Summary of Comments and Responses on the December 1984
Proposed Revisions to the Guideline on Air Quality Models
July 1986
Source Receptor Analysis Branch
Monitoring and Data Analysis Division
Office of Air Quality Planning and Standards
Environmental Protection Agency
-------
Table of Contents
Summary of Comments and Responses on the December 1984
Proposed Revisions to the Guideline on Air Quality Models
Page
OVERVIEW 1
1.0 INTRODUCTION 1-1
2.0 OVERVIEW OF MODEL USE 2-1
3.0 RECOMMENDED AIR QUALITY MODELS 3-1
3.1 Preferred Modeling Techniques 3-2
3.2 Use of Alternative Models 3-8
3.3 Availability of Supplementary Modeling Guidance 3-13
4.0 SIMPLE-TERRAIN STATIONARY-SOURCE MODELS 4-1
5.0 MODEL USE IN COMPLEX TERRAIN 5-1
6.0 MODELS FOR OZONE, CARBON MONOXIDE AND NITROGEN DIOXIDE 6-1
6.1 Discussion 6-1
6.2 Recommendations 6-2
6.2.1 Models for Ozone 6-2
6.2.2 Models for Carbon Monoxide 6-5
6.2.3 Models for Nitrogen Dioxide (Annual Average) 6-8
7.0 OTHER MODEL REQUIREMENTS 7-1
7.1 Discussion 7-1
7.2 Recommendations 7-1
7.2.1 Fugitive Dust/Fugitive Emissions 7-1
7.2.2 Particulate Matter 7-3
7.2.3 Lead 7-4
7.2.4 Visibility 7-5
7.2.5 Good Engineering Practice Stack Height 7-6
7.2.6 Long Range Transport 7-9
7.2.7 Modeling Guidance for Other Government Programs .. 7-10
8.0 GENERAL MODELING CONSIDERATIONS 8-1
8.1 Discussion 8-1
8.2 Recommendations 8-1
8.2.1 Design Concentrations 8-1
8.2.2 Critical Receptor Sites 8-2
8.2.3 Dispersion Coefficients 8-5
8.2.4 Stability Categories 8-8
-------
8.2.5 Plume Rise 8-9
8.2.6 Chemical Transformation 8-15
8.2.7 Gravitational Settling and Deposition 8-16
8.2.8 Urban/Rural Classification 8-17
8.2.9 Fumigation 8-18
8.2.11 Calibration of Models 8-19
9.0 MODEL INPUT DATA 9-1
9.1 Source Data 9-1
9.2 Background Concentrations 9-4
.9.3 Meteorological Input Data 9-8
9.3.1 Length of Record of Meteorological Data 9-8
9.3.2 National Weather Service Data 9-11
9.3.3 Site Specific Data 9-12
9.3.4 Treatment of Calms 9-19
10.0 ACCURACY AND UNCERTAINTY OF MODELS 10-1
11.0 REGULATORY APPLICATION OF MODELS 11-1
11.1 Discussion 11-1
11.2 Recommendations 11-1
11.2.1 Analysis Requirements 11-1
11.2.2 Use of Measured Data in Lieu of Model
Estimates 11-3
11.2.3 Emission Limits 11-5
APPENDIX A. SUMMARIES OF PREFERRED AIR QUALITY MODELS
A.O INTRODUCTION A-l
A.2 CALINE3 A-2
A.4 RAM A-2
A.5 ISC A-2
A.6 MPTER A-3
APPENDIX B. SUMMARIES OF ALTERNATIVE AIR QUALITY MODELS
B.3 APRAC-3 B-l
B.9 IMPACT (Sklarew) B-l
B.I 2 MESOPLUME B-l
B.18 SCSTER B-l
B.24 RTDM (Version 3.00) B-2
REFERENCES R-l
GLOSSARY OF COMMENTERS APPEARING IN DOCKET A-80-46 G-l
-------
Summary of Comments and Responses on the December 1984
Proposed Revisions to the Guideline on Air-Quality Models
OVERVIEW
Background
The Guideline on Air Quality Models was originally published in April
1978. It was incorporated by reference in the regulations for the Prevention
of Significant Deterioration of Air Quality in June 1978. The purpose of
the guideline is to promote consistency in the use of modeling within the
air management process. Consistency is a primary goal of the 1977 Clean
Air Act Amendments. The guideline provides model users with a common basis
for estimating pollutant concentrations, assessing control strategies and
specifying emission limits.
In October 1980, EPA proposed changes to the Guideline on Air Quality
Models and solicited comments on the changes. More than 80 comments were
submitted. EPA responded to the comments resulting from the draft guideline
and a summary of both the comments and the EPA responses is contained in a
document entitled "Summary of Comments and Responses on the October 1980
Proposed Revisions to the Guideline on Air Quality Models," February 1984.
A copy of this document as well as all supporting reports is available for
review in Docket A-80-46.
As a result of public comment, the guideline was further revised. On
December 7, 1984 [49 £R 48018], EPA announced the availability of the
"Guideline on Air Quality Models (Revised)" for further public comment.
Oral comments were presented by interested parties during the Third Conference
on Air Quality Modeling held in Washington, DC in January 1985. Proceedings
-------
of these hearings were transcribed verbatim. Written comments were received
until April 30, 1985. The purpose of this document is to summarize those
comments and to present EPA responses to the major issues.
Summary of Comments
In 1977, the "Guideline on Air Quality Models" was the subject of a
public hearing process in connection with promulgation of regulations for
the Prevention of Significant Deterioration. In 1984, EPA proposed to
change the reference in 40 CFR 51.24 and 40 CFR 52.21 from the 1978 edition
of that guidance to the "Guideline on Air Quality Models (Revised)" to be
completed and dated when the proposed changes become final and are promulgated
Several issues were identified in the Federal Register announcement:
0 Specific changes to 40 CFR 51 and 52
0 Revised format of the guideline
0 Recommendations for ozone models
0 Proposed changes to preferred models
0 Improving performance evaluations, especially for ozone models
0 Modeling uncertainty
0 Degree to which State or local regulatory agencies can have authority
to use nonguideline models
0 Degree of oversight or approval authority retained by EPA.
Sixty-six commenters responded. The Glossary for this document lists
the name of these commenters and indicates the docket reference number of
the entire text of the individual submittals. Over 450 separate comments
were extracted from these submittals and were later condensed and organized
by topic. Many commenters suggested changes to specific sections of the
guideline; at least one comment was received on every section of the proposed
-------
revisions. Most comments were very technical in nature and quite detailed.
The comments were varied; some were very specific while others were very
general. In nearly all comments, the issue of consistency versus flexibility
was inherent.
The commenters on the proposed revisions can be separated into three
major categories: private sector and industrial associations, State and
local air management agencies, and Federal-Government representatives. By
far the largest number of comments came from private industry. Although
some of these industrial commenters voiced general support for the guide-
line, several were critical of specific aspects of EPA guidance. However,
recommendations to EPA were contradictory; some suggested that modeling
analyses should be reviewed on a case-by-case basis while others said that
guidance was not detailed enough and that further guidance was necessary.
The Federal Agency comments supported the need for consistency and
provided suggestions for changes to the proposed revisions or suggested
improvements to EPA procedures. Comments were received from four EPA
regions and have been resolved internally. Comments from State and local
agencies were numerous. They appeared to support the concept of the guideline
although not necessarily the complete content.
EPA Responses to Comments
As the comments were received, it was evident that not only had comments
been submitted on the eight highlighted issues, but the entire proposal had
been considered by most of the commenters. It was impractical to organize
the responses by those eight issues. Therefore, the summary was organized
by topic and the topics listed in the order of the chapter and section that
appears in the guideline. Summarizing and characterizing the comments
-------
themselves required some interpretation in order to place them in manageable
response categories. Whenever possible the exact words used by the commenter
were used in the comment summary. However, in many instances it was
necessary to re-word or substantially condense the comment. Every effort
was made to maintain the exact meaning of the ccmmenter. Following the
summary of comments on each separate topic, the EPA response to those
comments is given. Where there are subtopics in the comment summary, the
response may be separated into subparagraphs. At the end of each comment
summary, the primary commenters1 name(s), in abbreviated form, is used and a
key to these abbreviations is shown in the Glossary. Not every commenter who
may have alluded to that issue is necessarily listed with the comment summary,
However, all issues have been addressed.
-------
1.0 INTRODUCTION
Comment Summary (Model Accuracy and Consistency)
A number of commenters urged the Agency to place model accuracy ahead
of model consistency. They argued that use of the most accurate models
should be promoted and that the need for consistency was overstated. Con-
sistency is not always possible, thus flexibility/adaptability should not
be sacrificed. The various commenters also noted that (1) the regulatory
program should not require use of a single model, (2) use of a single model
was based on an arbitrary selection process, and (3) this selection made
the Agency very inflexible in allowing use of nonguideline models, especially
those involving advances in technology. Furthermore, one comenter urged
that Regional Offices use a consistent framework for modeling decisions.
(AISI, AMC, APCA, API, DS, IPL, MSUS, SOC)
EPA Response
EPA's position is not that the "same answer" is preferable to the
"best answer". The model that most accurately estimates concentrations in
the area of interest is always sought. However, it is clear from the needs
expressed by the States and EPA Regional Offices, by many industries and trade
associations, and also by the deliberations of Congress, that uniform proce-
dures in the selection and application of models and data bases should also be
sought. Consistency ensures that air quality control agencies, affected
industries, and the general public have a common basis for estimating pollutant
concentrations, assessing control strategies and specifying emission limits.
Such consistency is not, however, promoted at the expense of model and data
base accuracy.
The modeling guideline provides a uniform basis for selection of the most
accurate models and data bases for use in air quality assessments. The ulti-
mate goal of modeling guidance is to ensure that the best possible scientific
procedures are implemented for operational use. Suitable mechanisms have been
provided to ensure that the realism, flexibility, accuracy and best technical
judgements, can be provided in a framework that satisfies the Clean Air Act
1-1
-------
requirements. This has been attempted through the general nature of guideline
requirements,2 guidance on model demonstrations,3,4 and the operation of a
Model Clearinghouse.5 Those activities are also intended to ensure consistency
in the overall procedures that are followed by EPA Regional Offices.
EPA has also solicited models from all model developers and since the
inception of the program has considered over 25 non-EPA models.6 Where appro-
priate, those models are recognized in the revised guideline. The limitations
under which those models are used should be part of the initial protocol estab-
lished between the applicant or model user and the appropriate EPA Regional
Office.
Based on its assessment of these models, EPA has designated certain
models as "preferred". This is consistent with requirements of Sections
301 (a) (2) (A) of the Clean Air Act requiring EPA to promulgate regulations
to "assure fairness and uniformity in the criteria, procedures, and policies
applied by the various regions in implementing and enforcing the Act," Section
165(e)(3){D) concerning PSD, and more general requirements to demonstrate the
adequacy of State Implementation Plans. The criteria for designating those
models are discussed under comments in Section 3.1 dealing with "Basis for
Model Selection." The designated standard models are frequently the EPA
developed models and the ones recommended for specific uses in the guideline.
However, as stated in the guideline, the model applied to a given situation
should be the one that is most accurate in simulating atmospheric transport
and dispersion in the area of interest. The PSD regulations specifically make
allowance for the use of nonguideline techniques and the framework for
consistent Regional Office decisions is already functioning.
1-2
-------
Comment Summary (Wording Changes)
Several commenters recommended specific wording changes in Chapter 1.
Those changes are listed below.
Page 1-1, Lines 11 to 12--The sentence should be revised to retain the
wording of the 1978 guideline: "Rather, it should serve as a basis by which
air quality managers, supported by sound scientific judgment, have a common
measure of acceptable technical analysis,"
Page 1-1, Line 13--The statement, "Due to limitations in the spatial
coverage . . ." should be changed to read, "Due to limitations in the spatial
and temporal coverage . . . ."
Page 1-1, Line 21--Change the word "suitable" to "preferable."
Page l-l--The second paragraph should be changed to state that . . .
"monitoring and modeling data should be used in a complementary manner, with
due regard for the strengths and weaknesses of each."
Page 1-2, Lines 16 to 23--The following text should be added at the end
of line 23. "The model that most accurately estimates concentrations in the
area of interest is always sought. However, designation of specific models
is needed to promote consistency in model selection and application. Such
consistency is not, however, promoted at the expense of model and data base
accuracy. This guide provides a consistent basis for the selection of the
most accurate models and data bases for use in air quality assessments."
Page 1-3, Lines 8 to 10--The sentence starting on line 8 should be
changed to: "In all cases, the model applied to a given situation should
be the one that provides the most accurate representation of atmospheric
transport, dispersion, and chemical transformations in the area of interest."
Pages 1-3, Lines 12 to 20--The Regional Meteorologists' workshops should
be held not only to ensure model consistency but also to "promote the use of
more accurate air quality models and data bases." (APCA, CONE, SOC, UARG)
EPA Response
EPA agrees with all of these suggested word changes and has included
them in the final text of the guideline.
1-3
-------
2.0 OVERVIEW OF MODEL USE
Comment Summary (General)
One commenter recommended that model estimates only be made by those
with the requisite technical competence. Several other commenters recom-
mended specific wording changes in Chapter 2. Those changes are listed
be!ow.
Page 2-2, Line 15--Change the sentence to read: "Air quality models are
applied with the least degree of uncertainty in areas with relatively simple
topography.
Page 2-2, Lines 16 to 17--This sentence should be reworded to read: "Air
quality models have been most accurately applied to simulations of long term
averages in areas with relatively simple topography."
Page 2-3, Line 12--An explanatory sentence should be added as follows:
"Further, it should be recognized that under some sets of physical circum-
stances and accuracy requirements, no present model may be appropriate."
Page 2-5, Lines 2 to 4--The sentence concerning physical modeling should
be revised to read: "Nevertheless, physical modeling may be useful for com-
plex flow situations, such as building, terrain or stack downwash conditions,
plume impact on elevated terrain, diffusion in an urban environment, or
diffusion in complex terrain."
Page 2-6, Line 6--EPA should state that "If screening techniques show
impacts that do not approach or exceed PSD or NAAQS standards, then no further
refined modeling will be necessary." (APCA, CDH, NYEC, SOC, SRP)
EPA Response
EPA agrees that competent and experienced personnel are a requirement
for modeling. EPA staff are highly trained in this area. A routine audit
program to test the adequacy and improve State modeling programs has been
implemented.
EPA agrees with all of the suggested word changes and has included them
in the final text of the guideline.
2-1
-------
3.0 RECOMMENDED AIR QUALITY MODELS
Comment Summary (General)
Several commenters recommended specific wording changes in Chapter 3.
Those suggested changes are listed below.
1. Page 3-4, lines 5 to 7--The text should be revised to read "Models
found to be clearly superior based on an evaluation using che same data bases
as used to evaluate the preferred models will be proposed for inclusion as
preferred models in future guideline revisions."
2. Page 3-4, add--"The modeler/meteorologist exercising the preferred
model should state the degree of accuracy and precision that is expected in
the given application, using standard statistical terminology. Such state-
ment will provide the regulatory decision-maker with a more meaningful basis
for his decisions."
3. Page 3-7—Use the term "at least as accurate as" wherever the term
"equivalent" appears. The term then becomes self explanatory. Suggest delet-
ing the third paragraph entirely and substituting "more accurate performance"
for the term "superior performance" in last paragraph.
4. Page 3-8--Change 3a to read: "Performance evaluations of the model
in similar circumstances have shown that the model is no less accurate or
precise, or." Change "superior" to "more accurate and precise" whenever it
appears in 3b.
5. Chapter 3--The Interim Procedures should not be cited if they have
not been peer reviewed. (APCA, IPL, SRP, SOC)
EPA Response
1. This change will be adopted.
2. The thrust of these words is already included at the end of Chapter 10.
No change will be made.
3 and 4. The suggestions to substitute "at least as accurate as" for
"equivalent" and "no less accurate" for "not biased toward underestimates"
indicate that the commenters have misunderstood the intent of these statements
(as discussed under comments in Section 3.2 dealing with "Equivalency and
Alternative Models"). No action will be taken. However, in Chapter 3 the
words "performs better than" will be used in lieu of either "more accurate" or
3-1
-------
"superior" to maintain consistency with terminology in the Interim Procedures
for Evaluating Air Quality Models (Revised.)4
5. The Interim Procedures document has been subjected to EPA's review
requirements and published as an EPA report. It will continue to be cited.
3.1 Preferred Modeling Tecnniques
Comment Summary (Basis for Model Selection)
Several commenters requested supplemental information on the basis for
selecting models (identified in Appendix A of the guideline) as preferred
for specific regulatory applications. The basis for selecting these models
over others is not evident, especially since there is no clearly superior
model for some applications. Also, the evaluations are not sufficiently
extensive, nor are any of the models accurate enough to justify using some
models as the basis for judging others. Further evaluations of rural and
urban models were requested due to limitations of the previous studies;
in particular it was suggested that selected factors be further evaluated,
especially the Me Elroy-Pooler dispersion coefficients which were based on
data from the same location as the urban evaluation. Additional documentation
concerning the previous studies was also requested. One commenter stated
that numerous procedural changes in the guideline which significantly affect
predicted concentrations have been neither supported on a technical basis nor
peer reviewed concerning model accuracy. (APCA, CMA, DS, IPL, ODEQ, SUC)
EPA Response
EPA agrees that there is no clearly superior model for the various
categories in which preferred models are identified. The models listed in
Appendix A were selected because (1) they are at least as accurate as other
available models; (2) in at least one case, a unique approach to specific
analysis problems is provided (e.g., ISC for industrial complexes); (3) they
have been widely used for regulatory applications in the past (similar models
identified in Appendix B have been used for a much narrower set of applications);
(4) they form the current basis for control regulations for many sources in
many parts of the country and their selection results in a minimum disruption
of those regulatory programs; (5) they have been widely released through
3-2
-------
UNAMAP? or through other means by government agencies and are readily available
at nominal cost; and (6) their wide use and current basis for regulatory programs
have resulted in high public familiarity with these models. The reader should
note, as indicated previously (under comments in Section 1.0 dealing with "Model
Accuracy and Consistency") that where consistent techniques can be used within
the context of obtaining the most accurate estimates, they are encouraged by
the CAA and by various governmental and industry parties.
The thrust of EPA's evaluations has not been to judge other models, but
to establish the relative accuracy of models, one to the other. As a result,
the models recommended by EPA have been found to be at least as accurate as,
if not better than, other available models. They thus form a sound basis
for a regulatory air quality program. Where the equivalency of models is at
issue (as discussed under comments in Section 3.2 dealing with "Equivalency"),
a test is implemented to determine if the models provide the same concentration
estimates. If the estimates are not the same, then a mechanism to determine
the best model has been provided.4
While more extensive evaluations of all models are desirable, the best
data bases available at the time have been used to evaluate the EPA preferred
models. In all cases the data bases were selected in coordination with the
Steering Committee of the AMS/EPA Cooperative Agreement. Documentation of
the evaluations and of the peer scientific reviews were extensive and have
been widely distributed; no further documentation is appropriate or plannedS-13.
These detailed analyses have tended to affirm the relative accuracy of EPA
modelsJ4,15 Furthermore, other data bases^6^7 have tended to support
the findings of EPA's evaluations. Those model evaluations prior to 1982
for all models in Appendices A and B that were documented by the developers
have been summarized in an EPA report.18 Nevertheless, EPA plans further
3-3
-------
evaluations, especially for rural, urban, and complex terrain models as
time and resources allow. However, the further evaluation of urban models
suggested by one commenter must wait for another data base such as the
EPRI-sponsored Indianapolis field experiment.^ All models were
operated for the urban evaluation as specified by the developers and any
further adjustment and evaluation for the same data-base would constitute
"tuning" a model for that data base. Also, it should be noted that the
McElroy-Pooler coefficients were developed for the same city, but not for
the data base used in the evaluation; thus, the "tuning" implied by the
commenter does not exist.
Contrary to one commenter1s statement, the procedural model changes that
are proposed (e.g. wind speed profile exponents) have a technical basis and
have been reviewed for accuracy. In most cases those changes were proposed
or reviewed by EPA's Office of Research and Development and were subjected
and supported by comments at public hearings held in 1980. The technical
basis, regulatory impact, sensitivity and accuracy of the proposed changes
were thoroughly documented and made available in Docket A-80-46 as part of
information publically released prior to the 1985 public hearing (appendices
to Summary of Comments and Responses) J In most cases, the sensitivity analyses
and accuracy assessments showed little change from current practice. Where a
more extensive change does occur, public comments are being assessed and are
discussed elsewhere in this document.
Comment Summary (Texas Models)
Several commenters argued that the Texas Models (TEM-8A and TCM-2)
should be retained in the guideline as preferred models. These models are
viewed as meeting EPA's criteria for selection, are economical to run and
have not been shown to be inferior to other models. Concern was also expressed
that failure to include these as preferred models would have an adverse effect
on the consistency of PSD permitting analyses where the Texas Models are
currently used. Other commenters variously suggested (1) that an option to
3-4
-------
use these models should be granted; (2) that they are adequate for pre-permit
review in "flat terrain" States, and (3) that these models were not treated
properly in the urban model evaluation. (APCA, ARCO, CC, DS, EPNG, TACB)
EPA Response
EPA carefully considered the Texas Models as discussed on page 110 of the
Summary of Comments and ResponsesJ The models were evaluated in a manner
consistent with similar models and were subjected to peer scientific reviews.
