Transformation
sure, ambient temperature, stack height,
stack diameter, stack exit velocity, stack RPM–IV has the flexibility of using any
gas temperature, and location. user input chemical kinetic mechanism. Cur-
Meteorological data requirements are: rently it is run using the chemistry of the
wind speeds, plume widths or stability class- Carbon Bond Mechanism, CBM–IV (Gery et

427
Pt. 51, App. W 40 CFR Ch. I (7–1–97 Edition)
al., 1988). The CBM–IV mechanism, as incor- standard NWS surface and upper air mete-
porated in RPM–IV, utilizes an updated sim- orological data through the RAMMET
ulation of PAN chemistry that includes a preprocessor.
peroxy-peroxy radical termination reaction, Receptor data: coordinates for each recep-
significant when the atmosphere is NOX-lim- tor.
ited (Gery et al., 1989). As stated above, the
current CBM–IV mechanism accommodates c. Output
34 species and 82 reactions focusing primarily
Printed output includes the MPTER model
on hydrocarbon/nitrogen oxides and ozone
output as well as: special shoreline fumiga-
photochemistry.
tion applicability report for each day and
m. Physical Removal source; high-five tables on the standard out-
put with ‘‘F’’ designation next to the con-
Not treated. centration if that averaging period includes
a fumigation event.
n. Evaluation Studies
Stewart, D.A. and M–K Liu, 1981. Develop- d. Type of Model
ment and Application of a Reactive Plume SDM is hybrid Gaussian model.
Model. Atmospheric Environment, 15: 2377–
2393. e. Pollutant Types
B.15 Shoreline Dispersion Model (SDM) SDM may be used to model primary pollut-
ants. Settling and deposition are not treated.
Reference
PEI Associates, 1988. User’s Guide to SDM– f. Source-Receptor Relationships
A Shoreline Dispersion Model. EPA Publica- SDM applies user-specified locations of
tion No. EPA–450/4–88–017. U.S. Environ- stationary point sources and receptors. User
mental Protection Agency, Research Tri- input stack height, shoreline orientation and
angle Park, NC. (NTIS No. PB 89–164305) source characteristics for each source. No
topographic elevation is input; flat terrain is
Availability assumed.
The model code is available on the Support
Center for Regulatory Air Models Bulletin g. Plume Behavior
Board System (see section B.0). SDM uses Briggs (1975) plume rise for final
Abstract rise. SDM does not treat stack tip or build-
ing downwash.
SDM is a hybrid multi-point Gaussian dis-
persion model that calculates source impact h. Horizontal Winds
for those hours during the year when fumiga-
Constant, uniform (steady-state) wind is
tion events are expected using a special fu-
assumed for an hour. Straight line plume
migation algorithm and the MPTER regu-
transport is assumed to all downwind dis-
latory model for the remaining hours (see
tances. Separate wind speed profile expo-
appendix A).
nents (EPA, 1980) for both rural and urban
a. Recommendations for Regulatory Use cases are assumed.

SDM may be used on a case-by-case basis i. Vertical Wind Speed


for the following applications:
Vertical wind speed is assumed equal to
• Tall stationary point sources located at a
zero.
shoreline of any large body of water;
• Rural or urban areas; j. Horizontal Dispersion
• Flat terrain;
• Transport distances less than 50 km; For the fumigation algorithm coefficients
• 1-hour to 1-year averaging times. based on Misra (1980) and Misra and McMil-
lan (1980) are used for plume transport in sta-
b. Input Requirements ble air above TIBL and based on Lamb (1978)
Source data: location, emission rate, phys- for transport in the unstable air below the
ical stack height, stack gas exit velocity, TIBL. An effective horizontal dispersion co-
stack inside diameter, stack gas tempera- efficient based on Misra and Onlock (1982) is
ture and shoreline coordinates. used. For nonfumigation periods, algorithms
Meteorological data: hourly values of contained in the MPTER model are used (see
mean wind speed within the Thermal Inter- appendix A).
nal Boundary Layer (TIBL) and at stack
k. Vertical Dispersion
height; mean potential temperature over
land and over water; over water lapse rate; For the fumigation algorithm, coefficients
and surface sensible heat flux. In addition to based on Misra (1980) and Misra and McMil-
these meteorological data, SDM access lan (1980) are used.