No substantive difference was found between TCM and other similar models for
annual average urban applications. However, even though reviewers found it
difficult to distinguish among the models evaluated, some differences were
apparent for TEM. For short-term urban applications TEM had a notably greater
bias towards overestimating observed concentrations than did RAM. For rural
applications, TEM tended to underestimate 24-hour S02 concentrations for which
the NAAQS apply; CRSTER and MPTER were relatively unbiased. While the Texas
models have been widely used in the State of Texas, they have not been used
extensively to set emission limits for sources in other parts of the country.
Thus, even though these models may satisfy criteria such as public familiarity,
cost, and availability, they would change the basis for regulation in most
parts of the country; such a change could not be justified given their statis-
tical performance relative to those models currently recommended in Appendix A.
The fact that the Texas Models are included in Appendix B, rather than
with the preferred models in Appendix A, should have no impact on consistency
of PSD permitting analyses in the State of Texas. EPA has met with Texas
State representatives and indicated that there is no intention to preclude
the use of these models in that regulatory program. These models will be
3-5
-------
tested by the State of Texas using an agreed upon protocol, and their use
will be allowed in -that state if these models can satisfy the demonstration
requirements. In the interim, these models may continue to be used there
because of long use and historical precedent. The issue concerning the
urban model evaluation is discussed in this section under comments dealing
with "Basis for Model Selection."
Comment Summary (Implementation of New Models)
Several commenters indicated that criteria EPA will use to replace a
preferred model with a new model are not clear. EPA was urged to make pro-
visions in the guideline for use of new models, for improvements to existing
models, and for models that are otherwise more appropriate in specific cases.
The proposed list of preferred models should not remain static. Also the
criteria to include other models should not be so heavily weighted in favor
of the current models that new models are faced with an undue burden to prove
superiority, or that state-of-the-art advances are inhibited. One commenter
suggested criteria that should be considered when deciding the superiority of
a new model, including: (1) performance measures that are more concise and
emphasize the ratio of predicted to observed values and the mean square error;
(2) comparisons at the upper end of the concentration frequency distribution
that are unpaired in space or time; and (3) a limit of one or two data sets
for sites typical of those to which the model is to be applied. This last
commenter also indicated that, for the above three criteria, if another model
performs as well or better than a preferred model listed in Appendix A, then
that model should also be given guideline status. (API, EPNG, MSUS, SOC,
UARG).
EPA Response
The guideline generally states that, in addition to the six items
listed in the Federal Register (45 FR 20157),6 new models will also be
subjected (1) to a performance evaluation which includes the data base(s)
used in the original EPA evaluation and (2) to a peer scientific review.
Models found to perform better for general applications will be proposed
3-6
-------
for recognition as a preferred model. The three criteria suggested by one
commenter are a viable starting point and are generally consistent with
criteria that EPA has considered in past analyses. However, they cannot be
used as the sole criteria; the soundness of the scientific principles in
models must be left to the judgment of peer reviewers. Since the scientific
community has not yet identified performance standards for models, the
Steering Committee for the AMS/EPA Cooperative Agreement has been asked to
identify factors that should be considered in establishing improved model
performance. It is not possible to be more specific about the criteria
that will be used to replace a preferred model. In fact, it may take
several experiences with new models before a specific formal procedure can
evolve.
One example of why the development of this process must proceed with
care is the PPSP model (listed in Appendix B of the guideline). This model
implements many of the improvements suggested by the peer review of rural
models.H However, when tested against the Clifty Creek data base,8 the model
systematically overestimated the highest concentrations. This clearly
illustrates the fallacy of any assumption that a new model with "a more
credible scientific basis" is necessarily a more accurate model.
EPA has made provisions in the guideline for adding new and improved
models. EPA encourages the use of these provisions and does not intend to
place an undue burden on new models or delay state-of-the-art advances that
are appropriate for implementation. Provisions include (1) use of models
for case-specific applications following procedures outlined in Interim
Procedures for Evaluating Air Quality Models (Revised),4 and (2) addition
of new models based on the general criteria identified above. The criteria
are int.ended to provide an even-handed technical assessment of models so as
3-7
-------
to select that which is most appropriate for the applications considered.
As stated in the guideline, it is not intended that the currently preferred
models are to be permanently used to the exclusion of others. The most
accurate estimates are always sought. Thus, where a new or improved model
is found to be better than a preferred model through the evaluation and
review process, it will be used to replace the preferred model or to fill a
niche not covered by the preferred models. However, including models just
because they are found to be "as good as" a preferred model, adds neither
to the accuracy of the estimates nor to the uniformity fostered by the CM
and can only lead to regulatory confusion.
3.2 Use of Alternative Models
Comment Summary (Equivalency)
Several commenters addressed the issue of "equivalency" between air
quality models. The comments fell into three general categories. First,
several commenters said that equivalency of 2% (difference between proposed
and preferred models) was too stringent and that this criteria is not even
satisfied by models in Appendix A. Alternative measures of equivalency were
suggested including those between 5% and 50%, those with statistical tests, and
those based on a wider set of temporal and spatial concentration comparisons.
Second, other commenters implied that the concept of equivalency was invalid
for determining the acceptability of a model and tended to preclude the use
of improved or alternative models. Third, three commenters indicated that
because of the inaccuracy of models and data bases, agreement within 2% was
within the "noise level" of estimates provided by models and was therefore
meaningless. (APCA, API, CC, CITG, DS, IEPA, MSUS, NYEC, OEPA, SOC)
EPA Response
As stated on page 60 of the Summary of Comments and Responses, EPA has
never required numerical agreement as a prerequisite to using a nonguideline
model. In dealing with an alternative nonguideline model, EPA is proposing
that a showing be made that the alternate performs better than the recommended
3-8
-------
model. This is a procedure not unlike that suggested by the American
Meteorological Society in the Woods Hole Workshop Report.20 However, in
response to requests from developers, EPA proposed criteria to identify
equivalent models, or models so nearly identical to those that are preferred
that they can be treated for practical purposes, as recommended models. It
is from this latter proposal which is not a requirement, but is meant to be
a reasonable response to requests from model developers, that much confusion
about "equivalent" models and "numerical agreement" has resulted. Neverthe-
less, three model developers successfully showed that when specific options
in their models are used, essentially identical estimates to those from recom-
mended models in UNAMAP (Version 5) can be achieved. Equivalence to recommended
models in UNAMAP (Version 6) will now be necessary, however.
To show that models are equivalent, or their estimates are nearly identical ,
fairly stringent criteria are necessary. Since the individual algorithms in
most air quality models are well known and reproducible with high precision,
differences greater than 10% are certainly significant. The preferred models
in Appendix A, where a comparison is appropriate, are equivalent to each other;
minor discrepancies have been reconciled for this promulgation. Also, the
comments about use of statistical tests and the use of a wider set of concen-
tration comparisons have merit. The way in which these criteria would be used,
though, was not made clear by the commenters. These additional criteria would
also make it more difficult to show equivalency, and would increase the likeli-
hood (as suggested by other commenters) that a minor artifact in the way the
models operate would preclude a demonstration of equivalency to each other.
Therefore, EPA will continue to use the maximum and the highest, second-highest
concentrations as a sufficient demonstration of equivalency. Similarly, since
no commenter directly refuted the basis for the 2% equivalency criteria, nor
3-9
-------
provided any data for an alternative, the 2% criteria will be maintained.
If in a special case the 2% criteria is found to be unreasonable, it will
be reassessed on a case-specific basis.
The commenters who are concerned that 2% is within the "noise level" of
model estimates are correct. However, thay have failed to recognize (1) the
regulatory need for a single, consistent modeling technique (as discussed
under comments in Section 1.0 dealing with "Model Consistency vs Accuracy"),
and (2) that equivalency as used here does not preclude the use of better
models. Thus, the need to be able to distinguish between models, or vice
versa, to identify models that provide essentially identical estimates,
remains and is discussed below.
Comment Summary (Alternative Models)
A number of commenters addressed the circumstances and criteria under
which alternatives to the preferred models may be used. In several cases
these comments overlap the issues of consistency and equivalency which are
addressed elsewhere. In general the commenters felt that Section 3.2.2
concerning alternative models is too restrictive, burdensome, and incomplete.
As long as an alternative model uses the same basic theories as a preferred
model, a statistical performance evaluation or an equivalency test are
unnecessary given the inherent inaccuracies of the preferred models; not
all the preferred models have been subjected to such an elaborate evaluation.
This function should be decentralized to the Regional Offices.
One commenter suggested that the burden of the equivalency test could be
lessened if the equivalent models and the appropriate options were listed by
EPA; also EPA was urged to allow simple changes to models without a full
equivalency demonstration provided the change did not affect the concentration
algorithms. Other commenters recommended that (1) an applicant be allowed to
demonstrate the superiority of alternative models following the procedures on
page 3-8 of the guideline (2) guidance be given on how physical modeling can
be used to evaluate mathematical models, and (3) where a preferred model does
not exist, no requirement for conservative estimates should be imposed. (ADEM,
AMC, APCA, EPNG, MSUS, SOC, UARG, WDNR)
EPA Response
The purpose of Section 3.2.2 is to provide an objective and technically
sound means for determining the acceptability of an alternative model for a
3-10
-------
regulatory application. Selection of the best technique is always encouraged.
However, demonstrating that an alternative model is both appropriate and
performs better than other models is a substantial undertaking requiring
major investment of time and resources. That is why options are provided
for demonstrating (1) equivalency to a-preferred model (as discussed under
comments in this section dealing with "Equivalency"), (2) performance better
than a preferred model, or (3) a reasonable level of performance where no
preferred model exists. The basis for these demonstrations is the Interim
Procedures for Evaluating Air Models (Revised) which basically follows
recommendations of the American Meteorological Society.20 Given the need for
consistency in the models used and the original basis for selection of the
preferred models, which is discussed elsewhere, no further options seem
available. Thus, the requirements of Section 3.2.2 appear to be technically
sound; within the limit of other requirements, they are not restrictive,
burdensome, or incomplete.
The Interim Procedures document^ provides a complete basis for documenting
the superiority of a given model and its use is encouraged. Development of
protocols to identify procedures and statistical tests is a major component
of this documentation. The use of protocols is encouraged; experience gained
from past use of such protocols has been summarized .21 The thrust of the
demonstration procedure is to encourage first-hand communications between the
source and the State or EPA Regional Office; the role of the Model Clearing-
house is one of general guidance and review. The Interim Procedures document
is appropriate whenever there are differences between models that affect the
concentration estimates. Just because models use the same general theories
similar concentration estimates are not insured (as discussed in Section 1.0
under comments dealing with "Model Consistency vs Accuracy"), and the need
3-11
-------
for a statistical evaluation to determine the more accurate model is not
eliminated. Where on-site data bases are not available, off-site data may be
considered on a case-specific basis, under limited circumstances documented
in the Interim Procedures. Finally, it should be noted that all models included
as preferred models have been subjected to evaluations comparable in detail
to those suggested in the Interim Procedures, if not by the "letter" of that
document.
Non-EPA models that met the equivalency test based on comparison of results
from recommended models in UNAMAP (Version 5) are MPSDM, PLUMES, and SCSTER.
Contrary to the implications of one commenter, a successful demonstration of
equivalency for COMPTER has not been completed, although the developer has
been given that opportunity by EPA. Since EPA does not control future
changes to these models which must maintain equivalence with improvements in
EPA's recommended models and more models may meet the test at a later time,
the list of equivalent models and options suggested by one commenter would
quickly become dated. However, EPA Regional Offices are kept informed about
models that have been shown to be equivalent. Further, it is the responsibility
of the Regional Office to determine those tests that are required for minor
input and output changes. Frequently a full equivalency test will not be
necessary but, even if required, the tests are rudimentary and normally
require no more than 10 simple model runs encompassing a variety of source
and climatic conditions.
Guidance on the use of fluid" modeling techniques is available.22-24
However, the technical community has little experience with use of such
techniques to evaluate mathematical models. Thus, any formal guidance on
how physical modeling can be used to evaluate mathematical models is premature.
3-12
-------
The requirement that an alternative model is not biased toward
underestimates, particularly where there is no preferred model, is necessary
to ensure that the NAAQS are met. This is not meant to require that a
grossly conservative model be used in lieu of one that is much less biased
and slightly underestimates design concentrations. However, compensation
for the underestimates must be provided; some means for doing this nave
evolved with applications of the Interim Procedures and are documented in a
supplement21 to that report.
3.3 Availability of Supplementary Modeling Guidance
Comment Summary (Model Clearinghouse)
Several commenters specifically endorsed the concept of a modeling center
or clearinghouse and urged that it include a formal advisory group that arbi-
trates modeling disputes. One commenter recommended that the clearinghouse
should provide guidelines and technical assistance and relegate final decisions
to EPA Regional Offices. Finally, a commenter felt that States should be free
to make minor deviations from the guidance where technically appropriate and
to consult EPA in complex situations. (ADEM, ADHS, APCA, CMA)
EPA Response
Jointly between the Model Clearinghouse of the Office of Air Quality
Plannng and Standards and the User's Network for Applied Modeling of Air Pol-
lution (UNAMAP) managed by the Office of Research and Development, the func-
tions of a clearinghouse for models are satisfied. The formal role of EPA's
Model Clearinghouse is to ensure consistency and technical adequacy of specific
modeling analysis. UNAMAP provides codes, user's guides and servicing for
a wide variety of air quality model applications. However, the Regional
Administrator is responsible for approval of any modeling technique used.
State deviations from the guideline concerning use of models for SIPs and
PSD are generally reviewed and approved by the EPA Regional Office; if the
3-13
-------
deviation is significant, a review by the Clearinghouse for conformance
with modeling policy may be requested. The Model Clearinghouse is always
available for consultation on complex issues. Also, refer to pages 79 and
80 of the Summary of Comments and Responses.1
Regulatory requirements for advisory groups tena to be burdensome for
both regulatory agencies ana the regulated industry. Although EPA has •
previously established advisory groups (SAB, CASAC, etc.), they are intended
to meet broad regulatory needs. Detailed administrative requirements must
be met in establishing such groups and the resource requirements to maintain
them are extensive. The time lost in using an advisory group to arbitrate
modeling disputes could result in undue delays in the decision-making process
and subsequent major costs to industry. Given resolution of a dispute by
such a group, the courts would be open for further argument if one of the
parties were not willing to accept the advisory group findings. For these
reasons, the formal use of advisory groups seems to be neither appropriate
nor practical in this instance.
Nevertheless, EPA recognizes the need for review by the scientific
community and has entered into a cooperative agreement with the American
Meteorological Society (AMS). Although a formal advisory procedure is not
used, the AMS through this agreement provides review and comment on the
scientific basis for the models, procedures and data bases required in regula-
tory processes. An example of the advisory aspects of the agreement are
found in the AMS publication entitled "Air Quality Modeling and the Clean Air
Act: Recommendations to EPA on Dispersion Modeling for Regulatory Applica-
tions. "25 EPA has subsequently initiated programs to implement many of these
recommendations. In conclusion, it appears that many of the positive
3-14
-------
contributions that a formal advisory group could make are being satisfied
through the AMS/EPA cooperative agreement, without introducing the liabili-
ties that can accompany such groups.
Comment Summary (Regional Meteorologists Workshops)
Several coimnenters argued that any changes in modeling policy or
"clarifications" resulting from the Regional Workshops on air quality model-
ing should be subjected to public comment and rulemaking before being
implemented. Otherwise, such workshops should be made open to the public.
Others urged State participation and conducting workshops within individual
Regions. Several suggested that a periodic newsletter should be published
releasing new information from the Workshops, Model Clearinghouse and the
Conference on Air Quality Modeling. Finally, one commenter expressed concern
about the impact of guideline revisions on previous decisions. (APCA, CMA,
FDER, KC, KOCH, MCC, MSUS, ODEQ, SOC, TVA, UARG, WC)
EPA Response
As stated in the modeling guideline, all changes to that guidance will
follow a formal rulemaking process. Prior regulatory decisions are normally
"grandfathered" and should not be affected by such changes or additions.
The purpose of the Regional Workshops is to ensure that guidance is
properly interpreted and applied. If as a result of such a workshop, model-
ing guidance and techniques are added or changed, these will be reflected
in the guideline and the rulemaking process will be followed as the commenters
suggested. However, if the result is to clarify procedures or make helpful
instructions, standard means of communication within EPA and with the
States, including appropriate policy memoranda, will be used. To improve
these communications, active participation by State representatives in
these workshops has been initiated, and is encouraged. Pre- or post-workshops
at the Regional/State level are also encouraged.
3-15
-------
To allow an efficient working atmosphere for these gatherings, they are
limited to EPA and State representatives who have a primary responsibility
for air quality impact assessments relating to SIPs and PSD. However, every
effort will be made to disseminate information of public interest as quickly
as possible.
Due to resource limitations, EPA does not have any immediate plans for a
modeling newsletter. As a minimum, distribution of information will continue
to be through NTIS, Federal Register notices, Docket A-80-46, and early
communications between the source and EPA/State authorities; these latter com-
munications are specifically encouraged in the guideline. All modeling guidance
and supporting information (including proceedings of all the Conferences on
Air Quality Modeling) have been made publically available through these
mechanisms.
3-16
-------
4.0 SIMPLE-TERRAIN STATIONARY-SOURCE MODELS
Comment Summary (Consistency Among Preferred Models)
One commenter suggested that EPA make certain model features consistent
in all preferred models. The commenter stated that in CRSTER, the calculated
effective stack height is reduced, by an amount equal to the elevation of
each receptor point above the stack base, before being compared to the mixing
height at that point. As a result, if a receptor is elevated sufficiently,
a plume that originally rose aoove the mixing height can be artificially
reintroduced into the mixed layer. A physically impossible result often is
the calculation of very high concentrations at the elevated receptor, but
no concentrations at all at another nearby receptor that might be only a
few meters lower. This problem does not exist in MPTER. (MES)
EPA Response
To foster consistency in the preferred models, the treatment of mixing
height in all EPA preferred models will be made to conform with MPTER and
this modification will be available in UNAMAP Version 6. In fact, all
preferred models with similar applications, as defined in Table 4.1 of the
guideline, have been modified to be internally consistent and the equivalence
established, as appropriate.
Comment Summary (Differences Between Short- and Long-term Models)
Many commenters indicated that different annual concentration estimates
will result depending on whether long term or short term models are being
used. In particular, the ISCLT and ISCST models were mentioned because of
the cost savings in running ISCLT over ISCST. The difference in predicted
concentrations is due to the data input requirements, i.e. short term models
require hourly meteorological data while long term models can use joint fre-
quencies. Comments varied, however, on this recommendation to EPA. Some
suggested that ISCLT (and CDMQC) be used for predicting monthly, seasonal and
annual average concentrations for complicated sources or urban areas, while
another suggested that it is more productive from a labor and resources stand-
point to allow for the option of using ISCST to obtain long term averages if
ISCST was already used for short term analyses. (APCA, CDH, NDDH, WCHD, SOC)
EPA Response
EPA recognizes that the two different meteorological data input require-
ments mentioned above will result in different concentrations. Each model
was developed for a specific type of application. Table 4-1 indicates
4-1
-------
a preference for ISCLT for estimating long term concentrations, however the
use of ISCST will be acceptable. Consequently, for long term applications
the user should evaluate the capabilities of each model in relation to the
problem at hand. If the modeler is interested in modeling complicated
sources for a pollutant for which short-term standards (i.e., 3-nour or
24-hour) are applicable, ISCST'may be used for all averaging times. Con-
versely, if the modeler is interested in such sources for a pollutant for
which long-term standards alone are applicable (i.e., quarterly or annual),
then ISCLT should be used.
Comment Summary (Modifications to Preferred Models)
Several commenters recommended modifications to the preferred models.
One recommended that EPA implement a rural version of the CDMQC Model.
Others recommended that EPA develop an urban version of BLP and develop a
refined model for shoreline and offshore sources. One commenter recommended
that ISC be modified to include urban dispersion coefficients. Another
recommended that EPA add to ISC an algorithm which accurately simulates the
behavior of buoyant emissions from roof monitors of industrial complexes.
(APCA, JCPL, AISI).
EPA Response
EPA does not recommend using the COM model for rural applications; ISC
is the preferred model for these aplications. EPA is not planning to develop
an urban version of BLP because EPA did not develop the BLP model and it is
the responsibility of the model developer to make this change. The BLP
model developer has been advised of revisions to EPA's preferred models so
that similar modifications might be made at the choosing of the developer;
however, there has been no indication that a change is planned. EPA has
conducted preliminary analyses of models for shoreline sources,26 but further
work is necessary before a model can be recommended for regulatory applica-
tions. A model dealing with offshore sources has been developed by the
4-2
-------
Department of the Interior, Minerals Management Services (50 £R 12248).
This model, the Offshore and Coastal Dispersion Model, will be submitted by
EPA under a supplemental notice of proposed rulemaking for public comment in
1986. The model is being considered for inclusion in the revised modeling
guideline as a preferred model. (Please also refer to comments in Section
7.2.7 dealing with "Other Governmental Programs").
The ISC model is being modified to include the RAM model urban dispersion
coefficients and is proposed in this guideline as a refined model for compli-
cated sources in urban areas ( refer to Table 4-1).
EPA has no plans to modifiy ISC to simulate buoyant emissions from roof
monitors. For simulating buoyant emissions from aluminum reduction plants,
or other similar applications, the BLP model is the preferred model.
Comment Summary (Changes to List of Preferred Models)
Several commenters suggested that EPA make changes to the list of preferred
models and their organizational content. A couple of these suggested phasing
out CRSTER; one suggested substituting for it with MPTER or ISCST and the
other said that there is no good basis for recommending CRSTER. Another said
that the MPTERU model should be a preferred model for urban applications and
should give identical results to RAM when both models have the exact inputs.
Another stated that more than one preferred model per category should be listed
in Table 4-1. Another suggested that for preferred models, the limitations
and proper applications should be clearly discussed in Chapter 4. (NYEC, NYCP,
IPL, CARB).
EPA Response
The CRSTER model is less time consuming to set up and substantially
less expensive to execute than the other two models mentioned. Therefore,
phasing out CRSTER cannot be justified at the present time, since it provides
concentration estimates equivalent to those from MPTER and ISCST for single
point sources. Also, the MPTER model is being modified by EPA to incorporate
urban dispersion coefficients. The urban option of MPTER gives results
equivalent to RAM for point sources.
4-3
-------
Table 4-1 does list several EPA models that yield equivalent results
when applied according to EPA recommendations (please also refer to comment
responses in Section 3.2 dealing with "Equivalency"). EPA's recommendations
on the proper applications of the preferred models is discussed throughout
the guideline, including Appendix A where many of the details of the features
of the recommended models are presented.
Comment Summary (New Screening Models)
A few commenters recommended that EPA modify existing screening models-.
One recommended that the PTPLU and PTCITY screening models be modified to
permit concentration predictions from multiple spatially separated sources
and not only a source at a single location. EPA should recommend how to
locate receptors and define worst case meteorology for these models.
Another recommended that EPA modify PTPLU to give concentrations at user-input
downwind distances and to run on a minicomputer. Another comment suggested
that EPA modify the wind speed ranges for each stability class in PTPLU so
they would be the same as those used in the refined models. One comment
wanted a description of the PTCITY model included in the guideline. (APCA,
FDER» IPL, NYEC).
EPA Response
An iterative application of these models, or a model such as MPTER, with
a qualified individual looking at the output and designing a subsequent run
if more information is required, is a better utilization of resources.
Determining receptor locations and defining worst case meteorology requires
judgement of a trained meteorologist, in consultation with the Regional
Meteorologist, on a case-by-case basis. With respect to modifying PTPLU and
PTCITY to give concentrations at user-input downwind distances, such a
change is unnecessary since MPTER can be used easily for this purpose.
4-4
-------
As to the third comment, EPA plans to modify the wind speed ranges in
PTPLU to correspond to those in refined models.
The PTCITY model will be combined in PTPLU-2 and will be made available
in UNAMAP Version 6. A description of screening models is generally not
included in the guideline but is included in the UNAMAP model package.
Comment Summary (Miscellaneous)
There were a few other miscellaneous comments. One suggested that the
guideline should either fully cover urban SIP modeling or reference documents
that do. Another recommended retaining the option to use CDM/Larson's analysis
techniques for determining 24-hour concentrations of TSP. Use of preferred
models may not provide better results for short-term calculations of TSP.
Another suggested that simple terrain be redefined to some specific level
below stack top elevation. (ADHS, IEPA, APCA).
EPA Response
A discussion on urban SIP modeling is available in several places in
the revised guideline. Case-by-case discussion of modeling concerns with
the Regional Office regarding specific urban areas is encouraged. EPA
disagrees that the CDM/Larson's analysis technique gives better results for
short term modeling of TSP concentrations. Since the comment did not
contain any validation data supporting its claim, there is no basis for
modifying EPA's proposal. RAM has been evaluated and its performance
documented.
EPA defines simple terrain to be an area where terrain features are
all lower in elevation than the top of the stack of the source. The com-
menter did not provide any analysis indicating how simple terrain should be
redefined.
4-5
-------
5.0 MODEL USE IN COMPLEX TERRAIN
Comment Summary (Model Development)
Many commenters suggested that EPA place a high priority, accelerate
its development efforts and recommend a refined complex terrain model based
on current research and review of candidate models because there is an
immediate need. Some of the commenters recommended that EPA routinely allow
the use of alternative, more realistic complex terrain models on a case-by-
case basis without validation as long as the source can demonstrate that
the model theoretically simulates the physics of plume behavior in hilly
terrain and the model is not biased towards underestimation. Another com-
menter asked how an alternate model can be evaluated since no reference model
is available for complex terrain. (API, TVA, NYEC, ADHS, EPNG, OEPA, PPL)
EPA Response
EPA appreciates the interest expressed in developing a suitable model
for complex terrain applications, and has conducted over the last five
years the Complex Terrain Model Development (CTDM) program.27,28 The goal
of this EPA/ORD program is to develop reliable atmospheric dispersion models
that are applicable to large pollutant sources located in complex terrain.
EPA has explained earlier the complexity of the problem on page 41 of the
Summary of Comments and Responses.1 As with other scientific endeavors, it
is difficult to accelerate research. Completion of the research project is
not expected before 1986.
EPA encourages the use of better alternate complex terrain models subject
to the requirements for consistency (please refer to comments in Section 1.0
dealing with "Model Consistency and Accuracy"). The use of a second or
alternate model for technical evaluation when no reference model is available
is fully described in Section 2.7 of the Interim Procedures.4
Comment Summary (RTDM)
Numerous commenters, notably from the utility industry, recommended that
EPA include the Rough Terrain Diffusion Model (RTDM) as a preferred model in
Appendix A. Among the reasons cited by those recommending the model are that
5-1
-------
the model has general acceptance of the modeling community and that EPA's
complex terrain model evaluation program has shown RTDM to have the best
overall statistical performance of all complex terrain models considered.
One comment stated that EPA should designate RTDM as either a screening tech-
nique or a refined model; RTDM's superiority to VALLEY and Complex I should
be recognized. EPA should allow use of RTDM in cases when VALLEY and
COMPLEX I predict that there will be an air quality impact problem. (UARG,
TVA, ERT, MSUS, TEGP, WC, PENE)
EPA Response
EPA has reviewed the performance evaluation of the RTDM model based on
additional information submitted during the public comment process. EPA will
issue a supplemental notice of proposed rulemaking that will seek public
comment on including RTDM with specific default parameters as a third level
screening model in the guideline. Full detail of the rationale for this action
will be given in a Federal Register notice. Modifications to the text in the
guideline will be proposed to reflect this change. RTDM can not be considered
as a refined model because of its tendency to substantially underpredict
concentrations as will be described in the above Federal Register notice.
Comment Summary (Treatment for Receptor Height)
Several commenters questioned the rationale for using a simple terrain
model on those receptors below plume height in complex terrain. One asked
how plume height should be established. If plume height is to be calculated
on an hourly basis using the standard model algorithm, substantial effort
would be required to address model selection on an hour-by-hour and receptor
basis. Two commenters questioned this approach from the physical sense and
stated that the atmospheric forces in complex terrain which cause increased
dispersion should effect concentrations at receptors below plume height as
well as above plume height. Recommendations to EPA included: (1) using an
average (constant) stable plume height for the entire year; (2) using the
complex terrain model only and specifically COMPLEX I which has been found
to overpredict maximum concentrations and, therefore, is sufficiently con-
servative; and (3) requiring the use of a simple terrain model with terrain
cut off at stack top when the only receptors above stack top are large
distances from the source being modeled.
Three commenters addressed the issue of when to recommend complex terrain
modeling. One recommended that complex terrain modeling should be required
whenever a receptor is above the legally permitted stack height. Another
said that complex terrain modeling should be performed when the receptor
5-2
-------
height is near the plume height. Another suggested that complex terrain
models should be applied for receptors at elevations between stack height
and plume height and that concentrations at all receptors above stack height
should be used in determining maximum source impacts.
One commenter recommended that a conservative estimate of the potential
temperature gradient (e.g. 0.035°K/m for F stability) be used for calculating
the critical streamline height. Plumes below this height should follow the
VALLEY 10m plume/terrain approach, while for plumes above this height a
refined modeling aoproacn should be used. (MES, CMA, APCA, NROC, CDH, NDDH,
MMES, NYEC, ODEQ)
EPA Response
EPA has recommended the use of complex terrain screening models until
research produces a refined model . Treatment of a situation where a receptor
on terrain is higher than actual stack height is uncertain. Perhaps the
present research activities will enable EPA to recommend a refined model
(such as CTDM) for those cases. However, in the absence of such an answer,
EPA recommends bracketing the highest concentrations by using both simple
and complex terrain models for receptors between stack height and plume
height. This technique would eliminate the need to determine the height
of the plume on an hourly basis, but would yield the worst case impact.
This method is accurate yet more simple than that proposed by the commenters.
The recommendation to use an average stable plume height for the entire
year is not supported by any data. The second recommendation to rely on
COMPLEX I model results alone may result in underpredictions at receptors
on terrain between stack height and effective plume height. This model has
not been evaluated for receptors below stack height. The third recommendation
to cut off terrain at "large distances" is ambiguous since that distance
cannot be accurately defined.
The definition of complex terrain, and thus when to use complex terrain
modeling, has been earlier explained on page 34 of the Summary of Comments and
5-3
-------
Responses^ and is reflected in guidance contained in the guideline. The
commenters do not present technical data to support any changes in the
definition of complex terrain.
An accurate calculation of the critical streamline height is one of the
goals of the present research activities. When such a method is developed
and adequately tested, it will be incorporated in cne refined complex
terrain model. The conservative estimate of the potential temperature
gradient suggested by one commenter for calculating the critical streamline
height is incorporated in the version of RTDM under consideration as an
additional screening technique.
Comment Summary (Modifications to Screening Models)
Several commenters suggested that EPA modify the screening models
COMPLEX I and VALLEY to make them less conservative and easier to use. Two
of these indicated that in the COMPLEX I model, the complex terrain option,
IOPT(25) should not equal 1 in all circumstances because it overpredicts by
a factor of 5. One of these two comments suggested only that EPA test
COMPLEX I with IOPT(25) equal to 2 or 3. The other comment recommended that,
until the CTDM model is available, IOPT(25) is set to 1 for true valley
applications, i.e., impaction on a tall mountain range. If an isolated
mountain exists, or if the maximum terrain elevation is near the effective
stack height, then set IOPT(25) to 2 or 3, whichever produces the higher
concentration. The comment further suggested that the Regional Meteorologist
make the decision whether or not to use this procedure.
One commenter suggested that the guideline should be reworded to state
"for screening analyses using VALLEY or COMPLEX I, full ground reflection
should always be used and a sector of 22 1/2° should be used unless represen-
tative, valid, site specific data of adequate temporal resolution demonstrate
that a broader or narrower sector is appropriate. Any such changes in sector
width would not affect the status of these models as preferred models."
Another comment recommended that EPA modify VALLEY to use NWS or on-site
meteorology to determine the maximum number of consecutive hours in which
the wind direction remains in a 22 1/2° sector with Class F stability and
use this maximum number of hours rather than 6 hours.
One commenter stated that the current form of the VALLEY model is difficult
to use, easy to make mistakes with and should be modified. Receptors should
not be restricted to the unusual grid with its scaling factor. Meteorological
input should be hourly data for short-term estimates instead of a frequency
distribution and have a set of default inputs for the 24-hour estimates.
5-4
-------
Furthermore, the assumption of a level plume (at constant elevation) and
sector averaging should apply to all stability categories. Another comment
suggested that the impingement treatment in VALLEY and COMPLEX I is only
applicable where steep terrain is inbedded in a uniform vertical temperature
structure. The use of these models, the comment stated, should be limited
to instances where these conditions are confirmed. (KC, MMES, APCA, ADHS,
NCNR).
EPA Response
EPA has recommended the use of the VALLEY and COMPLEX I models only as
screening techniques to determine concentrations from the impact of plumes
on hillsides. Setting the complex terrain option IOPT(2b) at 1 yields this
desired goal. On the other hand, option 3 assumes no "reflection" from the
ground (the estimates are half of those obtained from a standard application
of the Gaussian model algorithm). Also, neither option 2 nor 3 are physically
realistic and the commenter has not provided data to support this recommenda-
tion. Any changes to these models that will result in a decrease in the
conservativeness alone cannot be allowed as a substitute to developing an
accurate or refined model. As stated in the guideline, the present state
of knowledge on the interaction of plumes with complex terrain needs much
improvement, and thus there is no basis to support using options 2 and 3.
Furthermore, the commenters did not present any new data to support their
recommendations. Use of refined models, when developed, should remove the
concern for conservativeness that is present in the screening models.
The suggested word changes are consistent with the current content of
Section 5.2.1.4, with one major exception. The change implies that disper-
sion rates in the models can be changed without demonstrating the soundness
of this change following the Interim Procedures. Since this is inconsistent
with Section 3.2.2 of the guideline (as discussed in this document under
comments in Section 3.2 dealing with "Alternative Models"), no action will
5-5
-------
be taken on the word change. Furthermore, there is no need to modify the
VALLEY model as the second commenter has suggested because if on-site data
are available, they may be used in COMPLEX I. Also, if on-site data are
available, they can be used to specify a different worst case assumption
than that assumed in VALLEY.
EPA will modify COMPLEX I to include an option equivalent to VALLEY.
This would eliminate the commenter's concern with the receptor grid and
scaling factors. It is already possible to input hourly meteorological
data in COMPLEX I and no further change there is required. EPA does not
agree with the comment that VALLEY and COMPLEX I should be limited to steep
terrain and has provided its recommendation on the use of these models in the
guideline. The commenter did not provide any data to support this claim or
refute the EPA position. The issues of level plumes and impingement are
the subject of EPA's research program, and no changes will be made until
that program is completed.
Comment Summary (Guidance on Screening Models)
Several commenters indicated the need for further guidance when using
COMPLEX I and VALLEY screening models. One suggested that COMPLEX I should
be slightly modified by EPA to replace VALLEY as the primary screening
technique. The user could input the same meteorological assumptions as the
VALLEY model, use a 0.25 multiplication factor to convert 1-hour to 24-hour
concentrations and use a wind speed independent of height for F-stability.
Another commenter suggested that both VALLEY and COMPLEX I should be replaced
by SHORTZ. Another said that if only a few receptor points around a source
are above stack height, and preliminary screening indicates that they are
not likely to be critical receptors, consideration should be given to
"chopping-off" the receptors at the stack height and evaluating the source
with a refined simple-terrain model.
One commenter recommended that if the VALLEY model indicates a violation,
a second level screening must be mandated instead of "may be used" as is
now recommended by EPA.
Two commenters thought that the guidance on page 5-4, paragraph 4 which
discusses the use of receptor grids for VALLEY modeling is confusing. One
very often cannot tell whether a receptor lies above or below the plume
5-6
-------
centerline height. One will not know whether the worst-case receptor was
missed unless modeling with greater resolution, both vertically and horizon-
tally is performed.
A couple of commenters suggested there is a lack of guidance when model-
ing for unstable or neutral periods in a second-level screening analysis.
They noted that the discussion in Section 5.2.1 references stable conditions
only. When COMPLEX I (or SHORTZ) is utilized, computations will include
neutral and unstable periods when long-term averages are computed.
One commenter requested the use of the complex terrain algorithm in the
BLP model as an alternate screening technique to VALLEY and COMPLEX I when
treating emissions from aluminum plants which emit pollutants from short
stacks and roof vents atop short buildings.
Another commenter asked EPA to recommend a complex terrain modeling
technique for industrial sources subject to building downwash in rural
areas because the recommended screening models do not incorporate building
downwash algorithms. Furthermore, COMPLEX I cannot treat area sources,
such as fugitive emissions, and VALLEY is not recommended for seasonal or
long-term applications. The comment proceeded to recommend that modest
terrain be ignored and that ISCST and ISCLT be considered because the down-
wash algorithm mixes the plume rapidly so that concentrations are not likely
to be higher on elevated terrain than they are at flat terrain receptors.
In severe terrain, the comment recommended using the long-term version of
VALLEY. (MES, NCNR, NDDH, MMES, MCC)
EPA Response
EPA recommends using the COMPLEX I model as a second-level screening
technique when hourly on-site meteorological data are available. COMPLEX I
uses VALLEY as its basic algorithm but incorporates a half-height correction
for unstable plumes, adjusts wind speed with height and uses hourly wind data
as input. EPA plans to modify COMPLEX I as suggested by the commenter. How-
ever, replacement of VALLEY and COMPLEX I with SHORTZ is unwarranted at this
time since the SHORTZ model is a second-level screening technique for urban
areas where there are multiple stack, building and area sources. Also this
model did not perform better than COMPLEX I in the complex terrain evalua-
tion.10 Because of the relative complexity of SHORTZ, EPA cannot justify
denying the use of VALLEY when only a simple screening model is desired.
5-7
-------
The suggestion of "chopping-off" terrain if only a few receptors are
above stack height is partially consistent with EPA's recommendations for
receptors located between stack height and plume height. However, use of
simple terrain models in complex terrain could result in predicting arbitrarily
lower ground leva! concentrations. That is why EPA recommends applying both
flat terrain and complex terrain models in cases when there are receptors
above stack height and selecting the highest predicted concentrations. The
fact that there are only a few receptors cannot mitigate this situation.
There is no basis for making the word change regarding second level
screening as the commenter suggests because other alternatives may be avail-
able to the source without using a second level screening model. For example,
the source may reduce emissions until the predicted violation is no longer
present. Also, there may not be sufficient data available, such as one
year of on-site data, to warrant using a second level screening model. A
more conservative technique than that recommended is always an acceptable
basis for determining emission limits.
The location of receptors is established by first calculating final
plume height and comparing it to receptor terrain height. A fixed polar
grid may not include all the peaks in terrain and would need to be modified
to include additional receptors if they are within 10 meters from the center-
line of this final plume height. This problem will be mitigated by the
planned modification to COMPLEX I to include an option equivalent to VALLEY.
The discussion in the guideline has been limited to the stable condition
because this condition is the most likely to yield the highest ground level
concentrations for rural isolated sources. The formulations in the models
will be revised as knowledge about the behavior of the atmosphere in complex
5-8
-------
terrain increases and will be reflected in recommendations concerning any
refined model. For urban sources, guidance concerning neutral or unstable
conditions can implicitly be found in second-level screening models.
The complex terrain algorithm in the BLP model requires the use of
terrain correction factors. According to EPA evaluations, the choice of
terrain correction factors can have a significant effect on the estimated
second-high short-term concentration and thus the allowable emission rate.
Please refer to Appendix H of the Summary of Comments and Responses^ for a
discussion of these evaluations. Because of this uncertainty, EPA does not
recommend the BLP complex terrain algorithm for use, regardless of stability
class.
Due to lack of applicable technical information, EPA can not recommend
a complex terrain modeling technique for industrial sources as the commenter
suggests, but will accept a suitable model on a case-by-case basis. The
definition of "modest terrain" is very subjective and the use of a simple
terrain model could result in predicting arbitrarily low ground level
concentrations. Where the highest concentrations are shown to be near stack
base elevation in the wake, immediately downwind of a source complex, ISC may
be used. However, the feasibility of routinely using ISC in lieu of COMPLEX I,
as the last commenter suggested, can not be evaluated since the commenter
did not provide any data to substantiate this claim, nor made reference to
studies in scientific journals.
Comment Summary (SHORTZ/LONGZ)
Several commenters said that EPA has not justified specifying SHORTZ and
LONGZ for urban complex terrain applications. Two commenters pointed out that
the plume rise formulas and dispersion coefficient functions in SHORTZ and
LONGZ are different from those used in other EPA preferred models for urban
areas. Another commenter suggested that a few changes are warrented to
5-9
-------
enhance the utility of the models. One, it is desirable to install an option
that would allow the user to produce an hourly file of concentrations that
could be post-processed vi a a program such as CALMPRO to eliminate the
effects of calm winds and to produce high-five tables for each averaging
period. Two, an option is needed to select days to be included in the
analysis, as is possible with other guideline models. Three, the SHORTZ/LONGZ
meteorological data input formats should be modified to make them identical
to those of all the other models so that the same pre-processed meteorological
data sets can be used. (API, IPL, OEPA, APCA, SOC, AISI)
EPA Response
EPA recommends the SHORTZ/LONGZ model as a second level screening
technique for urbanized complex terrain areas because it provides a more
refined model than VALLEY; a better technique has not been submitted for
evaluation. Thus, this model fills a special niche. Evaluation studies
are cited in the SHORTZ/LONGZ user's manual. Although some of the options
are different from similar options in other EPA models, the SHORTZ/LONGZ
model has not been evaluated with these EPA options. Nor does EPA believe
it appropriate to require a developer to change the options without such an
evaluation.
EPA agrees that the utility of these models would be enhanced if the
commenter's suggestions were implemented. However, at the present time,
the vast majority of EPA's resources in this area are being used to develop
a refined model for complex terrain. If the commenters, or other interested
parties, are interested in making the recommended changes, EPA will provide
any technical guidance that may be necessary.
Comment Summary (Miscellaneous)
A couple of commenters indicated that paragraph 5 on page 5-4 of the
guideline should be more specific about how meteorological data should be
reviewed for spatial and temporal representativeness. A single site in
complex terrain is seldom ever representative of general conditions. One
commenter recommended that EPA provide additional guidance on the formulation
and evaluation of hybrid models involving a combination of a wind field model
and a diffusion model. (MFCS, NYEC, APCA)
5-10
-------
EPA Response
EPA acknowledges that establishing spatial and temporal representativeness
is difficult in a complex terrain setting and therefore has not issued any
detailed guidance on how to determine representativeness. Consultation by
experienced meteorologists with EPA Regional Offices is appropriate. Please
refer to page 48 of the Summary of Comments and Responses^ for further dis-
cussion on this position.
EPA can not give any guidance on hybrid models involving a combination
of a wind field model and a diffusion model because such modeling is still
in the research phase. However, EPA does not want to preclude applications
of this method on a case-by-case basis.