428
Environmental Protection Agency Pt. 51, App. W
l. Chemical Transformation b. Input Requirements
Chemical transformation is not included in Source data requirements are: for point,
the fumigation algorithm. building or area sources, location, elevation,
total emission rate (optionally classified by
m. Physical Removal gravitational settling velocity) and decay
coefficient; for stack sources, stack height,
Physical removal is not explicitly treated.
effluent temperature, effluent exit velocity,
n. Evaluation Studies stack radius (inner), actual volumetric flow
rate, and ground elevation (optional); for
Environmental Protection Agency, 1987. building sources, height, length and width,
Analysis and Evaluation of Statistical and orientation; for area sources, char-
Coastal Fumigation Models. EPA Publica- acteristic vertical dimension, and length,
tion No. EPA–450/4–87–002. U.S. Environ- width and orientation.
mental Protection Agency, Research Tri- Meteorological data requirements are:
angle Park, NC. (NTIS PB 87–175519) wind speed and measurement height, wind
profile exponents, wind direction, standard
B.16 SHORTZ deviations of vertical and horizontal wind di-
rections, (i.e., vertical and lateral turbulent
Reference intensities), mixing height, air temperature,
and vertical potential temperature gradient.
Bjorklund, J.R. and J.F. Bowers, 1982.
Receptor data requirements are: coordi-
User’s Instructions for the SHORTZ and
nates, ground elevation.
LONGZ Computer Programs, Volumes I and
II. EPA Publication No. EPA–903/9–82–004a c. Output
and b. U.S. Environmental Protection Agen-
cy, Region III, Philadelphia, PA. Printed output includes total concentra-
tion due to emissions from user-specified
Availability source groups, including the combined emis-
sions from all sources (with optional allow-
The computer code is available on the Sup- ance for depletion by deposition).
port Center for Regulatory Air Models Bul-
letin Board System and on diskette (as PB d. Type of Model
96–501986) from the National Technical Infor-
SHORTZ is a Gaussian plume model.
mation Service (see section B.0).
e. Pollutant Types
Abstract
SHORTZ may be used to model primary
SHORTZ utilizes the steady state bivariate pollutants. Settling and deposition of partic-
Gaussian plume formulation for both urban ulates are treated.
and rural areas in flat or complex terrain to
calculate ground-level ambient air con- f. Source-Receptor Relationships
centrations. The model can calculate 1-hour,
User specified locations for sources and re-
2-hour, 3-hour etc. average concentrations
ceptors are used.
due to emissions from stacks, buildings and
Receptors are assumed to be at ground
area sources for up to 300 arbitrarily placed
level.
sources. The output consists of total con-
centration at each receptor due to emissions g. Plume Behavior
from each user-specified source or group of
sources, including all sources. If the option Plume rise equations of Bjorklund and
for gravitational settling is invoked, analy- Bowers (1982) are used.
sis cannot be accomplished in complex ter- Stack tip downwash (Bjorklund and Bow-
rain without violating mass continuity. ers, 1982) is included.
All plumes move horizontally and will
a. Recommendations for Regulatory Use fully intercept elevated terrain.
Plumes above mixing height are ignored.
SHORTZ can be used if it can be dem- Perfect reflection at mixing height is as-
onstrated to estimate concentrations equiva- sumed for plumes below the mixing height.
lent to those provided by the preferred model Plume rise is limited when the mean wind
for a given application. SHORTZ must be ex- at stack height approaches or exceeds stack
ecuted in the equivalent mode. exit velocity.
SHORTZ can be used on a case-by-case Perfect reflection at ground is assumed for
basis in lieu of a preferred model if it can be pollutants with no settling velocity.
demonstrated, using the criteria in section Zero reflection at ground is assumed for
3.2, that SHORTZ is more appropriate for the pollutants with finite settling velocity.
specific application. In this case the model Tilted plume is used for pollutants with
options/modes which are most appropriate settling velocity specified. Buoyancy-in-
for the application should be used. duced dispersion (Briggs, 1972) is included.

429
Pt. 51, App. W 40 CFR Ch. I (7–1–97 Edition)
h. Horizontal Winds Abstract
Winds are assumed homogeneous and The Simple Line-Source Model is a simple
steady-state. steady-state Gaussian plume model which
Wind speed profile exponents are functions can be used to determine hourly (or half-
of both stability class and wind speed. De- hourly) averages of exhaust concentrations
fault values are specified in Bjorklund and within 100m from a roadway on a relatively
Bowers (1982). flat terrain. The model allows for plume rise
due to the heated exhaust, which can be im-
i. Vertical Wind Speed portant when the crossroad wind is very low.
The model also utilizes a new set of vertical
Vertical winds are assumed equal to zero. dispersion parameters which reflects the in-
fluence of traffic-induced turbulence.
j. Horizontal Dispersion
a. Recommendations for Regulatory Use
Horizontal plume size is derived from input
lateral turbulent intensities using adjust- The Simple Line-Source Model can be used
ments to plume height, and rate of plume if it can be demonstrated to estimate con-
growth with downwind distance specified in centrations equivalent to those provided by
Bjorklund and Bowers (1982). the preferred model for a given application.
The model must be executed in the equiva-
k. Vertical Dispersion lent mode.
The Simple Line-Source Model can be used
Vertical plume size is derived from input on a case-by-case basis in lieu of a preferred
vertical turbulent intensities using adjust- model if it can be demonstrated, using cri-
ments to plume height and rate of plume teria in section 3.2, that it is more appro-
growth with downwind distance specified in priate for the specific application. In this
Bjorklund and Bowers (1982). case the model options/modes which are
most appropriate for the application should
l. Chemical Transformation be used.
Chemical transformations are treated
b. Input Requirements
using exponential decay. Time constant is
input by the user. Source data requirements are: emission
rate per unit length per lane, the number of
m. Physical Removal lanes on each road, distances from lane cen-
ters to the receptor, source and receptor
Settling and deposition of particulates are
heights.
treated. Meteorological data requirements are:
n. Evaluation Studies buoyancy flux, ambient stability condition,
ambient wind and its direction relative to
Bjorklund, J.R. and J.F. Bowers, 1982. the road.
User’s Instructions for the SHORTZ and Receptor data requirements are: distance
LONGZ Computer Programs. EPA Publica- and height above ground.
tion No. EPA–903/9–82–004. EPA Environ-
mental Protection Agency, Region III, Phila- c. Output
delphia, PA. Printed output includes hourly or (half-
Wackter, D. and R. Londergan, 1984. Eval- hourly) concentrations at the receptor due
uation of Complex Terrain Air Quality Sim- to exhaust emission from a road (or a system
ulation Models. EPA Publication No. EPA– of roads by summing the results from re-
450/4–84–017. U.S. Environmental Protection peated model applications).
Agency, Research Triangle Park, NC.
d. Type of Model
B.17 Simple Line-Source Model The Simple Line-Source Model is a
Gaussian plume model.
Reference
Chock, D.P., 1980. User’s Guide for the Sim- e. Pollutant Types
ple Line-Source Model for Vehicle Exhaust The Simple Line-Source Model can be used
Dispersion Near a Road. Ford Research Lab- to model primary pollutants. Settling and
oratory, Dearborn, MI. deposition are not treated.

Availability f. Source-Receptor Relationship


Copies of the above reference are available The Simple Line-Source Model treats arbi-
without charge from: Dr. D.P. Chock, Ford trary location of line sources and receptors.
Research Laboratory, P.O. Box 2053; MD–
g. Plume Behavior
3083, Dearborn, MI 48121–2053. The short
model algorithm is contained in the User’s Plume-rise formula adequate for a heated
Guide. line source is used.

430
Environmental Protection Agency Pt. 51, App. W
h. Horizontal Winds elevated horizontal jet, a stack or elevated
vertical jet and an instantaneous volume
The Simple Line-Source Model uses user-
source. All sources except the evaporating
supplied hourly (or half-hourly) ambient
pool may be characterized as aerosols. Only
wind speed and direction. The wind measure-
one type of release can be processed in any
ments are from a height of 5 to 10m.
individual simulation. Also, the model simu-
i. Vertical Wind Speed lates only one set of meteorological condi-
tions; therefore direct application of the
Vertical wind speed is assumed equal to model over time periods longer than one or
zero. two hours is not recommended.
j. Dispersion Parameters a. Recommendations for use
Horizontal dispersion parameter is not The SLAB model should be used as a re-
used. fined model to estimate spatial and temporal
distribution of short-term ambient con-
k. Vertical Dispersion
centration (e.g., 1-hour or less averaging
A vertical dispersion parameter is used times) and the expected area of exposure to
which is a function of stability and wind- concentrations above specified threshold val-
road angle. Three stability classes are used: ues for toxic chemical releases where the re-
unstable, neutral and stable. The parameters lease is suspected to be denser than the am-
take into account the effect of traffic-gen- bient air.
erated turbulence (Chock, 1980).
b. Input Requirements
l. Chemical Transformation
The SLAB model is executed in the batch
Not treated. mode. Data are input directly from an exter-
nal input file. There are 29 input parameters
m. Physical Removal required to run each simulation. These pa-
Not treated. rameters are divided into 5 categories by the
user’s guide: source type, source properties,
n. Evaluation Studies spill properties, field properties, and mete-
Chock, D.P., 1978. A Simple Line-Source orological parameters. The model is not de-
Model for Dispersion Near Roadways. Atmos- signed to accept real-time meteorological
pheric Environment, 12: 823–829. data or convert units of input values. Chemi-
Sistla, G., P. Samson, M. Keenan and S.T. cal property data are not available within
Rao, 1979. A Study of Pollutant Dispersion the model and must be input by the user.
Near Highways. Atmospheric Environment, Some chemical and physical property data
13: 669–685. are available in the user’s guide.
Source type is chosen as one of the follow-
B.18 SLAB ing: evaporating pool release, horizontal jet
release, vertical jet or stack release, or in-
Reference: stantaneous or short duration evaporating
pool release.
Ermak, D.L., 1990. User’s Manual for
Source property data requirements are
SLAB: An Atmospheric Dispersion Model for
physical and chemical properties (molecular
Denser-than-Air Releases (UCRL–MA–105607),
weight, vapor heat capacity at constant
Lawrence Livermore National Laboratory.
pressure; boiling point; latent heat of vapor-
Availability ization; liquid heat capacity; liquid density;
saturation pressure constants), and initial
The computer code can be obtained from: liquid mass fraction in the release.
Energy Science and Technology Center, P.O. Spill properties include: source tempera-
Box 1020, Oak Ridge, TN 37830, Phone (615) ture, emission rate, source dimensions, in-
576–2606. stantaneous source mass, release duration,
The User’s Manual (as DE 91–008443) can be and elevation above ground level.
obtained from the National Technical Infor- Required field properties are: desired con-
mation Service. The computer code is also centration averaging time, maximum down-
available on the Support Center for Regu- wind distance (to stop the calculation), and
latory Air Models Bulletin Board System four separate heights at which the con-
(Public Upload/ Download Area; see section centration calculations are to be made.
B.0.) Meteorological parameter requirements
are: ambient measurement height, ambient
Abstract
wind speed at designated ambient measure-
The SLAB model is a computer model, PC- ment height, ambient temperature, surface
based, that simulates the atmospheric dis- roughness, relative humidity, atmospheric
persion of denser-than-air releases. The stability class, and inverse Monin-Obukhov
types of releases treated by the model in- length (optional, only used as an input pa-
clude a ground-level evaporating pool, an rameter when stability class is unknown).