5-11
-------
6.0 MODELS FOR OZONE, CARBON MONOXIDE AND NITROGEN DIOXIDE
6.1 Discussion
Comment Summary (Reactive Pollutants)
One commenter indicated that stationary sources need to be isolated
from mobile and area sources before point source models referred to in
Sections 4 and 5 of the guideline can be used.
Two commenters suggested further guidance be given on selection of HO2
to NOX ratios from monitoring data for areawide urban modeling of N02-
Another commenter stated that guidance is needed for estimating the
ozone impact of rural VOC point sources on nearby nonattainment areas.
(APCA, SOC, ADHS, CDH)
EPA Response
EPA does not agree that stationary sources need to be isolated from
other sources before they can be treated as point sources. EPA intends that
point source models as discussed in Sections 4 and 5 of the guideline are to
be used for estimating the air quality impact of CO and NOX emissions from
stationary sources in urban areas. This has been clarified in the revised guide-
line. Section 6.2.3 provides additional guidance on how NOX concentrations are
to be converted to N02 concentrations.
EPA requirements regarding quality assurance procedures, site selection,
and data capture should be adhered to in the measurement of annual average
N02 concentrations. These are spelled out in 40 CFR Part 58. Since N02
is measured as the difference between NOX and NO, there are no special
*
requirements for N02 to NOX ratios beyond those for the measurement of N02«
EPA recognizes the need for guidance regarding the potential impact of
rural VOC point sources on urban ozone concentrations in nonattainment areas,
as well as small urban areas on themselves. Reactive plume models have been
developed which may serve as suitable analysis tools under some circumstances
6-1
-------
(refer to Appendix B of the guideline). Until such time as specific guidance
is developed based upon an evaluation of available techniques, modeling
techniques will be considered on a case-by-case basis. Any such techniques
must consider the chemistry of the specific organic compounds emitted and
the interaction between the point source plume and other sources of VOC and
NOX emissions.
6.2.1 Models for Ozone
Comment Summary (Urban Airshed Model)
Several commenters requested justification for selection of the Urban
Airshed Model as the preferred model. (APCA, SOC, OEPA)
EPA Response
The Urban Airshed Model is the most widely applied and evaluated
photochemical dispersion model in existence. EPA believes the evaluation
studies referenced in Appendix A of the guideline represent sufficient
justification for the selection of the Urban Airshed model as the preferred
model.
Comment Summary (EKMA)
One commenter suggested that EPA develop a modeling approach which
accounts for year-to-year fluctuations in the meteorological potential for
ozone formation so that such fluctations do not lead to changing control
requi rements.
Another commenter suggested that EPA modify EKMA to factor in or account
for anomalous meteorological conditions such as unusually high temperature
and that EPA consider procedures for predicting the probability of attain-
ment using estimates of model uncertainty. One commenter noted the sensi-
tivity of EKMA to certain input parameters and, in light of this, suggested
that the reliability of EKMA needs to be investigated. Another commenter
questioned the validity of applying EKMA except in urban areas dominated by
motor vehicle emissions and suggested the use of PLMSTAR or RPM-II instead.
One commenter indicated alternatives to EKMA for determining individual
point source impacts are needed and mentioned PLMSTAR and RPM-II as possible
models but suggested that simplified screening techniques need to be developed.
(OKIG, API, ADEM, CARB, UARG)
6-2
-------
EPA Response
EPA fully recognizes the influence of meteorological conditions on ozone
concentrations and the effects that year-to-year meteorological variations
might have on determining allowable emissions of hydrocarbons and nitrogen
oxides. In this regard, EPA is examining possible means for explicitly
treating meteorological, fluctuations relative to a more long-term climatolo-
gical condition. Until this procedure is developed and evaluated, EPA
continues to recommend the present approach used with EKMA in which three
successive years are selected for modeling. This provides a base period
with meteorological conditions that are broadly representative of several
years. Thus, year-to-year fluctuations are considered implicitly and the
dominance of any single year in determining control requirements is somewhat
mitigated.
EPA is considering modifications to EKMA which would more completely
reflect the role of ambient temperature and additional phenomena. However,
as described in Appendix A to the Guideline on Air Quality Models, the Urban
Airshed Model (UAM) is a preferred model for cases where there is interest
in modeling day-specific meteorological conditions in a more comprehesive
manner. UAM is a data intensive model which treats day-specific meteorolo-
gical conditions including wind speed, temperature, solar radiation, atmospheric
stability, and mixing height. EPA will consider the methodologies suggested
by the commenter for predicting the probability of attainment once decisions
have been made on how such information might be used in the regulatory
process.
EPA agrees that EKMA can be sensitive to certain input parameters, and
these have been identified in EPA guidance documents on the use of EKMA as
6-3
-------
cited in the Guideline on Air Quality Models (see for example "Guideline
For Use of City-Specific EKMA in Preparing Ozone SIPs," EPA-450/4-80-027).
These guidelines indicate those model inputs that should be measured and the
inputs for which default values will suffice. These recommendations arose
from evaluations conducted with EKMA as suggested by the commenter.
EPA believes that EKMA is an acceptable approach in a variety of urban
areas. Although the EKMA approach originally relied on a photochemical
mechanism developed to characterize motor vehicle exhaust, EKMA can now be
used with the most recent version of the Carbon-Bond Mechanism.29 This
mechanism has been widely evaluated against smog chamber simulations of
surrogate urban atmospheres and has been successfully employed in photochem-
cal dispersion modeling of actual urban areas. PLMSTAR and RPM-II are point
source models and as such are inappropriate by themselves for areawide urban
ozone applications. For these applications, the Urban Airshed Model is
preferred, although EKMA may also be used.
EPA does not regard EKMA as an appropriate method for assessing ozone
impacts from individual point sources and, therefore, EKMA is not an accept-
able approach in this instance. EPA does not recommend a preferred model
for VOC point sources at this time. As indicated in Appendix B of the
guideline, there is no specific recommendation for the use of PLMSTAR or
RPM-II. These models may be applied on a case-by-case basis. EPA acknowl-
edges the need for screening techniques for estimating potential ozone impacts
from VOC point sources.
6-4
-------
6.2.2 Models for Carbon Monoxide
Comment Summary (CO Line Source Models)
Most commenters supported the use of CALINE3 as the preferred line
source model. Two commenters suggested*approval of all future versions of
CALINE. Two other commenters suggested that both CALINE3 and HIWAY-2 should
be listed as preferred models. One commenter suggested that recommendations
for CO modeling approaches be delayed until completion of EPA's current
evaluation program and that program be limited to tracer data evaluations.
One commenter suggested that the discussion of data needs for line source
models on page 9-4 of the guideline be made more general to ensure that the
data needs of the preferred model, CALINE3, match those mentioned. (WDNR, CDOT,
FHA, NYCP, GMC)
EPA Response
Each revision of CALINE must be evaluated to determine the appropriateness
of modifications before the model can be approved by EPA for general use.
Automatic inclusion of revisions of CALINE without first evaluating the
performance, technical adequacy, and effect of such revisions would not be
responsible. For example, CALINE4 is more difficult to use than CALINE3
because it requires some inputs for which data is not routinely available,
such as sigma theta. Thus, since the concentrations given by the two models
differ very little, the use of CALINE3 is recommended.
Although the differences between CALINE3 and HIWAY-2 in terms of theory,
data needs, and results may not be substantial, CALINE3 is more widely used
throughout the modeling community and has provided a broader basis for CO
control strategies. Thus, because of its widespread use, CALINE3 is listed
as the preferred model. Accuracy would not be increased by including models
not found to be better and would only lead to "shopping" for the model
which gives results closer to desired concentrations. An evaluation to test
6-5
-------
the relative performance of the CO models is underway and will be forwarded
for peer scientific review.
EPA's program for evaluation of mobile source CO models is intended to
review specific applications of the techniques under investigation. The
process of completing this review, making recommendations, and soliciting
public comment will require more time. The Agency believes that sufficient
information based on completed performance evaluations, past use and famili-
arity exists at this time to support the recommendations made. In addition,
EPA intends that the model evaluation program be a continuing process.
However, there is a need to prescribe a recommended model at this time.
Finally, EPA's current program for evaluation of mobile source impacts
relies on data bases containing both carbon monoxide air quality data as
well as appropriate tracer studies. The Agency recognizes the potential
effect.of background interference and believes that the process of data
selection in the model evaluation process ensures the proper use of such
data and therefore believes that it is proper to include results from both
types of analyses when evaluating the performance of such models.
The data needs for line source models mentioned on page 9-4 of the
guideline are general and match those required for CALINE3, with the excep-
tion of pollutant emissions where the (grams per second per meter) will be
dropped. Detailed information on data requirements for CALINE3 are intended
to be obtained from the user's guide.
Comment Summary (CO Model for Special Situations)
Two commenters suggested that EPA specify the model or technique to use
to evaluate occurrences when monitors have measured exceedances of the CO
8-hour ambient standard in the late night and early morning hours over a
6-6
-------
wide area. These exceedances are not related to specific "hot-spots."
(FDER, ADHS).
EPA Response
EPA recognizes that these exceedances occur and is investigating
suitable models to handle them. The most appropriate approach may involve
use of urban area modeling requiring considerable resources and technical
expertise. Guidance will be expanded at a future time as the information
base on the best and most cost effective approaches evolve. If a suitable
model is available for a specific application and the data and technical
competence for its use are available, then such a model should be considered.
Comment Summary (Techniques for Intersections)
Several commenters suggested that the guideline specify a technique for
modeling intersections. One commenter suggested the use of the Intersection
Midblock model, while another recommended the use of the Texin model. (ADHS,
FHA, NYDOT, NYEC, DOT).
EPA Response
EPA agrees with this comment and will revise the model guideline to
indicate that Worksheet 2 of the "Guidelines for Air Quality Maintenance
Planning and Analysis Volume 9 (Revised): Evaluating Indirect Sources" be
used to determine modal (acceleration, deceleration, idling, and cruise)
emission factors for input to the preferred dispersion model, CALINE3, when
intersections are modeled. The Intersection Midblock Model is not recommended
because it uses the outdated HIWAY dispersion model.
TEXIN has a simplified version of MOBILE2 built into the model and can
handle only simplified intersections. The version of MOBILE2 built into TEXIN
does not allow the consideration of Inspection/Maintenance controls. In
addition, the simplfied treatment of intersections does not allow the con-
6-7
-------
sideration of certain transportation control measures. Therefore, TEXIN is
not recommended for regulatory analysis.
6.2.3 Models for Nitrogen Dioxide (Annual Average)
Comment Summary (Stationary Source Models)
One commenter requested additional guidance on the use of Appendix A
models for N02« Another commenter recommended the ISCLT model. (APCA, HMES).
EPA Response
Additional guidance for selection of suitable Appendix A models for
stationary sources is covered in Sections 4 and 5 of the guideline. These
sections should be referred to in selecting a single source dispersion
model for use with the three-tiered screening approach for point sources or
a multiple source dispersion model for urban areas. The ISCLT model can be
used for either a first or second level screening analysis for point sources
in those situations for which it is recommended, as discussed in Section 4
of the guideline.
Comment Summary (Ozone Limiting Method)
One commenter requested additional guidance on the use of the Ozone
Limiting Method. Another suggests that more refined techniques be recognized
in the guideline as alternatives to the Ozone Limiting Method. Another com-
menter questions the use of the Ozone Limiting Method for determining annual
average concentrations from point sources. (APCA, EPNG, API, NYCP)
EPA Response
The Ozone Limiting Method is described in the reference cited in the
guideline. EPA believes that more refined techniques have yet to be shown
to be more suitable for point source applications.30 The commenter does not
identify alternative techniques for EPA to evaluate. However, as stated in
the guideline, more refined techniques may be used on a case-by-case basis.
6-8
-------
Comment Summary (N02 to NOX Ratios)
One commenter requested guidance on the use of N02 to NOX ratios when
future emission controls may alter the baseline ratios derived from ambient
measurements. (APCA)
EPA Response
When future emission controls are expected to substantially alter N02
to NOX ratios on an annual basis, more refined modeling techniques can be
considered on a case-by-case basis. As stated in the guideline, photo-
chemical dispersion models may be applied in situations that require more
refined techniques.
6-9
-------
7.0 OTHER MODEL REQUIREMENTS
7.1 Discussion
Comment Summary
One commenter suggested that the guideline state or reference topics
for which special regulatory program guidance documents have been prepared.
If such documents have not been prepared, then some interim form of guidance
should be provided in the guideline. (SOC)
EPA Response
Section 7 highlights several program areas important to modelers. It
is not the purpose of this guideline to serve as a compendium of information
because these programs are subject to change and references may become out-
dated. Such changes and any new guidance are subject to public review and
comment under the rules pertaining to the specific program.
7.1 .2 Fugitive Dust
Comment Summary
Many commenters recommended that EPA provide more guidance on fugitive
dust emissions estimation and modeling procedures. One said that fugitive
emissions should not be considered routinely, but only where the specific
information is available and it is likely that such emissions could be mak-
ing a significant air quality impact. Another said that since the release
of fugitive emissions is unique to each plant, emission factors developed
in a generic manner will not be representative. One submitted a couple of
reports on fugitive emissions from utility sources and suggested that they
be referenced in the guideline. Another suggested that EPA develop emission
factors for the forest product industry.
With respect to modeling, these commenters suggested that the upcoming
PM-|Q modeling efforts will require a need for particle size distribution
data, wind erosion rates, and fugitive modeling techniques that include
multiple-hour transport and accumulation. One asked if naturally occurring
dust sources should be modeled. Another stated that the discussion of fugi-
tive dust should be separated from fugitive emissions because fugitive dust
is different in terms of both physical and chemical properties. By inserting
the second paragraph between the first and third, EPA has implied that ISC
is recommended for modeling fugitive dust. This model should not be recom-
mended for any modeling where deposition is important because the ISC
treatment of deposition places a discontinuity in the Gaussian distribution
at the surface, a fundamental error which has no theoretical or empirical
7-1
-------
basis, and prevents it from giving accurate predictions. Another said that
for fugitive emissions from haul roads, the CALINE3 model should be recom-
mended instead of ISC. One other comment suggested that EPA incorporate
some methodology to treat pit retention in any model for mining operations.
(APCA, TVA, ADHS, ADEM, UARG, AMC, BAAQ, MMES, DPC, WC, OEPA)
EPA Response
If actual source-specific emissions data are available, ^rtese data may
*
be used. However, if these data are unavailable, guidelines on characterizing
fugitive dust emissions given in EPA's AP-42 pub!ication^l should be used.
This recent Fourth Edition includes new fugitive emission factors for unpaved,
paved urban and industrial paved roads (Sections 11.2.2, 11.2.5 and 11,2.6,
respectively). When estimates of emissions for load handling operations are
desired, AP-42 contains fugitive emission estimates based on studies conducted
for EPA at twelve major coal fields in Western states. Since AP-42 is EPA's
official vehicle to publish emission factors, users of AP-42 emission factors
should make sure that the most recent updated factors available are being used.
EPA has reviewed the submitted publications by utility sources but finds
that since certain portions are not in accord with information in EPA publica-
tions, they are not appropriate as references. Emission factors for the forest
product industry are given in Chapters 10 and 11 of AP-42.
EPA agrees that guidance is needed to model fugitive emissions from
non-traditional sources. EPA has developed guidance on characterizing PM-JQ
fugitive emissions from such sources.32*33 Until such guidance has been
promulgated, it is inappropriate to provide further air quality dispersion
modeling guidance. Naturally-occurring dust sources are a component of
background and are estimated from monitored data representative of the
site. Area source emissions are another component of background and are
established from air quality modeling.
7-2
-------
The ISC model has been evaluated in terms of its deposition algorithm,
against three deposition experiments.34 Results show that at least 80 per
cent of the ISC model calculations were within a factor of. two of the
experimental values. EPA agrees that there are several deposition algorithms
available in the literature. Moreover, scientific understanding of the
mechanisms involved in deposition is increasing. EPA plans to evaluate
further the category of complex industrial source models over the next two
years and will subsequently make recommendations concerning such models as
the commenter recommended.
EPA does not agree with commenter1s suggestions to reorganize the text.
The ISC model may be used to model fugitive dust sources such as coal and ash
storage piles as discussed in the user's manual for this model.
EPA has not received any information demonstrating that CALINE3 performs
better than ISC for haul roads. Thus, where haul roads are reasonably con-
sidered part of an industrial complex, the ISC model is recommended for use.
EPA is currently developing a methodology and model algorithm for
treating pit retention of particles during mining operations.35,36 This effort
is proceeding in cooperation with the National Coal Association. However,
a suitable data base against which the algorithm can be tested has not yet
developed. EPA will report its findings as this project proceeds.
7.2.2 Participate Matter
Comment Summary
One commenter suggested the use of chemical and physical analysis of
particulate matter samples in addition to or in place of source receptor models
while another recommended the use of receptor models.
Two comments requested clarification as to whether the ISC model should
be used in all urban modeling when particle settling and deposition are
involved. (SOC, MMES, ODEQ, NYCP)
7-3
-------
EPA Response
The use of chemical and physical analysis is not required for routine
modeling applications but can be used on a case-by-case basis when special
situations arise. The suitability of using receptor modeling is addressed
in the responses to comments in Chapter 11; EPA encourages receptor modeling
as an adjunct to dispersion modeling.
The ISC model should generally be limited to analyses of industrial
source complexes in either urban or rural areas. Although the model has not
been evaluated in terms of its performance for an entire urban area, v/hen
particle deposition and settling are involved, it employes standard algorithms
applicable to urban areas. In conjunction with the PM]Q NAAQS program, EPA is
developing a short term model for urban areas which.will treat dry deposition,
sedimentation and first-order chemical transformations including aerosol
formation. Performance evaluations are underway and the model should be
available in late 1986.
7.2.3 Lead
Comment Summary
One commenter recommended that the text which states that CALINE3 is
unable to account for particle deposition be changed because optional
deposition and settling algorithms are, in fact, included. Another stated
that lead can be accurately modeled using CALINE3 and APRAC-3 models with
simple modifications. Another questioned the significance of using 4.0
yg/m^ as a cut-off value since the NAAQS for lead is a quarterly average of
1.5 yg/m3. One recommended that models for estimating ambient lead levels
must be able to account for deposition and long term (three month) impacts.
(DOT, MMES, OEPA, NYEC)
EPA Response
EPA agrees with the first comment and has corrected the text on page 7-5.
However, CALINE3 and APRAC-3 have not been evaluated in terms of modeling
7-4
-------
quarterly lead concentrations from automobile sources along roadways. Until
such evaluations are undertaken and reviewed, ISCLT is the recommended
approach. EPA agrees that models for estimating ambient lead levels must
be able to account for deposition and ISCLT does so. The 4.0 pg/m3 cut-off
value is obtained from 40 CFR 51.83 which was subjected to prior public
comment and rulemaking. (Also see 42 FR 63U87).
7.2.4 Visibility
Comment Summary
Three comments stated that the guideline should provide guidance on
choosing the appropriate visibility models and input parameters (e.g.,
observer geometry, type of background, etc.) that give worst-case visual
impacts. One of these coments recommended that the guideline outlined in
1980 (40 CFR 51.300-307) for modeling visibility should either be presented
or mentioned in this section (ADHS, CDH, MMES).
EPA Response
The visibility regulations anticipate that the minimum requirements-for
worst-case visual impacts are determined by analogy from visually observing
sources of the same character as the proposed new sources. The state-of-
the-art in visibility models at the time of regulatory development did not
permit the Agency to require such analyses. States are, however, encouraged
to use the results of visibility modeling analyses when available, but are
not required to approve or disapprove a source permit on the basis of
specific modeling results. It is not appropriate to list in Appendix A of
the guideline any visibility models until such modeling is required by the
regulations. However, when modeling is done, models listed in Appendix B
may be used. EPA believes that visibility model inputs for worst case
impacts are best determined on a case-by-case basis, especially given the
7-5
-------
variety of Class I areas and special concerns that Federal Land Managers
may express. The text will be revised to include direct reference to
the regulation.
7.2.5 GEP Stack Height
Comment Summary (Stack Height Credit)
One commenter recommended that EPA require downwash modeling if the
stack height that a source claims for purposes of receiving credit is less
than the GEP height. Another said it is unreasonable to require an analysis
for stacks only marginally less than GEP height. A minimum value, such as
H + L, should be established. One comment recommended that when simulating
actual air quality (e.g., for model validation), the actual stack heights
should be used; but when setting emission standards, GEP height should be
employed. Another suggested that modeling existing point sources at GEP
height precludes the validation of model results by use of ambient measure-
ments, particularly in the case of tall stacks. The model should be verified
for existing stack heights by use of actual monitoring data before it is
used to model GEP stack height releases.
One commented that if EPA is relying on the downwash algorithm to
identify the need for additional control measures involving an existing
stack, it should also be willing to accept the algorithm in lieu of a fluid
model to justify a stack height increase.
One commenter asked if the wakes due to nearby terrain obstacles should
be considered in the building downwash analysis.(NRDC, FDER, MCC, MMES, OEPA)
EPA Response
Section 123 of the Clean Air Act, as well as 40 CFR 51 revised in July
1985, defines good engineering practice (GEP) stack height. Many of the
comments presented above have been addressed in the Response to Comments
document for that rulemaking and are available in Docket A-83-49. A
brief response is given below.