431
Pt. 51, App. W 40 CFR Ch. I (7–1–97 Edition)
c. Output The model contains submodels for the
source characterization of evaporating pools,
No graphical output is generated by the
elevated vertical or horizontal jets, and in-
current version of this program. The output
stantaneous volume sources.
print file is automatically saved and must be
sent to the appropriate printer by the user g. Plume Behavior
after program execution. Printed output in-
cludes in tabular form: Plume trajectory and dispersion is based
Listing of model input data; on crosswind-averaged mass, species, energy,
Instantaneous spatially-averaged cloud pa- and momentum balance equations. Sur-
rameters—time, downwind distance, mag- rounding terrain is assumed to be flat and of
nitude of peak concentration, cloud dimen- uniform surface roughness. No obstacle or
sions (including length for puff-type simula- building effects are taken into account.
tions), volume (or mole) and mass fractions,
h. Horizontal Winds
downwind velocity, vapor mass fraction, den-
sity, temperature, cloud velocity, vapor frac- A power law approximation of the loga-
tion, water content, gravity flow velocities, rithmic velocity profile which accounts for
and entrainment velocities; stability and surface roughness is used.
Time-averaged cloud parameters—param-
eters which may be used externally to cal- i. Vertical Wind Speed
culate time-averaged concentrations at any Not treated.
location within the simulation domain (tab-
ulated as functions of downwind distance); j. Vertical Dispersion
Time-averaged concentration values at
plume centerline and at five off-centerline The crosswind dispersion parameters are
distances (off-centerline distances are mul- calculated from formulas reported by Mor-
tiples of the effective cloud half-width, gan et al. (1983), which are based on experi-
which varies as a function of downwind dis- mental data from several sources. The for-
tance) at four user-specified heights and at mulas account for entrainment due to at-
the height of the plume centerline. mospheric turbulence, surface friction, ther-
mal convection due to ground heating, dif-
d. Type of Model ferential motion between the air and the
cloud, and damping due to stable density
As described by Ermak (1989), transport stratification within the cloud.
and dispersion are calculated by solving the
conservation equations for mass, species, en- k. Horizontal Dispersion
ergy, and momentum, with the cloud being
modeled as either a steady-state plume, a The horizontal dispersion parameters are
transient puff, or a combination of both, de- calculated from formulas similar to those
pending on the duration of the release. In the described for vertical dispersion, also from
steady-state plume mode, the crosswind- the work of Morgan et al. (1983).
averaged conservation equations are solved
l. Chemical Transformation
and all variables depend only on the down-
wind distance. In the transient puff mode, The thermodynamics of the mixing of the
the volume-averaged conservation equations dense gas or aerosol with ambient air (in-
are solved, and all variables depend only on cluding water vapor) are treated. The rela-
the downwind travel time of the puff center tionship between the vapor and liquid frac-
of mass. Time is related to downwind dis- tions within the cloud is treated using the
tance by the height-averaged ambient wind local thermodynamic equilibrium approxi-
speed. The basic conservation equations are mation. Reactions of released chemicals
solved via a numerical integration scheme in with water or ambient air are not treated.
space and time.
m. Physical Removal
e. Pollutant Types
Not treated.
Pollutants are assumed to be non-reactive
and non-depositing dense gases or liquid- n. Evaluation Studies
vapor mixtures (aerosols). Surface heat Blewitt, D.N., J.F. Yohn and D.L. Ermak,
transfer and water vapor flux are also in- 1987. An Evaluation of SLAB and DEGADIS
cluded in the model. Heavy Gas Dispersion Models Using the HF
Spill Test Data. Proceedings, AIChE Inter-
f. Source-Receptor Relationships
national Conference on Vapor Cloud Model-
Only one source can be modeled at a time. ing, Boston, MA, November, pp. 56–80.
There is no limitation to the number of re- Ermak, D.L., S.T. Chan, D.L. Morgan and
ceptors; the downwind receptor distances are L.K. Morris, 1982. A Comparison of Dense Gas
internally-calculated by the model. The Dispersion Model Simulations with Burro
SLAB calculation is carried out up to the Series LNG Spill Test Results. J. Haz.
user-specified maximum downwind distance. Matls., 6: 129–160.

432
Environmental Protection Agency Pt. 51, App. W
Zapert, J.G., R.J. Londergan and H. This- b. Input Requirements
tle, 1991. Evaluation of Dense Gas Simula-
tion Models. EPA Publication No. EPA–450/4– Input data, including model options, mod-
90–018. U.S. Environmental Protection Agen- eling domain boundaries, boundary condi-
cy, Research Triangle Park, NC. tions, receptor locations, source locations,
and emission rates, may be entered inter-
B.19 WYNDvalley Model actively, or through existing template files
from a previous run. Meteorological data, in-
Reference cluding wind speeds, wind directions, rain
Harrison, Halstead, 1992. ‘‘A User’s Guide rates (optionally, for wet deposition calcula-
to WYNDvalley 3.11, an Eulerian-Grid Air- tions), and time of day and year, may be of
Quality Dispersion Model with Versatile arbitrary time increment (usually an hour)
Boundaries, Sources, and Winds,’’ WYNDsoft and are entered into the model through an
Inc., Mercer Island, WA. external meteorological data file. Option-
ally, users may specify diffusivities and
Availability upper boundary conditions for each time in-
Copies of the user’s guide and the execut- crement. Source emission rates may be con-
able model computer codes are available at a stant or modulated on a daily, weekly, and/
cost of $295.00 from: WYNDsoft, Incor- or seasonal basis.
porated, 6333 77th Avenue, Mercer Island, WA
98040, Phone: (206) 232–1819. c. Output
Output from WYNDvalley includes gridded
Abstract
contour maps of the highest pollutant con-
WYNDvalley 3.11 is a multi-layer (up to centrations at each time step and the high-
five vertical layers) Eulerian grid dispersion est and second-highest 24-hour average con-
model that permits users flexibility in defin- centrations. Output also includes the deposi-
ing borders around the areas to be modeled, tion patterns for wet, dry, and total fluxes of
the boundary conditions at these borders, the pollutants to the surface, integrated over
the intensities and locations of emissions the simulation period. A running ‘‘movie’’ of
sources, and the winds and diffusivities that the concentration patterns is displayed on
affect the dispersion of atmospheric pollut- the screen (with optional printout) as they
ants. The model’s output includes gridded evolve during the simulation. Output files
contour plots of pollutant concentrations for include tables of daily-averaged pollutant
the highest brief episodes (during any single concentrations at every modeled grid cell,
time step), the highest and second-highest and of hourly concentrations at up to eight
24-hour averages, averaged dry and wet depo-
specified receptors. Statistical analyses are
sition fluxes, and a colored ‘‘movie’’ showing
performed on the hourly and daily data to
evolving dispersal of pollutant concentra-
estimate the probabilities that specified lev-
tions, together with temporal plots of the
els will be exceeded more than once during
concentrations at specified receptor sites
and statistical inference of the probabilities an arbitrary number of days with similar
that standards will be exceeded at those weather.
sites. WYNDvalley is implemented on IBM
d. Type of Model
compatible microcomputers, with inter-
active data input and color graphics display. WYNDvalley is a three dimensional
Eulerian grid model.
a. Recommendations for Regulatory Use
WYNDvalley may be used on a case-by-case e. Pollutant Types
basis to estimate concentrations during val- WYNDvalley may be used to model any
ley stagnation periods of 24 hours or longer. inert pollutant.
Recommended inputs are listed below.
f. Source-Receptor Relationships
Variable Recommended value
Source and receptors may be located any-
Horizontal cell dimension ...... 250 to 500 meters. where within the user-defined modeling do-
Vertical layers ........................ 3 to 5. main. All point and area sources, or portions
Layer depth ............................ 50 to 100 meters.
of an area source, within a given grid cell are
Background (internal to Zero (background should be
model). added externally to model summed to define a representative emission
estimates). rate for that cell. Concentrations are cal-
Lateral meander velocity ....... Default. culated for each and every grid cell in the
Diffusivities ............................. Default. modeling domain. Up to eight grid cells may
Ventilation parameter (upper Default. be selected as receptors, for which time his-
boundary condition). tories of concentration and deposition fluxes
Dry deposition velocity .......... Zero (site-specific).
are determined, and probabilities of
Washout ratio ........................ Zero (site-specific).
exceedance are calculated.

433
Pt. 51, App. W 40 CFR Ch. I (7–1–97 Edition)
g. Plume Behavior Puget Sound Water Quality Authority, Se-
attle, WA.
Emissions for buoyant point sources are
Yoshida, C., 1990. A Comparison of
placed by the user in a grid cell which best
WYNDvalley Versions 2.12 and 3.0 with PM–
reflects the expected effective plume height
10 Measurements in Six Cities in the Pacific
during stagnation conditions. Five vertical
Northwest. Lane Regional Air Pollution Au-
layers are available to the user.
thority, Springfield, OR.
h. Horizontal Winds
B. REF References
During each time step in the model, the
Beals, G.A., 1971. A Guide to Local Disper-
winds are assumed to be uniform throughout
sion of Air Pollutants. Air Weather Service
the modeling domain. Numerical diffusion is
Technical Report #214 (April 1971).
minimized in the advection algorithm. To
Bjorklund, J.R. and J.F. Bowers, 1982.
account for terrain effects on winds and dis-
User’s Instructions for the SHORTZ and
persion, an ad hoc algorithm is employed in
LONGZ Computer Programs. EPA Publica-
the model to distribute concentrations near
tion No. EPA–903/9–82–004a and b. U.S. Envi-
boundaries.
ronmental Protection Agency, Region III,
i. Vertical Wind Speed Philadelphia, PA.
Briggs, G.A., 1969. Plume Rise. U.S. Atomic
Winds are assumed to be constant with Energy Commission Critical Review Series,
height. Oak Ridge National Laboratory, Oak Ridge,
TN. (NTIS No. TID–25075)
j. Horizontal Dispersion
Briggs, G.A., 1971. Some Recent Analyses
Horizontal eddy diffusion coefficients may of Plume Rise Observations. Proceedings of
be entered explicitly by the user at every the Second International Clean Air Congress,
time step. Alternatively, a default algorithm edited by H.M. Englund and W.T. Berry. Aca-
may be invoked to estimate these coeffi- demic Press, New York, NY.
cients from the wind velocities and their Briggs, G.A., 1972. Discussion on Chimney
variances. Plumes in Neutral and Stable Surroundings.
Atmospheric Environment, 6: 507–510.
k. Vertical Dispersion Briggs, G.A., 1974. Diffusion Estimation for
Vertical eddy diffusion coefficients and a Small Emissions. USAEC Report ATDL–106.
top-of-model boundary condition may be en- U.S. Atomic Energy Commission, Oak Ridge,
tered explicitly by the user at every time TN.
step. Alternatively, a default algorithm may Briggs, G.A., 1975. Plume Rise Predictions.
be invoked to estimate these coefficients Lectures on Air Pollution and Environ-
from the horizontal wind velocities and their mental Impact Analyses. American Meteoro-
variances, and from an empirical time-of-day logical Society, Boston, MA, pp. 59–111.
correction derived from temperature gra- Briggs, G.A., 1984. Plume Rise and Buoy-
dient measurements and Monin-Obukhov ancy Effects. Atmospheric Science and
similarities. Power Production, Darryl Randerson (Ed.).
DOE Report DOE/TIC–27601, Technical Infor-
l. Chemical Transformation mation Center, Oak Ridge, TN. (NTIS No.
DE84005177)
Chemical transformation is not explicitly
Carpenter, S.B., T.L. Montgomery, J.M.
treated by WYNDvalley.
Leavitt, W.C. Colbaugh and F.W. Thomas,
m. Physical Removal 1971. Principal Plume Dispersion Models:
TVA Power Plants. Journal of Air Pollution
WYNDvalley optionally simulates both wet Control Association, 21: 491–495.
and dry deposition. Dry deposition is propor- Chock, D.P., 1980. User’s Guide for the Sim-
tional to concentration in the lowest layer, ple Line-Source Model for Vehicle Exhaust
while wet deposition is proportional to rain Dispersion Near a Road. Environmental
rate and concentration in each layer. Appro- Science Department, General Motors Re-
priate coefficients (deposition velocities and search Laboratories, Warren, MI.
washout ratios) are input by the user. Colenbrander, G.W., 1980. A Mathematical
Model for the Transient Behavior of Dense
n. Evaluation Studies
Vapor Clouds, 3rd International Symposium
Harrison, H., G. Pade, C. Bowman and R. on Loss Prevention and Safety Promotion in
Wilson, 1990. Air Quality During Stagna- the Process Industries, Basel, Switzerland.
tions: A Comparison of RAM and DeMarrais, G.A., 1959. Wind Speed Profiles
WYNDvalley with PM–10 Measurements at at Brookhaven National Laboratory. Journal
Five Sites. Journal of the Air & Waste Man- of Applied Meteorology, 16: 181–189.
agement Association, 40: 47–52. Ermak, D.L., 1989. A Description of the
Maykut, N. et al., 1990. Evaluation of the SLAB Model, presented at JANNAF Safety
Atmospheric Deposition of Toxic Contami- and Environmental Protection Subcommit-
nants to Puget Sound. State of Washington, tee Meeting, San Antonio, TX, April, 1989.