When the stack height of a source is below what EPA has set as GEP,
downwash is suspected to occur and ambient standards or PSD increments may be
violated. The modeling guideline requires downwash analysis for sources with
stacks less than the height defined by EPA's refined formula for determining
7-6
-------
GEP heights. There is no rational basis to exclude stacks "marginally less
than GEP height" as the commenter desires. For a more detailed explanation
of all the requirements please refer to the Guideline for Determination of
Good Engineering Practice Stack Height.23
EPA agrees that modeling existing point sources at GEP height may
preclude the validation of model results by use of ambient measurements in
the future; however, no model evaluations thus far have been conducted in
this manner and EPA is working to develop strategies to address this problem.
EPA requires the use of fluid modeling for sources who wish to receive
a greater stack height credit that can be provided by the applicable formula.
Control measures may be based on the results of the downwash algorithm if
downwash is the controlling meteorological condition. Criteria for when
wakes due to nearby terrain obstacles may be considered in the overall
downwash analysis are given in the regulation and described in the Guideline
for Determination of Good Engineering Practice Stack Height.23 The downwash
algorithm in ISC assumes downwash does not occur for any stack that meets
the height criterion. Wind tunnel results indicate that the excess concen-
tration associated with not meeting that criterion may range from 20% to 80%.
Since the GEP regulations specify 40%, only a wind tunnel demonstration
can determine the stack height at which that criterion is satisfied. EPA
is very concerned that credit for increasing stack height not be granted
without a comprehensive demonstration.
Comment Summary (Modify ISC Downwash Algorithm)
Several comments suggested improvements to the ISC model algorithm for
treating building downwash. A couple recommended that the building dimensions
used in the model should be made directionally dependent, i.e., for each
wind direction the model should be able to specify a different set of cross-
wind building dimensions. Also, EPA should limit or reduce the extent of the
downwash analysis for wind speeds less than some designated critical wind
speed. Another stated that the ratios of stack height to building height
7-7
-------
often do not occur within the range for which the ISC downwash routine was
developed and recommended that either downwash routine be modified or these
stacks be exempt from downwash analysis.
A couple of commenters stated that the downwash algorithm has not been
properly validated and should not be required for use. One of these stated
that recent measurements have demonstrated the performance error in the ISC
downwash algorithm and that the new downwash algorithm in BLP model be
recommended.
One commenter recommended that more guidance should be provided OR
what screening models to use for sources with less than GEP stacks located
in complex terrain while another suggested that the detailed downwash screen-
ing procedure contained in the "Regional Workshop on Air Quality Modeling"
be included in the guideline. (APCA, TVA, ADEM, AISI, WC, AMC, CDH)
EPA Response
EPA agrees that using building dimensions in the ISC model that are
dependent on the wind direction as the commenter suggests appears to be
technically sound and plans to propose for public comment a modified version
of ISC recently submitted to the Agency that contains this improvement.
However, there is no technical justification for arbitrarily reducing the
extent of downwash for any wind speed because downwash has been shown to
occur in field studies with wind speeds as low as 1.8 m/s.37 As stated
earlier, stacks can not be arbitrarily exempted from a GEP/downwash analysis
because this is part of the stack height regulation.
EPA agrees that obtaining additional data bases to allow further valida-
tion of the ISC algorithm is desirable. However, existing evaluations
provide adequate support for using the algorithm, although underpredictions
are indicated. Evaluations of the BLP downwash algorithm are much more
i
limited. As noted above, EPA plans to propose a modification to the ISC
downwash algorithm sponsored by the American Petroleum Institute. Evaluation
studies indicate superior performance of this algorithm, although generally
higher air quality estimates result from its use.
7-8
-------
For sources in complex terrain with less than GEP stack height, the
impact analysis should determine which situation e.g., terrain impaction,
stability category A, or downwash, etc. produces the highest concentration
estimate for downwash. The emission limit should be determined therefrom.
Because the guideline does not contain a description of any screening
model, the reference to the Regional Workshop report for the downwash screening
procedure is sufficient.
7,2.6 Long Range Transport
Comment Summary
A few commenters suggested that EPA should recommend appropriate long-
range transport models (LRT). A few others, however, disagreed. One stated
that since the use of such models could not be recommended for regulatory
applications, any reference to these models should be deleted. Another
stated that the accuracy of Gaussian models decreases after 20 to 30 km and
they are inaccurate to use to 50 km. Another suggested that EPA specify a
maximum distance at which such models may be used to determine the impact
on a Class I area. One commenter recommended that a significant impact on
a Class I area be defined as 5%'of the applicable PSD increment or some
other specific value. Then, any source located more than 50 km from a
Class I area whose impact falls below the significance level at a distance
of 50km or less could be exempt from the requirement for LRT modeling. This
would limit the number of sources subject to the case-by-case selection of
a model to only those which it is most justified. The commenter also stated
that it is burdensome to require the Federal Land Manager, the EPA Regional
Office, and the PSD permitting authority to confer on procedures for evaluat-
ing the long-range impacts of all projects subject to PSD review, within
100 km distance from a Class I area.
A few commenters asked for text clarification. One stated that the last
sentence of the first paragraph is unclear. A couple questioned the last
sentence in the second paragraph. There have been no field studies on LRT in
complex terrain and this makes the evaluation and use of these models according
to Section 3.2 procedures impossible in some cases. Another stated that the
first sentence in the second paragraph is unclear since EPA does not intend to
allow the use of LRT models for regulatory applications to determine the effects
of S02 emissions from sources in one region on ambient levels of another pollu-
tant (e.g., sulfates, TSP) or on deposition level in another region. (TVA,
APCA, NYEC, CDH, ADHS, MMES, CHEV, PHC, ASRC, IEPA, FDER)
7-9
-------
EPA Response
EPA is currently in the process of evaluating eight short-term, LRT models
againsj two data bases. These are: TVA's (ARRPA) model; ERT's (MESOPUFF,
MESOPUFF II, MESOPLUME) models; North Dakota's (MSPUFF) model; Combustion
Engineering's (HTDDIS) model; Dames & Moore's (RADM) model, and SAI's (RTM-II)
model. A description of some of these models is included in Appendix B of the
revised guideline. The Department of Energy is also evaluating several other
LRT models against a third data base. Pending a review of these and other
evaluation studies reported in the literature, EPA intends to prepare a guidance
document on the application of LRT models for appropriate regulatory issues at
some future date. EPA is also developing plans to evaluate, and if necessary,
improve LRT models applicable to determining Class I area impacts for PSD sources.
Models for LRT applications should use meteorological data of sufficient spatial
coverage to overcome the difficulties mentioned by the commenter. EPA disagrees
with the comment that there should be no reference to LRT models in the guideline
since such models are needed under the Clean Air Act, and Section 165 of this
act does not specify any distance limitation beyond which the air quality impact
of a source on a Class I area need not be determined.
Using any specific value, such as 5% of the applicable PSD increment,
to exempt sources has no basis in the current regulation and is not a subject
for consideration in this rule-making. Instead, the regulation specifies
significant emissions as the means of exempting sources from PSD review.
The Clean Air Act requires the inclusion of the Federal Land Manager, the
State agency and the EPA in reviewing sources that may impact Class I areas.
-i
A change has been made to clarify the last sentence of the first
paragraph. The first sentence of the second paragraph will be clarified.
7-10
-------
7.2.7 Modeling Guidance for Other Government Programs
Comment Summary
One commenter recommended that the guideline provide a listing by refer-
ence of the air quality modeling requirements of all other Federal agencies.
Another suggested that since the State normally has jurisdiction in
PSD permit applications, text should be changed to state that in Class I
areas, the FLM should consult with the State(s) involved on all modeling
questions.
One commenter requested that EPA include the Offshore and Coastal
Dispersion (OCD) Model as a recommended model for application to sources
located over water or in nearshore coastal areas. (APCA, MMS, CHEV, SRP)
EPA Response
The guideline does reference the air quality models used by some other
Federal agencies; however, the applicant should review the modeling require-
ment with the government agency in question. The role of the Federal Land
Manager (FLM) with respect to handling air quality impacts on Class I areas
is defined in Clean Air Act such as in Section 160, 169A, etc. EPA regula-
tions require the State to consult with the FLM regarding PSD permit appli-
cations. EPA will propose in a supplementary notice of proposed rulemaking
to include the OCD model as an Appendix A model in the guideline. The
model would be limited in application to off-shore oil/gas facilities and
their on-shore impact.
7-11
-------
8.0 GENERAL MODELING CONSIDERATIONS
8.1 Discussion
Comment Summary
One commenter stated that the guideline should allow more sophisticated
treatment of the mixing height phenomena than the Holzworth method. Another
requested guidance on how mixing heights should be considered when on «?I>P
data are used. (ADHS, MMES) *>"-:> ite
EPA Response
Several methods are available in the literature for the calculation of
mixing heights. To date, no comparative analyses have been presented to
EPA in terms of the effect on model results. Therefore, a change is not
warranted until such analyses have been done. Although EPA will continue
to use the CRSTER user's guide method for the calculation of mixing heights
this does not preclude the use of other systems or the use of on-site data
Other systems and on-site data will be evaluated on a case-by-case basis
EPA is also presently investigating a new approach to develop hourly mixing
height for future air quality models that use on-site turbulence information
8.2.1 Design Concentration
Comment Summary
A few commenters requested that EPA define what is meant by "highest
second-highest short term concentration and provide further guidance on how
to determine this design concentration, i.e. whether it includes short term
background values, and whether it was determined for each year separately
or for the combined period. One commenter recommended that EPA should count
the second-highest concentration in the receptor network as the design con-
centration. Emissions should be rolled back to prevent exceedance at the
second-high receptor in the network. One comment suggested that the EPA
explicitly state how the PSD increment consumption should be calculated
using air quality models. (NRDC, NYCP, MMES)
EPA Response
The design concentration is the sum of the short-term background value
(except for PSD) and the highest, second-highest source impact. EPA deter-
8-1
-------
mines the highest, second-highest short term concentration by (1) ranking
the predicted concentration at all receptors (2) selecting the second-high-
est value at each receptor and (3) subsequently selecting the highest of
the values identified under (2). For annual averages, the source impact
is determined for each year separately, and the highest value is selected.
EPA considers the commenter's recommendation to consider the second-highest
concentration in the receptor network as the design concentration to be
inconsistent with the NAAQS.
EPA recommends modeling short-term PSD increment consumption on both a
spatially and temporally consistent basis. The maximum amount of PSD incre-
ment consumed must be determined by modeling the net changes in emissions
(between the baseline and future cases) sequentially for each time period
with at least a full year of meteorological data. The^ resulting maximum
impacts of this type of analysis specify the maximum amount of increment
consumption at each receptor. Please also refer to responses to comments
in Section 11.2.3. It is not feasible to provide more explicit guidance;
only broad principles can be stated in the guideline. A case-by-case
determination is needed.
8.2.2 Critical Receptor Sites
Comment Summary (Ambient Air)
There were several recommendations that the guideline specifically
define the areas accessible to the general public where the NAAQS and PSD
increments apply and, hence, where receptors should be located. Some
suggested that text be added stating receptors need not be placed within
plant property while another stated that inherently, the receptor array is
limited by the operational definition of "ambient air."
A couple of comments addressed the issue of locating receptors within
100 meters from a stationary source. One suggested that no receptors be
8-2
-------
placed within 100 meters because of the limitations of the dispersion
parameters. However, the other recommended that EPA allow placing receptors
within 100 meters of the source, but said that EPA should propose the most
appropriate way to estimate concentrations within this distance. (APCA, TVA
WDNR, ADHS, FDER, NYCP)
EPA Response
The placement of receptors in all ambient air locations (as defined in
40 CFR 50.1(e)) should be considered. It is EPA's policy (outlined in a
letter from Costle to Randolph on December 19, 1980) that the exemption
from ambient air is available only for the atmosphere over land owned or
controlled by the source and to which public access is precluded by a fence
or other physical barriers. Therefore, for modeling purposes the air
everywhere outside of contiguous plant property to which public access is
precluded by a fence or other effective physical barrier should be considered
in locating receptors. Specifically, for stationary source modeling, receptors
should be placed anywhere outside inaccessible plant property. For example,
receptors should be included over bodies of water, over unfenced plant
property, on buildings, over roadways, and over property owned by other
sources. For mobile source modeling (i.e., CO modeling), receptors should
continue to be sited in accordance with Volume 9 of the "Guideline for Air
Quality Maintenance Planning." EPA will continue to review individual
situations on a case-by-case basis to ensure that the public is adequately
protected and that there is no attempt by sources to circumvent requirements
of Section 123 of the Clean Air Act.
The EPA model RAM currently allows receptors to be located beginning
at 1 meter from any source where that is necessary to meet the ambient air
criteria. EPA is planning to introduce this capability into the regulatory
option for all of its models. However, because the ISC model is not appro-
8-3
-------
priate for estimating concentrations within the cavity region of buildings,
receptors within this region will be precluded from the calculation.
Comment Summary (Receptor Density)
There were several comments for additional guidance on how to locate
receptors and determine receptor grid size and spacing. One stated that
the geometric progression method to determine the downwind distance of
receptor rings should not be used, but a better method is needed. This
method should also show how to locate receptors that will determine the
combined maximum concentrations produced by two or more sources. Also, the
definition of "large sources" for receptor location purposes should be in
terms of total heat input or emission rate.
Another commenter stated that it is burdensome, costly and unnecessary
to require the use of too many receptors and recommended that in flat
terrain cases where there are no complicating source factors, EPA should
accept fewer than 400 receptors. In complex terrain or other, more compli-
cated cases, EPA should not require more than 400 receptors, as long as
this set includes many receptors that showed significant impacts in previous
modeling efforts. As to receptor spacing, one comment suggested a maximum
receptor spacing of 100 meters be used for final modeling of high impact
locations. A couple of other commenters suggested that the location and
number of receptor sites be determined from the results of a screening model.
(APCA, WDNR, SOC, UARG, ODEQ)
EPA Response
Due to the subjective nature of judgments about the location of the
highest concentration, the guideline provides only general direction and
allows for a reasonable amount of flexibility. More detail is provided on
page 101 of the Summary of Comments and Responses.1
The geometric progression method was a specific method recommended by
EPA.3 This technique has limitations, as the commenters point out, and is
therefore not being considered further. EPA believes that no additional
generic guidance is required and that the final decision on the choice of
critical receptor sites should be arrived at between the applicant and the
regulatory reviewing authority on a case-by-case basis using good professional
judgement.
8-4
-------
8.2.3 Dispersion Coefficients
Comment Summary (Averaging Period)
Some commenters stated that the Pasquil1-Gifford (P-G) rural horizontal
dispersion coefficients are based on 10-minute averages and should not be
used to represent hourly averages. To remedy this, some comments suggested
the use of empirical averaging-time conversion factors to produce one hour
averages while others suggested that EPA conduct an additional rural model
evaluation study to determine whether these P-G coefficients should be
increased to represent 60-minute averaging times. One commenter suggested
the use of the Brookhaven dispersion coefficients instead of the P-G coeffi-
cients. Another stated that dispersion coefficients developed from site-
specific studies should be given preferential use over the P-G or any other
non-site specific coefficients. (API, AMC, DS, APCA, SOC, PEPC, IPL)
EPA Response
EPA does not recommend the general use of empirical averaging-time
conversions for periods of less than 1-hour because they are limited to the
data set from which they are derived. Acceptance of such a technique is
provided for on a case-by-case basis. Until better data become available,
the P-G coefficients will continue to be used in their present form. Since
the models are typically used for estimating values toward the extremes in the
distribution, assumptions typifying extremes for the hourly concentrations are
justified. Model evaluation by Turner38 indicates that the second-highest
estimates based on sigmas assumed to represent 1-hour averages, although
having considerable scatter, appear to have little bias. Also, EPA compared
the results of using the P-G coefficients with other alternatives, such as
the Brookhaven dispersion coefficients, on concentration estimates as
documented in Addendum D to Appendix H of the Summary of Comments and
Responses.1 The P-G coefficients performed best. Furthermore, dispersion
coefficients derived from site-specific studies may be used for air quality
impact analyses at these sites if an evaluation demonstrates better performance
in a model than when using an EPA recommended model.
8-5
-------
TEM Is an example of a model that utilizes an empirical averaging-time
conversion factor. However, when evaluated,8 this model did not perform as
well as the EPA models that do not use this factor. This further detracts
from the credibility of this suggested.empirial method of adjusting the
10-minute averages.
Comment Summary (McElroy-Pooler Dispersion Coefficients)
Several commenters stated that the McElroy-Pooler (M-P) dispersion
coefficients are not appropriate for elevated buoyant sources in an urban
environment. According to these commenters, the EPA sponsored RAM model
evaluation is unreliable because the study lacked a good area source
emissions inventory and the air quality monitoring locations were inappro-
priate. Some commenters suggested that EPA conduct an additional urban model
evaluation program with monitors placed within 2 km of major point sources
with tall stacks. Another commenter recommended that EPA await the results
of an Electric Power Research Institute sponsored field experiment for eval-
uating urban plume dispersion planned for 1985 in Indianapolis. Other
commenters suggested that the entire model should be validated and quoted
an EPA statement that an improvement in one component of a model will not
necessarily improve overall model performance. Since the only model for
which any validation data for the M-P curves has been presented in RAM, M-P
dispersion curves should not be substituted into models already validated
with P-G dispersion coefficients unless the performance with M-P coefficients
has been demonstrated to be superior.
One commenter stated that the reference to the M-P coefficients in the
CRSTER and MPTER models are inappropriate since these models should be
confined to rural applications. (APCA, SOC, NYEC, UARG, CONE)
EPA Response
These coefficients were derived from the best scientifically validated
data available. Urban model evaluations39*40 indicate model under-estimates
of extreme concentrations, especially under unstable conditions. In the EPA
model evaluation study, RAM did not consistently show a tendency for over or
underprediction of peak values. The urban dispersion coefficients in the
other EPA models are the same as in RAM. No alternative to the M-P coeffi-
cients has been presented by the commenters.
8-b
-------
EPA also agrees that additional urban model evaluation is desirable
with several close-in monitors as suggested. In the EPA urban model evalua-
tion study, only one monitor was available that was close to a large source.
Pending the availability of high quality monitoring data close in to a
large urban point source, EPA will consider additional model evaluation
studies and encourages others to conduct them.
Comment Summary (Turbulence Intensity)
Some comments stated that dispersion coefficients should be based on
direct measurement of turbulence intensity and suggested that EPA place
high priority on developing guidance on how dispersion coefficients should
be computed from measurements of turbulence intensity. The Mineral Manage-
ment Service's Over Water Dispersion Model has implemented the direct tur-
bulence measurement concepts to characterize over water and land dispersion.
However, one comment said that EPA should retain the position that the
collection and use of such data are optional. (SOC, CHEV, UARG, CMA, APCA,
WC)
EPA Response
The use of dispersion coefficients based on direct measurements of
turbulence may be preferable to the use of discrete stability classes.
However, this involves highly subjective technical procedures and no
consistent methodology has evolved from the scientific community. Thus,
EPA has begun research to develop such a scheme, through its Office of
Research and Development, for relatively flat terrain to estimate dispersion
using horizontal and vertical fluctuation statistics measured or estimated
for the effective height of the plume. Pending completion of this program,
models of this kind must be tested and evaluated before they can be endorsed
for regulatory use. In the interim, the proposed dispersion coefficients
will continue to be used.
8-7
-------
Comment Summary (Buoyancy Induced Dispersion)
According to one comment, the use of Buoyancy-Induced Dispersion (BID)
should be limited to single or multiple point sources with buoyant plume
rise and not to complex sources or multiple source/urban applications where
its use could be considered enhanced dispersion. (NYEC)
EPA Response
BID is only being applied to point sources and EPA guidance is consistent
with the comment.
8.2.4 Stability Categories
Comment Summary (Split Sigmas)
There were several comments suggesting an alternative approach to the
Turner scheme for determining stability classification when on-site
measurements of horizontal and vertical turbulence intensity are available.
Specifically, the "split sigma" approach was recommended to independently
characterize horizontal and vertical stability classes. One commenter
recommended that the refined models should be re-evaluated with this classi-
fication scheme and if the revised models predict more accurately than the
currently preferred models, these revised models should be adopted as the.
preferred models. Several different approaches, however, were presented as
to how to determine these stability categories.
One commenter suggested that the guideline provide recommendations on
characterizing over water stability. (API, AMC, CONE, APCA, SOC, UARG, CARB)
EPA Response
The Turner classification is a widely used scheme because it can be
simply applied to National Weather Service data. There has been no convincing
demonstration that other stability classification schemes allow more accurate
concentration estimates to be made. EPA is presently developing a method
to avoid calculating stability categories altogether and allow for use of
the on-site turbulence data directly in the Gaussian equation. This program
is in the development stage at this time and extensive testing is required
before it can be released to the public.
8-8
-------
There has been no convincing demonstration that split sigmas allow
more accurate concentration estimates to be made as described on page 14 of
the Summary of Comments and Responses.!
Recently the Minerals Management Service^l has released the OCD model
which can treat dispersion over water bodies. EPA plans to propose for
public comment the incorporation of this model in Appendix A of the revised
guideline. (Please refer to responses in Section 7.2.7). EPA will continue
to assess experience gained from model validation studies in this area and
will issue guidance when sufficient experience is gained.
8.2.5 Plume Rise
Comment Summary (Bjorklund and Bowers Algorithm)
EPA's proposed use of the Bjorklund and Bowers (B-B) stack-tip downwash
algorithm received numerous comments that recommended not adopting this
algorithm. Some objected to the use of this algorithm because the study from
which this method was devised is semi-empirical and does not explicitly con-
sider the physics of the downwash process. Even so, some comments agreed
with the use of the B-B method for Froude number less than 1.0. However,
they noted ambiguity in the documentation for Froude number between 1 and 3.