434
Environmental Protection Agency Pt. 51, App. W
Gery, M.W., G.Z. Whitten and J.P. Killus, 030b. U.S. Environmental Protection Agency,
1988. Development and Testing of CBM-IV for Research Triangle Park, NC.
Urban and Regional Modeling. EPA Publica- Slade, D.H., 1968. Meteorology and Atomic
tion No. EPA–600/3–88–012. U.S. Environ- Energy, U.S. Atomic Energy Commission, 445
mental Protection Agency, Research Tri- pp. (NTIS No. TID–24190)
angle Park, NC. (NTIS No. PB 88–180039) Turner, D.B., 1964. A Diffusion Model of An
Gery, M.W., G.Z. Whitten, J.P. Killus and Urban Area. Journal of Applied Meteorology,
M.C. Dodge, 1989. A Photochemical Kinetics 3: 83–91.
Mechanism for Urban and Regional Scale Turner, D.B., 1969. Workbook of Atmos-
Computer Modeling. Journal of Geophysical pheric Dispersion Estimates. PHS Publica-
Research, 94: 12,925–12,956. tion No. 999–AP–26. U.S. Environmental Pro-
Gifford, F.A. and S.R. Hanna, 1970. Urban tection Agency, Research Triangle Park, NC.
Air Pollution Modeling. Proceedings of the Van Dop, H., 1992. Buoyant Plume Rise in
Second International Clean Air Congress, a Lagrangian Frame Work. Atmospheric En-
Academic Press, Washington, D.C.; pp. 140– vironment, 26A: 1335–1346.
1151.
APPENDIX C TO APPENDIX W OF PART 51—
Gifford, F.A., 1975. Atmospheric Dispersion
EXAMPLE AIR QUALITY ANALYSIS CHECKLIST
Models for Environmental Pollution Applica-
tions. Lectures on Air Pollution and Envi- C.0 Introduction
ronmental Impact Analyses. American Mete-
orological Society, Boston, MA. This checklist recommends a standardized
Green, A.E., Singhal R.P. and R. set of data and a standard basic level of anal-
Venkateswar, 1980. Analytical Extensions of ysis needed for PSD applications and SIP re-
the Gaussian Plume Model. Journal of the visions. The checklist implies a level of de-
Air Pollution Control Association, 30: 773– tail required to assess both PSD increments
776. and the NAAQS. Individual cases may re-
Heffter, J.L., 1965. The Variations of Hori- quire more or less information and the Re-
zontal Diffusion Parameters with Time for gional Meteorologist should be consulted at
Travel Periods of One Hour or Longer. Jour- an early stage in the development of a data
nal of Applied Meteorology, 4: 153–156. base for a modeling analysis.
Heffter, J.L., 1980. Air Resources Labora- At pre-application meetings between
tories Atmospheric Transport and Dispersion source owner and reviewing authority, this
Model (ARL-ATAD). NOAA Technical Memo- checklist should prove useful in developing a
randum ERL ARL–81. Air Resources Labora- consensus on the data base, modeling tech-
tories, Silver Spring, MD. niques and overall technical approach prior
to the actual analyses. Such agreement will
Irwin, J.S., 1979a. Estimating Plume Dis-
help avoid misunderstandings concerning the
persion—A Recommended Generalized
final results and may reduce the later need
Scheme. Fourth Symposium on Turbulence,
for additional analyses.
Diffusion and Air Pollution, Reno, Nevada.
Irwin, J.S., 1979b. A Theoretical Variation EXAMPLE AIR QUALITY ANALYSIS
of the Wind Profile Power-Law Exponent as CHECKLIST 1
a Function of Surface Roughness and Stabil-
ity. Atmospheric Environment, 13: 191–194. 1. Source location map(s) showing location
MacCready, P.B., Baboolal, L.B. and P.B. with respect to:
Lissaman, 1974. Diffusion and Turbulence • Urban areas 2
Aloft Over Complex Terrain. Preprint Vol- • PSD Class I areas
ume, AMS Symposium on Atmospheric Dif- • Nonattainment areas 2
fusion and Air Pollution, Santa Barbara, CA. • Topographic features (terrain, lakes,
American Meteorological Society, Boston, river valleys, etc.) 2
MA. • Other major existing sources 2
Moore, G.E., T.E. Stoeckenius and D.A. • Other major sources subject to PSD re-
Stewart, 1982. A Survey of Statistical Meas- quirements
ures of Model Performance and Accuracy for • NWS meteorological observations (sur-
Several Air Quality Models. EPA Publica- face and upper air)
tion No. EPA–450/4–83–001. U.S. Environ- • On-site/local meteorological observations
mental Protection Agency, Research Tri- (surface and upper air)
angle Park, NC.
Morgan, D.L., Jr., L.K. Morris and D.L. 1 The ‘‘Screening Procedures for Estimat-