Other comments stated that this algorithm has its greatest impact on plume
size determination for smaller buoyant sources and would cause the model to
calculate no plume rise at all for a source when the wind speed exceeds the
exit velocity. This, however, they noted is contrary to public literature.
Some commenters stated that this algorithm represents a radical departure
from previous EPA guidance on this issue and that this change may invalidate
some of EPA's validation studies of the preferred guideline models. Many
suggested that EPA not change its previously used stack-tip downwash algorithm
until the models incorporating this algorithm have been sufficiently validated.
One stated that EPA's sensitivity study presented in Addendum E does not con-
stitute a true validation study. Another suggested that a sensitivity study
was applied to one model only and not to the other models that EPA is proposing
to incorporate this algorithm. Others noted that EPA has previously suggested
that improving any single model algorithm does not necessarily ensure better
model results. Instead, the entire model as revised must be evaluated in
accordance with the "Interim Procedures" document. The commenter suggested
that EPA should follow its own policy on the proposed (B-B) stack-tip downwash
algorithm change and evaluate all preferred models.
Another commenter further stated that stack-tip downwash should not be
invoked when building wake effects are being simulated with a downwash
model. (API, UARG, APCA, SOC, ISBH, CMA, NYEC, ODEQ, TEGP, MMES).
8-9
-------
EPA Response
EPA has proposed the use of the Bjorklund and Bowers42 stack-tip downwash
algorithm (developed originally by Cramer43 to replace the Briggs44 stack-tip
downwash equation now in use in the EPA air quality models. A number of
responses to this proposal have been received. Many of the commentors
objected to the Bjorklund and Bowers algorithm due both to its semi-empirical
basis and- to the lack of testing which had been done using that equation.
Arguments favoring the use of the Bjorklund and Bowers algorithm are based,
in part, on evidence that the downwash effects on final plume height are
substantially greater than is accounted for by the Briggs equation, and
that tests with the ISC and SHORTZ models show very little bias in the
ground level concentrations.
It is well known that stack-tip downwash occurs when the wind speed
becomes large relative to the stack gas exit velocity. According to Bjorklund
and Bowers,42 their stack-tip downwash algorithm is a semi-empirical correc-
tion to the plume rise which "is based on a combination of visual observations
of plume behavior, the results of wind tunnel studies reported by Briggs44
comparisons of concurrent calculated and measured short-term and long-term
ground level S02 concentrations in Lansing [MI], Allegheny County [PA], and
elsewhere, analysis of the Bringfelt4^ plume rise data, and limited compari-
sons of calculated and measured plume rises for two coal-fired power plants
(Bowers and Cramer, 1976)." It is not entirely clear whether the effects
being treated by these correction formulas are in fact stack-tip downwash
effects, building downwash effects, or both. The Briggs stack-tip downwash
equation currently used in the EPA models does not reduce the calculated
plume height sufficiently to account for the plume height reduction and
higher concentrations noted by Bjorklund and Bowers42 and others.
8-10
-------
It should be noted that stack-tip downwash includes both a lowering of
the height of the plume immediately after it leaves the stack, and a decrease
in the plume rise. The latter results from shear effects and increased plume
size in the turbulence in the lee of the stack. The Briggs equation accounts
for the reduction in initial height only. The Bjorklund and Bowers algorithm
explicitly accounts for the reduction in plume rise. However, most of the
data on which the Bjorklund and Bowers equation was developed and tested
include building downwash effects. The stack height (hs) to building height
(hb) ratios for those data sets are between 1.2 and 2.5. It is not entirely
clear whether the observed increased concentrations are due to stack-tip
downwash or to building downwash. The Huber and Snyder46 downwash correction
implicitly includes stack-tip downwash effects.47 H. E. Cramer Co.43 states
that, for the DOW data, both the Cramer (Bjorklund and Bowers) stack-tip
downwash correction and the Schulman and Scire48 building downwash correction
yield significant and very similar improvements in the correspondence between
calculated and observed concentrations. In the same report the authors state
that "the Cramer,49 stack-tip downwash correction appears to account for the
combined effects [of building downwash and stack-tip downwash] if the stack
height to building height ratio is greater than or equal to about 1.2." The
sample size is, however, very limited.
Thus, the current EPA recommended approach, using the Briggs44 stack-tip
downwash correction and the Huber and Snyder46 building downwash correction,
almost certainly accounts for the combined effects of building and stack-tip
downwash for cases where the hs to hb ratio is less than 2.5. Larger plants
with a hs to hb ratio greater than 2.5 are not treated by the Huber and Snyder
approach. These plants generally have high stack gas exit velocities so that
stack-tip downwash is unlikely to result in significant air quality impacts.
8-11
-------
Smaller plants are the most severely affected by the Bjorklund and Bowers
algorithm; however, little or no testing of the algorithm has been done
with small plants.
Thus, on re-evaluation, we are withdrawing the proposal to recommend
the use of the Bjorklund and Bowers^? stack-tip downwash algorithm in the
preferred models. In doing so, we note the following:
0 For hs to hj., ratios less than 2.5, the Huber and Snyder building
downwash equations treat the stack-tip downwash case implicitly.
0 For the remainder of the cases, the larger sources are unlikely to
have a stack-tip downwash problem, while smaller sources may have
such a problem not addressed by the current EPA approach.
0 The Bjorklund and Bowers42 stack-tip downwash algorithm has not
been adequately tested, if at all, on such smaller sources.
In the interim, EPA continues to recommend the use of Briggs^4 stack-
tip downwash correction for those cases when the use of stack-tip downwash
is appropriate and is considering other methods for future use.
Comment Summary (Plume Penetration)
One commenter recommended that EPA re-evaluate the assumption that
plume rise penetrates the mixing layer, and reformulate its models to take
into account the tendency of the mixing layer to suppress plume rise.
Another commenter recommended that EPA should provide for partial plume
penetration of the elevated stable layer to overcome the tendency to predict
zero concentration for some hours when substantial concentrations are
measured. Another recommended that plumes should have considerable plume
rise (e.g. greater than 15% of the mixing height) before their impacts are
neglected. Less buoyant plumes, the comment added, should carry a fraction
of the plume within the mixing height layer. (NRDC, APCA, SOC, ODEQ)
EPA Response
There have been several formulas reported in the scientific literature
which have addressed the treatment of partial plume penetration. However,
8-12
-------
these formulas have not given better results in model validation studies.
Recently, EPA examined the performance of the PPSP model where partial
penetration of buoyant plumes into the capping inversion is a distinguishing
feature of this model50. Generally the results show that concentrations
estimated by PPSP are almost uniformly higher than the measured SUj data
values. Until such an improved formula is developed, EPA recommends continu-
ing the use of the existing method.
Comment Summary (Gradual Plume Rise)
A couple of comments dealt with whether or not to allow for use of
gradual plume rise in non-complex terrain. One cited the need for this
method in urban areas where there are many close-in elevated receptors
(buildings) and using final plume rise will cause large underpredictions of
the actual impacts on these receptors. Another recommended that the gradual
plume rise option should not be allowed because there is insufficient data
to verify the procedure. (NYCP, CHEV)
EPA Response
There have been no validation studies demonstrating the accuracy of
the gradual plume rise formula at these close-in receptor sites. Small
errors in the gradual plume rise formula could significantly influence
concentrations near plume centerline at these nearby receptors. EPA is not
recommending the general use of gradual plume rise for estimating effective
height because of uncertainty regarding dispersion (plume growth) under
such conditions.
Comment Summary (Plume Rise)
A couple of comments suggested that EPA incorporate plume rise enhance-
ment into the models for industrial source complexes where there are multiple
adjacent stacks. One of the comments suggested using the algorithm proposed
by Briggs while the other suggested that EPA first establish protocols and
then test the revised models for performance relative to currently preferred
models. A couple of commenters suggested that EPA re-examine the stable pi
rise formula used in the RAM Model since this model does not account for
8-13
ume
-------
the urban heat island effect at night. The approach used by urban CRSTER,
urban ISC, and COM is more consistent with research findings on lapse rates
in cities and EPA should modify the RAM model to make it consistent with
the other three. (APCA, SOC, AISI, UARG, CONE)
EPA Response
Although plume rise may be enhanced when there are multiple adjacent
stacks, enhancement of buoyancy due to merging of effluents from these
stacks should consider relations of wind direction and line of stack orienta-
i
tion. Although theories exist to consider this phenomenon, such as the Briggs
technique mentioned in the comment, quality data to justify the specific
procedures included in such a model are not available. EPA agrees with the
second commenter as to the need for evaluating this phenomenon and will
consider the merits of applicable on-site studies on a case-by-case basis.
EPA is investigating stable flow over urban areas. When this reaearch
is completed, EPA may make recommendations on changes to the stable plume
rise formula.
Comment Summary (Building Downwash)
One comment recommended that for treatment of building downwash, the
existing version of ISC (based on Huber's method) should be used for sources
with tall stacks. For sources with short stacks, the building downwash
algorithm in ISC should be replaced with the Schulman and Scire method
which is used in BLP model.
Another comment asked for more research on plume rise from shorter
stacks and unique sources such as flares. (API)
EPA Response
The Industrial Source Complex (ISC) model is being significantly
modified by industry (i.e. American Petroleum Institute) for applications
8-14
-------
involving stacks subject to downwash. Extensive testing is underway prior
to being considered for recommendation by EPA as a preferred model for
these applications. If accepted, this modification will replace existing
algorithms.
EPA agrees that more research on plume rise from unique sources such
as flares is needed and encourages those affected industries to pursue such
studies.
8.2.6 Chemical Transformation
Comment Summary
Several commenters requested a different method of treating SC>2 and 1%
half life than the present transformation scheme of a 4-hour half life for
SOg only. One of these comments recommended the transformation scheme
contained in the MMS OCD model where transformation for both S02 and NOg is
based on latitude, season, and time of day. Another comment asked how
should days with appreciable precipitation be modeled for pollutants like
S02- (SOC, MMES)
EPA Response
The transformation scheme used in the OCD model is based on data obtained
over bodies of water and there has been no convincing demonstration that this
technique is valid for rural or urban applications. EPA will continue to
assess experience gained from model validation studies of the OCD and other
models and will issue guidance when sufficient experience is gained.
The basis for the present transformation scheme is given on pages 29
and 104 of the Summary of Comments and Responses.1 The 4-hour half life has
been used for a long time; a better assumption has not been universally
adopted. However, EPA accepts the use of a different half life on a case-
by-case basis if on-site data are available. There is no recommended half
life for N02 (please refer to Section 6.0 and 6.2.3 of the guideline). EPA
8-15
-------
models do not explicitly consider the effects of precipitation on S02
removal. Any removal is assumed to be included in the decay term.
8.2.7 Gravitational Settling and Deposition
Comment Summary
Two commenters stated that the gradient-transfer model' by Rao for
calculating particle deposition appears to be more physically realistic
than ISO's method and is included in several UNAMAP 5 models (MPTER-DS and
PAL-DS). The comment suggested that ISC model be modified to incorporate
this method or withhold a recommendation of any modeling technique including
ISC for fugitive dust until more research is completed. Another comment said
that based on their theoretical analysis, the settling-deposition algorithm
in the ISC model may overpredict concentrations by a factor of 3.7. The
comment also added that the ISC model fails to account for pit retention
which may introduce an additional overprediction factor. Together, these
cumulative systematic errors may cause a combined overprediction of TSP
concentrations of a factor of 3.8. (CDH, BAAQ, AMC, NCA)
EPA Response
EPA has recommended only the specific particle deposition algorithm in
ISC. The model by Rao, i.e., MPTER-DS, has not yet been sufficiently eval-
uated using ambient data and it is premature to recommend this technique.
Listing a model in UNAMAP does not infer that the model is necessarily
approved for regulatory applications. EPA's evaluation of the ISC model, as
a whole, indicates a net underprediction when the deposition option is used
based on actual data.
EPA is presently sponsoring research that should help determine the
percent retention of particulate matter in surface coal mines.36 once an
algorithm is developed, it must be validated with field data before it can
be recommended for regulatory applications.
8-16
-------
8V2.8 Urban/Rural Classification
Comment Summary
There were a number of comments questioning the merit of the Auer land-
use scheme. One suggested that it should be deleted because it is arbitrary.
Another stated that the Auer scheme was not based on investigations of tur-
bulence or dispersion and that there have been no studies that demonstrate a
direct relationship between Auer's urban land-use classes and observed tur-
bulence or dispersion. Further, the comment added, Auer makes no reference
to a "3km radius circle about the source" or to "50 percent or more" of the
area being in urban land uses for a region to be classified as urban. Another
comment stated that the population density procedure would make suburban areas
and towns urban, when previously these areas were modeled as rural. One
commenter, however, did recommended the use of the Auer scheme.
A number of comments requested more guidance on how to model or classify
areas in between rural and urban. One suggested that sources that are several
kilometers from the heart of the urban area but are still in the overall major
metropolitan area should be modeled as rural sources. For such sources, the
comment recommended using concentrations derived from MPTER or rural RAM
super imposed on the concentrations resulting from separate modeling of the
urban sources. Another comment suggested that EPA develop a hybrid model
reflecting transition between rural and urban dispersion based on plume
location.
As to area classification, one comment suggested that the described
classification procedures would classify urban area sources located next to
large water bodies as rural. Another suggested that there is no good
procedure for urban or rural classification and that this can be best made
by performing an on-site visit and then a rational evaluation of the area
characteristics. (API, SOC, NYEC, APCA, UARG, OPEA, NYEC)
EPA Response
The Auer land-use classification scheme has been documented based on
technical arguments as discussed on page 28 of the Summary of Comments,and
Responses.1 EPA welcomes research results from field studies that more reliably
identify urban boundaries. In the interim, EPA, sees no basis for change in
guidance at this time because no viable alternative has been proposed.
EPA is sympathetic to the commenters assertion that the population
density procedure may mis-classify suburban areas. That is why the Auer land
use scheme is recommended as the first method of choice while the population
density is the second method of choice.
8-17
-------
As to the commenters request for guidance on how to classify areas
between rural and urban, EPA recommends that the whole urban area should be
considered as one entity. Otherwise, the use of such procedures as the
commenter has suggested would result in a complicated and arbitrary analysis.
EPA presently has no plans to develop a hybrid model as described by the
commenter, but will assist in examining technical accuracy and regulatory
applicability if such a model is submitted by developers. EPA's research on
estimating dispersion directly from fluctuation statistics should overcome
artificial urban-rural differences.
As to area classification, some subjectivity is always present in simple
rules of thumb. Therefore, once an area is identified as being potentially
within an urban boundary, EPA recommends early discussion between permit
granting authority and the applicant, including an on-site visit if possible
to resolve many issues including final area classification.
8.2.9 Fumigation
Comment Summary
Two comments requested improved guidance on inversion break-up fumigation
because of the importance of this phenomenon for certain model applications.
(ADHS, MMES)
EPA Response
EPA agrees that models dealing with plume fumigation phenomena are
needed. However, no method has been convincingly demonstrated with monitored
data. In the interim, other techniques such as the manual calculation method
shown in the Turner Workbook may be used.
8-18
-------
8.2.11 Calibration of Models
Comment Summary
Two comments recommended the use of calibration factors in situations
where all modeling techniques significantly underpredict or when nearby
ambient air quality data are available. (CITG)
EPA Response
Calibration for short-term air quality concentrations is not recommended
for various reasons as outlined on page 71 of the Summary of Comments and
Responses.^
8-19
-------
9.0 MODEL INPUT DATA
9.1 Source Data
Comment Summary (Change Definition of Emission Rates)
Several comments said that the use of maximum emission rates for every
hour throughout the year is unrealistic and overestimates air quality impacts.
This conservative assumption is further compounded by the probability of
occurrence of a specific wind speed, wind direction and stability class. As
an alternative, one suggested that EPA should model proposed new sources at
maximum emissions, while using actual emissions for other nearby sources.
Two suggested that actual hourly operating conditions should be used instead
of design capacity. Another suggested that the maximum operating rates should
be used only when a more representative rate cannot be defined. Otherwise,
the highest historical (e.g., three year) operating rate for the averaging
time being evaluated should be used. For example, the impact on a 24-hour
standard would be modeled using a 24-hour average emission rate. Another
recommended that EPA revert to the practice of identifying a single critial
load condition in a screening analysis and then using the critical load
(rather than both partial and maximum load) in refined modeling.
A few comments stated that, for multi-source applications when several
plants in a system are being modeled, the modeling should recognize any
system-wide limitations on load. The probability of a combination of 100%
load at each source (and 50 to 75% constant maximum loads) with worst case
meteorological conditions is mathematically almost impossible. One recom-
mended using hourly load data consistent with the meteorological record or
selecting an appropriate seasonal capacity.
One comment said that the description of long-term emissions (on page
5-4, paragraph 3) does not indicate the type of modeling study to which it
relates. For instance, the PSD regulation describes the "actual emissions"
as the average rate, in tons per year, at which the unit actually emitted the
pollutant during a two year period which precedes the particular date and
which is representative of normal source operation. The text in the guide-
line refers to maximum emissons based on 3 years.
One comment said that guidance should be provided to define the phrase
"future time period" during which growth of emissions should be considered.
This phrase appeared on page 9-4 of the guideline. (API, TVA, UARG, SOC, CMA,
MCC, OEPA, CONE).
EPA Response
According to 40 CFR 51.22, EPA is required to adopt "emission limitations
and other measures necessary for attainment and maintenance of any national
standard." To achieve this, stationary source control strategies for State
implementation plans (SIP's) must be determined using the maximum emission
9-1
-------
rate allowed under the federally enforceable permit. The actual emission
rate may be used only if it is federally enforceable. This requirement
applies to the source(s) subject to the SIP emission limit evaluation,
nearby sources, and other sources that contribute to the background concen-
tration of sulfur dioxide (SO?), To ensure attainment and- maintenance of
the ambient standards, as provided for in Section 110 of the Clean Air Act,
actual or design capacity {whichever is greater) should be used to simulate
operating conditions of the source(s) subject to evaluation and nearby
sources. The possible interacting impact at the same receptor is thus
accounted for. Other operating conditions may be used only if they are
federally enforceable permit conditions. Load conditions used as model
input should ensure maintenance of the ambient standard during all operating
conditions. For other than nearby sources that contribute to the background
concentration of $03, annual levels determined when the source actually
operates, averaged over the most recent 2 years may be used (See Comments
on Section 9.2).
For the prevention of significant deterioration of air quality, Section
163 of the Act requires that "each applicable implementation plan shall
contain measures assuring that . . . maximum allowable concentrations of
[sulfur oxide] . . . shall not be exceeded." For this reason, dispersion
model results should be based on the operating and meteorological conditions
that cause the highest ground-level concentrations of $03. Because the
model input data must represent worst-case conditions, no system-wide
limitations on load can be recognized.
The guideline text on evaluating SIP's for compliance with long term
ambient standards has been revised and a table added which incorporates the
following: Annual and quarterly emission limits should be tested using
9-2
-------
the maximum al lowab'le emission limit or other federally enforceable permit
limit. For source(s) under evaluation and nearby sources, the operating
conditions used should represent actual or design capacity (whichever is
greater) or another federally enforceable permit condition averaged over the
most recent 2 years. For other sources, actual annual operating levels
averaged over the most recent 2 years should be used.
The future time period during which the impact of growth on emissions
should be considered is the period during which known or anticipated growth
is expected to occur. Any new source construction or existing source
reconstruction or modification that has been proposed for the area being
evaluated and that has the potential to affect emissions in that area
should be included. In area-wide analyses, the data by which' the standard
must be attained defines the future time period.
Comment Summary (Use of Statistical Emission Rates)
A few commenters indicated that they disagreed with the use of fixed
maximum emission rates and that the use of statistically based methods and
variable emission rates should be encouraged. One suggested that EPA should
adopt the ExEx method or allow new sources in multiple source areas to model
their emissions at maximum capacity and the existing sources at historical
capacity usage. While the latter is not as rigorous as the former (probabi-
listic) approach, it does provide a more probable description of an emission
scenario than current practice. Another suggested the use of average
emission rates or some statistical approach which would account for emission
rate variations. Use of statistical probabilities in air quality analysis
is gaining some acceptance as demonstrated by a recent court decision (Kamp
vs Hernandez). One commenter recommended a statistical approach in dealing
with mining emissions because a mining operation is a complex array of
sources, some of which are simultaneous, and some of which are not and the
peak activity level may not occur simultaneously with the worst meteorological
conditions. (API, SOC, CHEV, AMC)
EPA Response
The EPA is currently investigating the use of statistically based
methods and variable emission rates. Thus far, no such approach has been
approved for use in mathematical air quality modeling sources to determine
9-3
-------
compliance with ambient air quality standards because statistical modeling
methods do not provide a mechanism for evaluating the sources against short
term ambient standards. Also, the deterministic nature of the models
presently used to test emission limits is consistent with the deterministic
nature of the standard. Currently, compliance with ambient standards should
be-determined using maximum allowable emission rates at actual operating
levels or design capacity (whichever is greater) to assure attainment and
maintenance of the ambient standard.
The use of statistical probabilities in air quality analysis is addressed
in the response to comments in Chapter 10 dealing with "Model Uncertainty."
9.2 Background Concentrations
Comment Summary (Single Sources)
Several comments asked EPA for more guidance on determining background
concentrations for isolated single sources. One comment agreed with the
Option One technique of determining background concentrations (described in
the guideline), but usually the meteorological data needed to carry it out
are not available. Also, in attempting to use this technique, users have
experienced difficulty in characterizing the "meteorological conditions of
concern" in a way that allows monitoring days to be selected for averaging.
The comment recommended that in the absence of sufficient data to employ
Option One, the annual mean concentration at the selected background monitor-
ing site (or the average annual mean concentration over the background
monitoring network) be used as an estimate of the background concentration
for all averaging periods. Another commenter said that the second option is
complicated and experience has produced situations where background concen-
trations calculated by this method are actually higher than the measured
impact at a site. The comment suggested that EPA consider, as the background,
the minimum concentration reported at any monitor within the regional monitor-
ing network. Or, EPA can adopt one of the following methods: First, EPA can
specify that if the background exceeds the measured concentration that
results from using the EPA-recommended calculational method, the background
should be set at less than or equal to the measured concentration. Another
option would be not to factor into any modeling analysis those measured
concentrations in the non-impacted sector which are unusually high and are
most probably due to local conditions. The final option would be to set
background at a given monitor in an amount not to exceed the average concen-
tration at that monitor when the source is impacting the same monitor.
Another comment stated that Option One may not be viable in complex
terrain, or complicated flow patterns and where there is indirect transport
9-4
-------
of pollutants. One comment stated that Option Two should explicitly state
how to determine background concentrations for various averaging times as
in Option One. (FDER, UARG, MCC, DPC)
EPA Response
Guidance on determining background concentrations is by necessity made
general because of the complexity and variety of modeling situations. This
flexibility allows for a case-by-case determination in consultation with
the Regional Office. Therefore, only a brief response to the above comments
can be given.
The meteorological conditions of concern are those conditions that
result in determining the highest, second-highest concentration (i.e.
neutral stability and 4-6 meters per second, or stable conditions with 2-3
meters per second, etc.). During these conditions, background is determined
and then averaged over the period of record. When sufficient data are not
available to employ this method, as the commenter asks, then a case-by-case
determination in consultation with the Regional Office is needed. The other
two options presented by the commenter can not be made into general guidance
because they employ arbitrary criteria. Again, EPA recommends a case-by-case
analysis of the problem.
In complex terrain, it must first be determined that the data available
are representative of the site. If data are not applicable, as the commenter
suggests, then the entire modeling analysis, not just background determination
must be reviewed with the Regional Office. As to the second comment, averaging
times for Option Two should be the same as those in Option One.
Comment Summary (Multi-Sources)
Several comments also asked EPA for more guidance in determining
background concentrations in multiple source areas. Two comments stated
that the proposed procedures are complex and unworkable in dense urban
areas with many hundreds or thousands of relatively small sources and
9-5
-------
recommended tnat the procedures for treating these sources be determined on
a case-by-case basis. Two comments recommended: (a) characterize the
meteorology of the periods of predicted highest combined impact from the
sources being modeled; (b) identify similar periods from the most recent
representative period of air quality monitoring and model these periods
using coincident hourly emissions and meteorological data; (c) subtract the
explicit coincident impact; (d) the resulting short term concentrations are
ranked and the highest of these is used as the representative background
value.
Another comment stated that for modeling nearoy sources, tne text
should be modified and made consistent with Section 9.2.1. For evaluation
against annual standards the term "worst case emissions" should be replaced
by "maximum historical emissions during the last 3 years" as stated on page
9-4. Sources permanently shut down should not be included. For evaluation
against short term standards, it is unrealistic to model all sources at
maximum allowable emissions. Only a realistic worst case emission scenario
should be required. (APCA, UARG, CONE)
EPA Response
Guidance on determining background concentrations in multiple source
areas is as explicit as can be stated. In urban areas where there are
numerous sources and source categories, a case-by-case determination in
consultation with the Regional Office is needed. It is unclear how the
approach (a through d) recommended by the commenters fits in with the
guidance i.e., whether it is a replacement or supplemental or on a case-by-
case basis. Nor was any proof presented that this approach was any better
than that given in the guideline.
The text in Section 9.2.3 has been revised to maintain consistency
with previous sections, and a new Table 9-1 has been added to further
clarify the data input requirements for nearby and other background sources.
Nearby sources should be explicitly modeled only when they are "expected to
cause a significant concentration gradient" and when they are "few in number.1
The resulting modeled concentrations should be used in concert with ambient
monitoring data to determine concentrations of S02 due to all background
9-6
-------
sources in the vicinity of the source or sources being evaluated. For
evaluation against annual standards, as the commenter suggests, nearby
sources should be modeled using maximum allowable emissions at actual
operating conditions averaged over the most recent 2 years. Short term
emission limits should be evaluated using maximum allowable emissions at
actual operating conditions for all hours of each time period under
consideration. If operation does not occur for all hours of the time period
and the source operation is constrained by a federally enforceable permit
condition, an appropriate adjustment to the modeled emission rate may be
made. EPA agrees that sources permanently shutdown should not be included.
Comment Summary. (Miscellaneous)
One commenter recommended that to ensure that a health-related NAAQS
will not be exceeded, the proposed method of using the average background
concentrations that occur during meteorological conditions of concern should
be replaced by using the highest background concentration that occurred
during meteorological conditions of concern.
Another commenter said that receptor modeling should be included as a
method for determining background concentrations. (CDH, ADHS)
EPA Response
The commenter's suggestion to use the highest background concentrations
would result in a very conservative estimate of a source's impact. The
commenter did not provide any analysis to support this position. EPA
believes that its approach is rational and is better suited for practical
applications.
Receptor models may be used to determine background concentrations on
a case-by-case basis. Please refer to comment responses in Section 11.1 on
"Receptor Models."
9-7
-------
9.3 Meteorological Input Data
9.3.1 Length of Record of Meteorological Data
Comment Summary (Use Less Than Five Years)
Many commented that the requirement for five years of NWS meteorological
data- seems excessive. One said that three years of CRSTER modeling is nearly
as effective in identifying peak concentrations as five years and the loss
of accuracy is insignificant compared to uncertainty of the model calculation
itself. For example, the ISC model is very expensive to run and furthermore,
downwash conditions would occur so frequently that one is certain to find
maximum concentrations in three years of modeling that are virtually as
large as maximum values obtained using five years of data. Another urged
EPA to devise screening procedures to identify the critical meteorological
conditions which produce maximum concentrations. Procedures should be
promulgated to select just those cases with potential for producing maximum
concentrations. Another stated that when three years of data are available,
the highest second highest concentration may be used. However, another
commenter said that the use of five years of data in which to pick a maximum
level results in an effective standard of not to exceed more than once in
five years and recommended the use of two years of NWS data instead.
One commenter recommended that if at least one year of quality assurable
data are available, the guideline should require its use. The source's
option of using the most beneficial result of either on-site or NWS data
should be eliminated. However, another said that the modeler should have
the discretion to select which year or years of data are most appropriate
for the application.
One of the commenters said that the language of all portions of the
guideline should be consistent in requiring that not more than five years
of data be used in any modeling analysis using off-site data and not more
than one year of data when on-site data are used. However, another commented
that one year of on-site data should be demonstrated to be representative
of the worst short term impacts over a five year period.
Another commenter suggested that the guideline allow flexibility for
use of older NWS data when the use of hourly data are appropriate. (TACB,
MES, AISI, UARG, NYCP, MSUS, CITG, CARB, WDNR)
EPA Response
The length of record should be adequate for EPA to determine the
adequacy of the emission limitations. EPA has presented its analysis and
rationale for the use of five years of NWS or one year of on-site data on pages
45-50 of the Summary of Comments and ResponsesJ Results from recent investi*
9-8
-------
gations support that conclusion.51 The commenters have not presented any
data to support their claim that three years (or two years) of data are as
climatologically representative and sufficient to protect the NAAQS as
five years of data.
The cost of modeling can not be used as the sole reason for using less
than five years of data; the deterministic nature of the standards require
EPA to make certain that the standards are never violated (please refer to
response to comments in Section 9.1 on "Use of Statistical Emission Rates";
and "Use of Best Estimates" in Chapter 10). Furthermore, the requirements
for consistency (refer to comment responses on page 1-1) do not allow EPA to
accept results from one model (e.g. ISC) with less data while more data are
required for input into another model. The use of screening procedures as
a first level conservative estimate is recommended in the guideline. However,
the use of screening procedures to reduce the number of hours to be modeled,
as the commenter suggests, is not recommended because there is a great deal
of uncertainty in arriving at such a screening method, especially when more
than one source is being modeled.
EPA agrees with the commenter and has changed its guidance to recommend
that if at least one year of quality assured on-site data are available,
they are preferable to NWS data and should be used.
EPA is consistent in stating that at least five years of NWS or one year
of on-site data is required in any modeling analysis. Up to five years of
on-site data should be used if available and valid. EPA has examined the
issue of determining representativeness of one year of on-site data to five
years of NWS data and has concluded that a scientifically available method
does not exist at this time. Considering the cost of acquiring quality-
9-9
-------
assured on-site data and the lead time required for this effort, it is not
feasible to recommend a longer period of record.
As to the age of the NWS data record, EPA recommends that the most
recent, readily available and consecutive five year period of record should be
used, provided that an earlier period has not been identified already by a
permit granting authority to contain the controlling period (i.e., has the
worst case impact).
Comment Summary (Differentiate by Size)
A few commenters suggested that the decision of what record length to
choose should be related to the plant size. One of these comments said
that it was unclear whether to use five years of meteorological data in all
cases or only in the case of a large source, with consideration given to
the uncertainties involved in cases when less than five years of data are
used. If the exact location of the controlling concentrations is not
important, one year of data may be used when the predicted concentration is
less than one-half of the standard. Another commenter said that requiring
the use of five years of data is a waste of resources for the majority of
sources which are small and consume only a fraction of an applicable PSD
increment or NAAQS. Another said that for large sources (e.g., 500 MW
power plant), one year of on-site data is preferable but should not be
required for small sources in uncomplicated terrain where the nearest NWS
station may adequately represent local conditions. (FDER, ADEM, MMES, OEPA)
EPA Response
EPA recommends that (1) five years of data should be used (no limitation
on source size) and (2) for large sources, e.g., a 500 MW plant, five years of
NWS data or at least one year of site-specific data are required. This
implies that the use of five years of NWS data is the norm, but on occasion an
exception for a small source might be justified. EPA rationale has been
previously stated on pages 45-50 of the Summary of Comments and Responses.!
If a source cannot demonstrate compliance with a screening model, the five
years of NWS data should be used in a refined model analysis. The commenter
has not provided any data to explain how cl imatological variability is
9-10
-------
addressed if less than five years of NWS data are used. For the majority of
small sources that the other commenter refers to, the question of length of
meteorological record never arises because these sources will most likely
pass the screening test and further refined analysis would not be necessary.
If the screening analysis indicated the possibility of a violation, then a
full analysis using five years of NWS or one year of site-specific data is
necessary.
9.3.2 National Weather Service Data
Comment Summary
One commenter recommended that the guideline discuss how mixing depth
data are obtained from NWS upper air stations.
A few comments noted that the NWS wind measurements are averages for a
few minutes estimated by the observer and that fluctuations during the hour
are not reflected by those short observation times. The guideline seems to
prefer NWS data to on-site data, which are better. EPA should provide infor-
mation on the sensitivity of modeling results to the use of NWS data versus
on-site data. A couple of other comments recommended that meteorological
data from other sources (companies, universities, FAA and military stations)
should not be excluded if they are equivalent in accuracy and detail to NWS
data, particularly if these sources are closer than the nearest NWS site.
One commenter urged EPA to allow the continued use of off-site NWS data,
at least until the numerous difficulties are resolved concerning the use of
on-site data, (e.g., until turbulence data can be input directly into the
model). (APCA, SOC, WCHD, ADHS, UARG, MCC)
EPA Response
A discussion of how mixing height is determined is available in the
documents cited in the guideline and the report by Holzworth.52
EPA recognizes the expense involved in collecting and processing site-
specific data and has allowed the use of NWS.data which are much less
costly. The technical community has not performed a sensitivity analysis
to determine model differences when using NWS versus on-site data and none
has been presented for review by EPA. Site-specific data are preferred and
9-11
-------
should be used when available. Equivalent meteorological data sets such as
those from Department of Defense or FAA, if appropriate, may be used in
lieu of NWS data and text will be changed to reflect this comment response.
EPA does not agree with the commenter's statement that there are
unresolved difficulties concerning use of on-site data. EPA is continuing
to develop techniques to improve the method of input of on-site data into
the EPA models (e.g., the use of turbulence parameters directly into the
model) and will issue guidance when these techniques have been adequately
tested.
9.3.3 Site Specific Data
Comment Summary (Alternate Stability Classes and Sampling Rates)
A couple of commenters recommended that the guideline indicate that
alternative stability categorization schemes can be used on a case-by-case
basis. Stability determination made with turbulence intensity are not com-
parable to, nor should be judged by, the subjective P-G-T method.
A few commenters stated that the requirement of a sampling frequency of
once per second (3600 per hour) presents significant memory requirements
for monitoring stations that store many parameters since most data acquisition
systems do not allow different sampling frequencies for different parameters.
Not only is this requirement burdensome, but also it is unnecessary. Studies
have shown that the sampling interval could be increased to up to 10 seconds
without sacrificing more than one percent in the accuracy of sigma E calcula-
tions, (APCA, SOC, UAR6)
EPA Response
EPA's research program has not yet resulted in recommending alternative
stability categorization schemes such as classifying atmospheric dispersion
directly from wind turbulence measurements, nor were such criteria recommended
by the commenters. Until uniform criteria are developed and tested in a
dispersion model, EPA preferred models continue to use the existing methods.
However, the use of alternative stability categorization schemes in dispersion
modeling is acceptable subject to an evaluation on a case-by-case basis.
9-12
-------
Criteria for such an evaluation are given in the Interim Procedures {please
refer to comment responses in Section 3 dealing with "Implementation of New
Models").
In response to the comments on sampling frequency, EPA is changing its
guidance on data sampling frequency and averaging method and will modify
the text accordingly. EPA recommends that the hourly standard deviation of
the horizontal wind direction (sigma-A) be based on four 15-minute averages
SO that the effect of plume meander can be accounted for. However, 360
samples are needed during each 15-minute period. According to the EPA
workshop,53 "three hundred and sixty or more samples will estimate the hourly
standard deviation within 5-10%." This is consistent with the results of
other researchers. For the standards deviation of the vertical wind direction
(sigma-E), EPA continues to recommend an hourly average with at least 360
samples; a higher frequency of sampling is encouraged.
Comment Summary (Other Measurement Methods)
One commenter suggested the use of doppler acoustic soundings as an
acceptable method for determining wind parameters at stack height because
the technology is proven and offers substantial cost and safety advantages
over tall towers. Another recommended the use of net radiometer instruments
for stability class determination. These instruments are reliable, correlate
well with diural cycles and cloud cover and would best correlate with the
Turner insolation method. One commented that guidance for instrument measure-
ments for stability recommended for urban areas is lacking.
One commenter recommended that all atmospheric dispersion model estimates
should be corrected to standard temperature and pressure (STP) because the
ambient air quality standards are given in terms of STP. It is a mistake
not to recommend this correction for STP because, depending on elevation,
not correcting modeled concentrations to STP would allow the ambient standards
to be exceeded by between 10 and 30 percent. (CDH, UARG, NYCD)
EPA Response
Based on an evaluation of wind measurements by doppler sodar,54 £PA
accepts the use of doppler sodar (subject to quality assurance requirements)
9-13
-------
to measure average wind speed and direction but not for determining sigma-A
or sigma-E values. The use of net radiometer measurements a]one for determin-
ing stability class is insufficiently tested; thus, it is premature to
recommend the use of this method until further experience is gained. However,
the use of these and other new methods of on-site meteorological measurements
will be considered on a case-by-case basis. Guidance for determining atmos-
pheric stability in urban areas is the same as that in rural areas.
Given the manner in which the $02 ambient air quality standards are
written, and to be consistent with the handling of monitoring data, some
States have suggested that model estimates should be corrected to STP.
From a historical standpoint, EPA has not recommended correcting model
estimates to STP. However, if a State wishes to make the STP correction,
EPA has no objections, since the State has the right to be more stringent
than EPA.
Comment Summary (Stability Classification Scheme)
Several comments questioned the order of preferrence for the stability
classification. One said that sigma-E measurements should not be placed
above sigma-A as a means of calculating stability class given the difficulty
and uncertainty involved in measuring sigma-E. Wind direction meander should
not be presented as a significant problem with sigma-A measurements. Another
recommended that all three sigmas be measured and used to independently
classify both vertical and horizontal dispersion.
Another objected to using the sigma-E method of determining stability
categories because the stability class limits were not selected on the
basis of any comparison between sigma-E values and Turner stability classes.
Sigma-E values as recommended by EPA will often indicate category A stability
conditions, but visual observations of power plant plumes during such hours
reveal no tendency for behavior characteristics of category A conditions.
EPA should not recommend this technique for determining stability classifi-
cation until further comparisons and evaluations are conducted.
One recommended that the preferred method of determining stability
category should be changed in the order of site specific measurements used
(e.g. change the order on page 9-17 so that number 3 is first, followed by
2, 1 and 4). Stability determined from on-site data should never be smoothed
because this negates one of the main purposes of collecting on-site data
9-14
-------
which is to measure what actually occurs at the site. If those data are
used for some other area, it is "off-site" for that application, and thus
should be smoothed for that application. However, another comment recommended
that stability data derived from on-site measurements should be smoothed in
a manner consistent with the CRSTER preprocessor so that the stability class
from hour to hour changes no more abruptly than by one class. Still another
recommended a different order of preference than that given on page 9-17
(i.e. change the order so that number 1 is followed by 4, 2 and 3). On-Site
information on cloud cover, etc. is almost never available and this in itself
should not force the use of on-site data with nearby NWS data last in the
order of preference since the latter method will give better representation
of the categories than the. sigma-A or sigma-E methods. The sigma-A method
is preferred because of familiarity, past use and improvements in monitoring.
One stated that the EPA recommendation of a temporary program of
stability class spot-checks of on-site data is futile since there is little
correspondence between the various methods except possibly in the overall
frequency distribution of the classes.
Another comment asked that since Pasquill defined nighttime as the
period from one hour before sunset to one hour after dawn, but Turner's
algorithm defines nighttime as being between sunset and sunrise, which
definition is appropriate. Also, since Pasquill based his work on the
neutral hour (D) preceding and following nighttime while Turner classifies
these hours as C, which procedure should be followed. (CDH, UARG, SOC,
NCNR, NYEC, UOF)
EPA Response
Although both sigma-E and sigma-A are of importance in estimating the
impact of point sources, sigma-E determines how near the source an elevated
plume will have maximum impact and consequently the magnitude of that
impact. To make estimates with the current models (which derive dispersion
parameter values from a stability class), EPA gives emphasis to best estimating
sigma-E rather than sigma-A. Therefore, methods closely related to vertical
dispersion, i.e. Pasquill stability classification and vertical fluctuation
measurements, are preferred over methods more closely related to horizontal
dispersion, i.e. horizontal fluctuations. EPA does not present wind direct-
ion meander as a significant problem with sigma-A measurements; only that
wind direction meander must be considered in the calculation. EPA does not
9-15
-------
object to the measurements of all three sigmas on a case-by-case basis. How-
ever, the use of these parameters for a split-sigma approach is not recommended,
A preferred method for determining atmospheric stability is shown on new
pages 9-21, 22 of the guideline.
There is no need to compare statiblity class limi.ts between sigma-E
values and Turner stability classes. EPA prefers classifying stability
based on Turner's method using site-specific data. If those data are unavail-
able, then EPA prefers three other schemes shown, in the order of preference,
in the guideline. With respect to correlating the stability categories
based on sigma-E with visual observations of power plant plumes, the comment
neglected to mention whether or not these parameters were extrapolated to
the same height. The conclusion reached by the comment may have resulted
because of the change in parameters with height.
EPA is not planning to change the preferrence order of determining
stability category in the guideline for the reasons stated above. Also
seen in the comments, the commenters are divided in their recommendations
on a preferred method. With respect to the smoothing of hour-to-hour
variations in stability as measured by sigma-A, EPA accepts smoothing of
these variations so that the stability class from hour-to-hour changes no
more abruptly than one class. This approach is comparable to that used for
•
determining P-G stability from cloud cover data. When measured sigmas are
used as direct turbulence inputs to models (and stability class is not
calculated), the actual hour-to-hour variations in turbulence should be
used. Commenters1 recommendations are varied; it appears that there is no
consistent recommendation for characterizing stability categories. EPA's
rationale is still valid; the Pasquil 1-Gifford method is preferred, and
9-16
-------
methods based on on-site measurements are more preferred than those based
on off-site data.
EPA's research program on dispersion coefficients is designed to
eliminate the need for determining stability classification entirely and
instead use on-site turublence da'ta directly in the dispersion model.
Thus, many of the difficulties noted by these comments will be overcome as
future models become operational.
The concept of a spot check is only a suggestion to ensure that data
collected are of highest quality. The text in the guideline has been
modified. As noted in the Workbook of Atmospheric Dispersion Estimates,
EPA continues to apply the objective system of classifying stability from
hourly meteorological observations based on the method published by Turner.55
EPA recommends that the stability class does not change by more than one
class during each consecutive hour and therefore, the commenter's second
argument is not recommended.