Ermak, 1983. SLAB: A Time-Dependent Com- ing the Air Quality Impact of Stationary
puter Model for the Dispersion of Heavy Gas Sources, Revised’’, October 1992 (EPA–450/R–
Released in the Atmosphere, UCRL–53383, 92–019), should be used as a screening tool to
Lawrence Livermore National Laboratory, determine whether modeling analyses are re-
Livermore, CA. quired. Screening procedures should be re-
Pasquill, F., 1976. Atmospheric Dispersion fined by the user to be site/problem specific.
Parameters in Gaussian Plume Modeling, 2 Within 50km or distance to which source

Part II. EPA Publication No. EPA–600/4–76– has a significant impact, whichever is less.

435
Pt. 51, App. W 40 CFR Ch. I (7–1–97 Edition)
• State/local/on-site air quality monitoring • Anticipated growth changes
locations 2 4. Air quality monitoring data:
• Plant layout on a topographic map cov- • Summary of existing observations for
ering a 1km radius of the source with infor- latest five years (including any additional
mation sufficient to determine GEP stack quality assured measured data which can be
heights obtained from any state or local agency or
2. Information on urban/rural characteris- company) 4
tics: • Comparison with standards
• Land use within 3km of source classified • Discussion of background due to
according to Auer (1978): Correlation of land uninventoried sources and contributions
use and cover with meteorological anoma- from outside the inventoried area and de-
lies. Journal of Applied Meteorology, 17: 636– scription of the method used for determina-
643. tion of background (should be consistent
• Population with the Guideline)
¥> total 5. Meteorological data:
¥> density • Five consecutive years of the most re-
• Based on current guidance determination cent representative sequential hourly Na-
of whether the area should be addressed tional Weather Service (NWS) data, or one or
using urban or rural modeling methodology more years of hourly sequential on-site data
3. Emission inventory and operating/design • Discussion of meteorological conditions
parameters for major sources within region observed (as applied or modified for the site-
of significant impact of proposed site (same specific area, i.e., identify possible vari-
as required for applicant) ations due to difference between the mon-
• Actual and allowable annual emission itoring site and the specific site of the
rates (g/s) and operating rates 3 source)
• Maximum design load short-term emis- • Discussion of topographic/land use influ-
sion rate (g/s) 3 ences
• Associated emissions/stack characteris- 6. Air quality modeling analyses:
tics as a function of load for maximum, aver- • Model each individual year for which
age, and nominal operating conditions if data are available with a recommended
stack height is less than GEP or located in model or model demonstrated to be accept-
complex terrain. Screening analyses as able on a case-by-case basis
footnoted above or detailed analyses, if nec- —urban dispersion coefficients for urban
essary, must be employed to determine the areas
constraining load condition (e.g., 50%, 75%, —rural dispersion coefficients for rural
or 100% load) to be relied upon in the short- areas
term modeling analysis. • Evaluate downwash if stack height is less
—location (UTM’s) than GEP
—height of stack (m) and grade level above • Define worst case meteorology
MSL • Determine background and document
—stack exit diameter (m) method
—exit velocity (m/s) —long-term
—exit temperature (°K) —short-term
• Area source emissions (rates, size of area, • Provide topographic map(s) of receptor
height of area source)3 network with respect to location of all
• Location and dimensions of buildings sources
(plant layout drawing) • Follow current guidance on selection of
—to determine GEP stack height receptor sites for refined analyses
—to determine potential building • Include receptor terrain heights (if appli-
downwash considerations for stack heights cable) used in analyses
less than GEP • Compare model estimates with measure-
• Associated parameters ments considering the upper ends of the fre-
—boiler size (megawatts, pounds/hr. steam, quency distribution
fuel consumption, etc.) • Determine extent of significant impact;
—boiler parameters (% excess air, boiler provide maps
type, type of firing, etc.) • Define areas of maximum and highest,
—operating conditions (pollutant content second-highest impacts due to applicant
in fuel, hours of operation, capacity factor, source (refer to format suggested in Air
% load for winter, summer, etc.) Quality Summary Tables)
—pollutant control equipment parameters ¥> long-term
(design efficiency, operation record, e.g., can ¥> short-term
it be bypassed?, etc.) 7. Comparison with acceptable air quality
levels:
3 Particulate emissions should be specified • NAAQS
as a function of particulate diameter and
density ranges. 4 See footnote 2 of this appendix C.

436
Environmental Protection Agency Pt. 51, App. W
• PSD increments ¥> ‘‘Guideline for Determination of Good
• Emission offset impacts if nonattain- Engineering Practice Stack Height (Tech-
ment nical Support Document for the Stack
8. Documentation and guidelines for mod- Height Regulations)’’ (EPA–450/4–80–023R),
eling methodology: 1985
• Follow guidance documents ¥> ‘‘Ambient Monitoring Guidelines for
¥> appendix W to 40 CFR part 51 PSD’’ (EPA–450/4–87–007), 1987
¥> ‘‘Screening Procedures for Estimating
¥> Applicable sections of 40 CFR parts 51
the Air Quality Impact of Stationary
and 52.
Sources, Revised’’ (EPA–450/R–92–019), 1992
AIR QUALITY SUMMARY—FOR NEW SOURCE ALONE
Pollutant: lllllllll1 lllllllll2 lllllllll2

Highest Highest
Highest Highest Annual
2d high 2d high

Concentration Due to Modeled Source (µg/m3 ........


Background Concentration (µg/m3 ............................
Total Concentration (µg/m3 .......................................
Receptor Distance (km) (or UTM easting) .................
Receptor Direction (°) (or UTM northing) ..................
Receptor Elevation (m) ..............................................
Wind Speed (m/s) ......................................................
Wind Direction (°) .......................................................
Mixing Depth (m) ........................................................
Temperature (°K) .......................................................
Stability .......................................................................
Day/Month/Year of Occurrence .................................

Surface Air Data From llllllllll Surface Station Elevation (m) llllllllll

Anemometer Height Above Local Ground Level (m) llllllllll


Upper Air Data From llllllllllllllllllllllll
Period of Record Analyzed lllllllllllllllllllll
Model Used llllllllllllllllllllllllllll
Recommended Model lllllllllllllllllllllll
1 Use separate sheet for each pollutant (SO2, PM–10, CO, NOX, HC, Pb, Hg, Asbestos, etc.).
2 List all appropriate averaging periods (1-hr, 3-hr, 8-hr, 24-hr, 30-day, 90-day, etc.) for which an air quality standard exists.

AIR QUALITY SUMMARY—FOR ALL NEW SOURCES


Pollutant: lllllllll1 lllllllll2 lllllllll2

Highest 2nd Highest 2nd


Highest Highest Annual
high high

Concentration Due to Modeled Source (µg/m3 ........


Background Concentration (µg/m3 ............................
Total Concentration (µg/m .......................................
3

Receptor Distance (km) (or UTM easting) .................


Receptor Direction (°) (or UTM northing) ..................
Receptor Elevation (m) ..............................................
Wind Speed (m/s) ......................................................
Wind Direction (°) .......................................................
Mixing Depth (m) ........................................................
Temperature (°K) .......................................................
Stability .......................................................................
Day/Month/Year of Occurrence .................................

Surface Air Data From llllllllll Surface Station Elevation (m) llllllllll

Anemometer Height Above Local Ground Level (m) llllllllll


Upper Air Data From llllllllllllllllllllllll
Period of Record Analyzed lllllllllllllllllllll
Model Used llllllllllllllllllllllllllll
Recommended Model lllllllllllllllllllllll
1 Use separate sheet for each pollutant (SO2, PM–10, CO, NOX, HC, Pb, Hg, Asbestos, etc.).
2 List all appropriate averaging periods (l-hr, 3-hr, 8-hr, 24-hr, 30-day, 90-day, etc.) for which an air quality standard exists.

437
Pt. 51, App. X 40 CFR Ch. I (7–1–97 Edition)

AIR QUALITY SUMMARY—FOR ALL SOURCES


Pollutant: lllllllll1 lllllllll2 lllllllll2

Highest 2nd Highest 2nd


Highest Highest Annual
high high

Concentration Due to Modeled Source (µg/m3 ........


Background Concentration (µg/m3 ............................
Total Concentration (µg/m3 .......................................
Receptor Distance (km) (or UTM easting) .................
Receptor Direction (°) (or UTM northing) ..................
Receptor Elevation (m) ..............................................
Wind Speed (m/s) ......................................................
Wind Direction (°) .......................................................
Mixing Depth (m) ........................................................
Temperature (°K) .......................................................
Stability .......................................................................
Day/Month/Year of Occurrence .................................

Surface Air Data From llllllllll Surface Station Elevation (m) llllllllll

Anemometer Height Above Local Ground Level (m) llllllllll


Upper Air Data From llllllllllllllllllllllll
Period of Record Analyzed lllllllllllllllllllll
Model Used llllllllllllllllllllllllllll
Recommended Model lllllllllllllllllllllll
1 Use separate sheet for each pollutant (SO2, PM–10, CO, NOX, HC, Pb, Hg, Asbestos, etc.)
2 List all appropriate averaging periods (1-hr, 3-hr, 8-hr, 24-hr, 30-day, 90-day, etc.) for which an air quality standard exists.

STACK PARAMETERS FOR ANNUAL MODELING


Emis- Building dimensions (m)
Stack
sion Stack
Stack Stack exit
rate for Phys- GEP base
Stack exit di- exit ve- tem- Stack
Serving each ical stack ele-
No. ameter locity pera- (m)
pollut- height ht. (m) vation Height Width Length
(m) (m/s) ture
ant (m)
(°K)
(g/s)

STACK PARAMETERS FOR SHORT-TERM MODELING 1


Emis- Building dimensions (m)
Stack
sion Stack
Stack Stack exit
rate for Phys- GEP base
Stack exit di- exit ve- tem- Stack
Serving each ical stack ele-
No. ameter locity pera- (m)
pollut- height ht. (m) vation Height Width Length
(m) (m/s) ture
ant (m)
(°K)
(g/s)

1 Separate tables for 50%, 75%, 100% of full operating condition (and any other operating conditions as determined by screen-
ing or detailed modeling analyses to represent constraining operating conditions) should be provided.

[61 FR 41840, Aug. 12, 1996]

438