Comment Summary (Wind Direction Meander)
Several commenters said that it is inappropriate to eliminate the wind
direction meander contribution to sigma-A. Such wind direction meander
contributes to the effective horizontal dispersion of the plume but does not
bear any relationship to the vertical dispersion. Use of split sigmas would
eliminate this concern. Meander can be overcome by the use of micropro-
cessors with sufficiently small sampling time, and running 5-minute averages
summed and averaged over a one-hour period. However, another comment said
that the standard averaging time for the determination of hourly average
sigmas should be 15-minutes regardless of meander because these sigma ranges
are based on experiments averaging 10 to 15-minutes. (APCA, SOC, NYEC, CONE)
EPA Response
Estimation of stability class from hourly sigma-A without eliminating
meander will result in overestimation of the frequency of unstable stability
classes especially classes A and B, resulting in increased frequencies of
9-17
-------
estimated large ground-level impacts from elevated sources. There is no
long term meander in the vertical. That is why EPA is justified in recommend-
ing this technique. EPA can not accept the recommendation to use split
sigmas (please refer to response to comments in Section 8.2.4 dealing with
"Split Sigma.") EPA recommends the use of four 15-min.ute averages to get an
hourly average. The commenter did not supply any data to justify the use
of a 5-minute averaging time to obtain an hourly average. The last comment
to use only one 15-minute average is not practical because the models compute
hourly averages. The issue of averaging times for sigma ranges was addressed
in the response to comments in Section 8.2.3 dealing with "Averaging Period."
Comment Summary (Miscellaneous)
One commenter suggested that EPA clarify that the limitations on the
possible stability categories in the footnotes of Tables 9-2 and 9-3 are
necessary in order to ensure conformance of the sigma-method stability
classes to the P-G classes. However, the occurrence of stable conditions
during daytime hours is not uncommon in real situations such as in complex
terrain.
Another commenter said that the corrections to the Table 9-2 and 9-3
ranges for surface roughness considerations should not be based on the
average roughness length. For situations with large roughness differences
surrounding the site (e.g., river and forests) the use of average roughness
can lead to erroneous stability cases. Thus the roughness correction should
be made wind direction sector dependent. Furthermore, the roughness length
should be determined within 1 km of the instrument tower and not within 1
to 3 km as recommended by EPA.
One commenter said that the sigma-A and sigma-E stability classification
schemes, as presented in Table 9-2 and 9-3 are inconsistent with respect to
D stability during the day. If sigma-A is used and wind speed exceeds 6 m/s,
then stability class D should be assumed. However, sigma-E has no dependence
on wind speed, so if high wind speeds are observed during the day then stabiTl"
class D will not automatically be assumed. The comment recommended that Table
9-2 for sigma-A be more clearly presented. (NYEC, MMES)
EPA Response
There are limitations to any method for deriving a stability class.
The occurrence of stable conditions during daytime hours is possible during
9-18
-------
some situations. However, until another method is shown to be superior,
EPA contiues to recommend the Pasquill-Gifford-Turner method for categorizing
atmospheric stability which has been used for the last 15 years.
EPA has recommended the use of surface roughness corrections because of
previous public comment (please refer to pages 50-55 of the Summary of Comments
and Responses J) The. procedure for determining the roughness correction for
a given site is general and should be applied with care in situations where
there are large roughness differences surrounding the site similar to those
noted by the commenter. There is usually no need to make roughness correction
dependent on the wind direction sector. However, such a determination may
be made on a case-by-case basis. Determining roughness length within 1 km
of the tower, as the commenter suggested, is unjustified since this distance
is too short to allow the roughness modification to atmospheric flow from
reaching a steady-state condition. A distance of 3 km is more appropriate
(See reference 59 in the revised guideline).
As to the inconsistency in the tables, the appropriate footnote has
been revised to state that during the daytime, conditions are neutral for
10-m wind speed equal to or greater than 6 m/s.
9.3.4 Treatment of Calms
Comment Summary
One commenter suggested that EPA modify CALMPRO to make its final report
easier to read when predicted concentrations from ISC are included. The
final report should include the source group number and receptor location.
Another suggested that software be modified to a cut-off speed less than or
equal to 1 m/s to allow easy automation of the entire procedure. A couple
of commenters asked EPA to clarify how wind direction Is treated. One asked
why the assignment of wind direction is treated differently between on-site
and NWS data! Another said that guidance should require a case-by-case
evaluation any time wind speeds are less than 3 m/s. Another agreed with
EPA's method when NWS data are used.
A couple of commenters said that many periods of high observed air
quality exceedances are observed during calm winds and disregarding calm
9-19
-------
winds during these periods is unrealistic. Another said that if the total
number of noncalm hours is less than 18 for 24-hour averages, less than six
for 8-hour averages, or less than three for 3-hour averages, concentrations
determined for these periods should be disregarded. Another said that
eliminating calm hours from consideration could lead to inadequate data
recovery for some NWS data and that a better scheme is needed. EPA should
allow the use of alternative procedures to determine hourly averages, such
as subdividing the hour into shorter averaging times, if data are available.
*i
One commenter said that a definition of indeterminate should be included
in the guideline (e.g., the wind direction should be defined indeterminate
when the hourly average wind speed is below the response threshold of the
wind vane). Another asked for more specific guidance for handling stagnation
episodes rather than on a case-by-case basis. (AISI, WDNR, FDER, UARG, UOF,
ADHS, APCA, NYEC, CDH).
EPA Response
The procedures which make up the CALMPRO processor for handling calms
are being incorporated into the recommended EPA preferred models MPTER,
CRSTER, RAM, and ISCST. As the commenter suggests, the calm treatment, as
implemented in ISC, will be applied to the source and source group contri-
butions as well as the total concentration. The various tables will be
adequately labeled. Each of the four models is normally run using prepro-
cessed (by RAMMET) NWS data. Calms are not explicitly identified in the
RAMMET preprocessed data, since all calm cases are assigned, for modeling
purposes, a speed of 1.00 m/s and a direction identical to that of the
previous hour. The calms treatment takes advantage of this procedure to
identify those hours which were originally calm. Specifically, any hour
with a wind speed of exactly 1.00 m/s and wind direction identical to the
previous hour is treated as calm. This procedure is not appropriate for on-
site data. For these data, observations less than 1 m/s should be input as
1 m/s and the corresponding wind direction is used. EPA does not agree
with the commenters statement that a case-by-case determination is needed
whenever wind speed is less than 3 m/s. The definition of calm winds is
9-20
-------
related to the instrument threshold and most present day instrumentation
has a threshold of 1 m/s.
Gaussian models are not valid for calm conditions. Even though it may
be physically unrealistic to disregard calm winds, as the commenter suggests,
it is the only option available since there is no meaningful way to account
for this condition. Until a satisfactory method is provided in the scientific
literature, EPA will continue with the practice of disregarding model concen-
trations when wind speeds are calm. EPA does not agree with the commenter
who suggests deleting entire periods when less than the desirable amount of
data exists. Data should be used whenever available. The commenter did not
present any evidence to prove that the EPA method is incorrect. EPA agrees
with the commenter that eliminating too much data from the record is undesire-
able, however, all valid data should be used. The commenters suggestion to
subdivide the hour into shorter averaging times, when such data are available,
may be acceptable on a case-by-case basis.
EPA agrees to incorporate the definition of indeterminate, that the
commenter suggested, in the guideline. EPA is not aware of any model or
analysis methodology available in the scientific literature that has been
evaluated and demonstrated as being capable of handling stagnation episodes.
The synoptic conditions and mesoscale circulations possible within a stagna-
tion episode result in a truly complicated modeling problem.
9-21
-------
10.0. ACCURACY AND UNCERTAINTY OF MODELS
Comment Summary (Discussion of Accuracy/Uncertainty)
Several commenters requested a clearer distinction between the accuracy
and uncertainty of models. One commenter recommended that this discussion
take place earlier in the guideline, i.e., Chapter 2, and that accuracy of
models should be clearly documented. Other commenters asked for clarifica-
tion of the effect of averaging time on accuracy and uncertainty associated
with models since the reference provided (i.e., Hanna) was for hourly
concentrations only. (APCA, CMA, MCC, NYEC)
EPA Response
The terms "uncertainty" and "accuracy," although not associated with
precise definitions as applied to air quality models, have been defined in
the guideline. They are frequently used interchangeably without objection
by technical experts. Several workshops/conferences have been sponsored
by EPA which dealt with these issues; a clear definition has not evolved.
Generally, "accuracy" is a statistical expression of the results of com-
paring estimates and observations, e.g., bias and scatter. "Uncertainty,"
which encompasses "accuracy," is a more general expression of those factors
that influence "accuracy" and how much the resulting estimates depart from
"truth." A clearer distinction in these terms does not appear possible at
this time. Variations in model accuracy for various averaging times is
well documented elsewhere (see discussion in Section 3.1 concerning "Basis
for Model Selection") and there is not a need for further elaboration in
the guideline. The Hanna reference in the chapter on model uncertainty re-
fers to variations between estimates and observations for 1-hour estimates;
extension to other averaging times is not intended by the guideline
commentary.
EPA has given careful consideration to the general discussion of model
accuracy and its placement in the guideline text. Location earlier in the
10-1
-------
text and more extensive intertwining would be desirable. However, it was
found that this major topic cuts across all techniques used and requires
a separate chapter. Located elsewhere in the guideline text, it would
interfere with the clarity of guidance on appropriate models and data bases
for specific regulatory programs.
Reports on the accuracy of each model included in the guideline are
identified and referenced in the model summaries. Thus, further documenta-
tion of model accuracy seems unnecessary at this time.
Comment Summary (Performance Measures)
Several commenters recommended that model evaluation schemes should
be simplified and that statistical performance measures should be reduced,
standardized and documented in the guideline. Emphasis should be on tests
that evaluate performance at the high end of the concentration frequency
distribution and redundant statistical tests should be eliminated. Two
commenters felt the evaluation should focus on the design value, or second
highest concentration. Other commenters suggested that emphasis should
be given to (1) use of the entire frequency distribution, (2) use of the
mean square error, (3) use of variance ratios and (4) use of confidence
limits in uncertainty determination. One comment approved of EPA's model
performance program but suggested that future model evaluations should
include sources that have stack and processes different from a power plant.
One comment suggested the use of empirical adjustments (calibration) to
improve the fit between observed and predicted values and provide zero
bias. (APCA, CMA, NYEC, PEPC, SOC, TACB)
EPA Response
EPA agrees that it would be desirable to propose simplified and
standardized model evaluation schemes in the guideline. However, there has
•
not been sufficient experience with model evaluation at this time to provide
such guidance in order to eliminate all redundancy. The AMS workshop on judg-
ing models also suggested that it was premature to take such a step. Through
additional guidance from the AMS and a variety of evaluation exercises EPA
is beginning to focus on the usefulness of selected tests. These tests
tend to be at the high end of the frequency distribution. However, tests
10-2
-------
for a wider range of concentrations including perhaps the entire frequency
distribution may also be of use for some applications. Tests for the second-
highest value have been used, but not to the exclusion of other analyses
since they do not satisfy the criteria for robustness in statistical testing,
Thus it seems prudent to await further experience from EPA's model evalution
program, before standardized guidance is attempted.
The mean square error is a powerful statistical test. However, it is
based on paired observed and predicted concentrations and does not allow
the type of comparisons suggested by other commenters and used by EPA for
the upper end of the frequency distribution. Variance ratio analysis is
another useful tool that would supplement, but not replace, other tests
already being used. Confidence limits are available for use as a result
of the model evaluation programs already completed. EPA agrees that
models should be evaluated for point sources other than power plants and
will consider such sources in future evaluations. The use of adjustment
factors to reduce bias can not be recommended because of the arbitrary
nature of such adjustments.
Comment Summary (Use of Best Estimates)
Several commenters recommended that decision-makers be allowed to
depart from the "best estimate" and not be required to use conservative
estimates which lead to inefficient controls and results in costs which
are not proportional to the risks. One commenter requested that multi-
point rollback and the ExEx methodology in setting emission limits be
addressed in the guideline and their use be authorized. Another commenter
questioned the use of inaccurate models where multiple sources are involved
and where space and time comparisons are important because this will lead
to holding the wrong source responsible. (API, DOE, SOC, UARG)
EPA Response
The purpose of the guideline is to develop the "best estimate" for use
by the decision maker, i.e., the design concentration estimated by a model
10-3
-------
recommended in this guideline or an alternate model of known accuracy.
However, given a lack of (1) substantial precedent, (2) proven techniques
for quantifying uncertainty, (3) procedures for using uncertainty information,
and also given the requirement for uniformity in Section 301 of the CAA, it
is not likely that decision-makers will depart significantly from the best
estimate.
Multi-point rollback is a narrowly approved application to smelters and
is not consistent with the more widely used techniques recommended in the
guideline. The ExEx methodology, originally developed for large coal-fired
power plants, appears to be inconsistent with the current deterministic form
of the $62 standard and has never been approved by EPA. Thus, it does not
appear appropriate to extend the use of either technique in the guideline.
It is true that point source Gaussian dispersion models do poorly in
space and time comparisons. As noted in Chapter 10 of the guideline, due
to inherent uncertainty, it is unlikely that even a "perfect" model would
do well for such comparisons. Thus, it does not appear that a change in
preferred models would improve this deficiency. Where monitoring or other
techniques show inaccurate estimates, the guideline (Chapter 11) allows for
the use of alternate approaches to set emission limits.
10-4
-------
11.0 REGULATORY APPLICATION OF MODELS
11 .1 Discussion
Comment Summary (Receptor Models)
One comment endorsed the use of receptor models to set emission
limitations where meteorological dispersion conditions are difficult to
characterize appropriately and where accurately and completely quantifying
emissions is impossible. (AISI)
EPA Response
The use of receptor models to assess control strategies for particulate
matter is discussed in the PM1Q SIP guideline.32 Like dispersion modeling,
usefulness of receptor modeling is limited in the situations where the
emission characteristics of an area are difficult to define. Properly
applied, receptor models are most effective when used in concert with dis-
persion models to provide important new information to regulatory agencies
on an emissions inventory and on source specific impacts calculated by a
dispersion model. Complete documentation and guidance on receptor models
is provided in several EPA publications.56'60
11.2.1 Analysis Requirements
Comment Summary (Selection of Significant Receptors)
A few comments suggested the use of significant receptors rather than
all receptors when remodeling sources. One of these questioned the need to
model all previous receptor sites and suggested that it is sufficient to
select just those receptors needed for a rigorous worst-case impact analysis.
Another said that to require'applicants for PSD permits to model all receptors
considered in previous permit applications when the sources are separated by
an appreciable distance can result 1n far more receptors than is reasonable
or necessary. The area of analysis should be restricted to that where the
impact of the new source or modification is expected to be significant.
Another recommended using a set of 400 receptors including significant, but
not all, receptors from previous applications.
A rn.miP of comments requested more specific guidance on what constitutes
"all source' " ?eqS™ for PSD modeling because that will depend on location,
s?ze Snd Expected impact. One said that much of the contribution from "all
sources" is background.
11-1
-------
Another comment suggested that the guideline address the use of the
significance levels for air quality impact as a screening procedure for
ambient air quality analysis. For sources whose air quality impact is
insignificant, no further analysis should be required to demonstrate
compliance with the NAAQS or PSD increments. If a source impact is found
to be less than the significance levels, an emission inventory for other
sources in the area would not have to be developed nor additional modeling
performed. (APCA, SOC, NYEC, UARG, CONE, ADEM)
EPA Response
Reasonable efforts to estimate the maximum impact of the new source,
as well as the maximum impact, or increment used, of all sources considered,
are necessary. However, guidance on subjective judgment of what receptors
to use cannot be recommended in a regulatory program because of ambiguity
in defining what is significant. EPA recommends early discussion between
the applicant and regulatory control agencies. Selecting receptors based
on a worst-case modeling analysis alone is not recommended because screening
models do not require wind data which determines the location of the most
heavily affected receptors. When refined models are subsequently used, the
area of maximum impact may well be different and the coincident area of
maximum impact of the source with other nearby sources will vary. The
complexity of selecting appropriate receptors in PSD applications is acknow-
ledged by EPA and the guideline has provided a general direction to allow
for a reasonable amount of flexibility. Further explanation of EPA's
rationale is found on page 99-102 of the Summary of Comments and Responses.1
•
"All sources" generally refers to those which have been included in
previous PSD analyses and for which joint impact with the new source is
possible. The sources considered will depend on their location relative to
the source being modeled. Therefore, good judgment in selection of sources
and receptors and the early discussion with control agencies referenced
11-2
-------
above are required. The contribution from all other sources not explicitly
considered could be classified as background. However, as the number of
"background" sources contributing to use of the PSD increment grows, their
impact should be explicitly considered (See Section 9.1 and specifically
Table 9-1 of the revised guideline on how to mode] "nearby" background
sources vs. "other" background sources). Careful consideration must be given
to the problem and a case-specific analysis conducted, as appropriate.
There is no need to address significant levels for air quality impact
because they are already identified elsewhere in regulations dealing with
PSD, bubbles, etc. Screening models can be used to estimate impact of a
specific source, or source category, on the significance levels, as appropriate
(Refer to Chapter 2 of the guideline). No further analysis is required if
the screening showed that the source impact is below the significance level.
When multi-source impacts are involved, screening techniques may be of limited
utility. EPA recommends remodeling all sources with each new application,
unless the State has an increment tracking procedure which would make this
unnecessary.
11.2.2 Use of Measured Data in Lieu of Model Estimates
Comment Summary (Reliance on Measured Data)
A few commenters recommended that for existing sources which already have
ambient monitoring networks meeting EPA requirements actual measurements
should be Preferred over model estimates. One year of ambient monitoring at
a Kiel's m x mum impact point should be adequate One stated that EPA has
relected certain air quality analyses because of the design of the monitoring
network Another requested guidance on how monitor siting should be performed
toTocaie £1^0? 2«1mum impact if a model. Is not appropr ate to a specific
aoo ication Another said that the proposed requirement to identify the most
impo tanl JndividuaTsources as a condition for using measured data cannot be
met In urbl!I areas and recommended deleting criteria (d) from page 11-5.
Another commenter said that the States are nearly equally divided on the
questtS;^whether both air quality data and modeling or Jir quality data
alone should be used for determining attainment status. Should EPA begin
11-3
-------
requiring air quality modeling as part of the attainment demonstration, it
is clear that many areas will have^ to be modeled which have not been
scrutinized previously using that procedure.
One comment asked EPA to define a "regional site" which can be used
when there are no monitors in the vicinity of an isolated single source.
(MSUS, APCA, MCC, CONE, OEPA)
EPA Response
Where models are available and appropriate, modeling is the preferred
method for determining emission limitations, consistent with Clean Air Act
requirements. In some cases when the modeling technique available is only
a screening technique, the addition of air quality data to the analysis may
lend credence or replace model estimates. Suitable criteria are provided in
Section 11.2.2 of the guideline. The design of the air quality monitoring
network can be questioned if there are not sufficient monitors present, if
the monitors are not located in areas that are likely to include the maximum
impact for the various pollutants and emission sources or if the monitoring
period is short or unrepresentative of typical meteorological conditions;
modeling with multiple years of data provides a superior set of information.
However, if monitoring data show a value higher than the modeled value, the
monitoring data should be used if the source inventory is proven to be
correct. There is no basis for deleting item (d) because the analysis must
allow impact of the most important individual sources to be identified, if
more than one source or emission point is involved.
The use of monitor siting to locate points of maximum impact is a
highly subjective procedure for which EPA does not propose guidance. However
the location of those monitors must be based on information developed from
prior monitoring, preliminary model estimates, experience, etc.; the other
alternative is to establish a comprehensive network which allows the maximum
site to be bracketed.
11-4
-------
EPA acknowledges that identifying the most important individual sources
in an urban area is difficult and case-by-case discussion with the Regional
Meteorologist is needed. This identification, however, is not impossible as
the commenter implies since techniques such as tracer gas, receptor analysis,
etc. could be used.
It is current EPA regulatory policy that redesignation from nonattainment
to attainment, requires valid air quality data showing no NAAQS violations
and must be supplemented with a demonstration that an approved SIP control
strategy which provides for attainment has been implemented. Where only the
most recent four quarters of data showing attainment are available, a state-
of-the-art modeling analysis must be provided which quantifies that the SIP
strategy is sound and that actual enforceable emission reductions are
responsible for the air quality improvements. It is not anticipated that
the promulgation of the guideline will result in the need for new air quality
modeling demonstrations.
The definition of a background regional site is given on page 9-8 of
the guideline and should represent (1) the impact of all sources not modeled
because of small size, and (2) natural background concentrations.
11.2.3 Emission Limits
Comment Summary (Miscellaneous)
One comment stated that the guideline recommends that the averaging
time for the design concentration be determined from the ratio of predicted
concentration (P) plus background (B) to the applicable NAAQS(s). The
averaging time with the highest ratio (P+B)/S identifies the most limiting
emission standard for large multi-source areas and not for a.single source.
The limiting emission standard for a single source is determined by calcu-
lating (S-BJ/P for each averaging time and setting the emission standard
from design concentration that minimizes this ratio.
11-5
-------
One comment suggested deleting the "temporal and spatial" modeling
procedures for PSD but supported thisjjse for the emission trading policy.
Another stated the method for calculating the PSD increment consumption is
not clearly stated and should not require the use of sequential modeling.
The comment suggested the use of screening models for this application.
One comment stated that with the exception of ozone, there is no
provision in the regulations which allows consideration of the frequency
with which NAAQS are exceeded over a number of years in determinations of
compliance and the statement on paragraph 2 should be clarified. (APCA,
FDER, MMES, NYEC).
EPA Response
40 CFR Part 51 .13(e)(2)(i) defines the following equation to determine
the fractional reduction needed to attain a standard: (P+B)-S. The S-B
~P T"
formula cited by the commentors is the fraction of emissions which will yield
attainment and is calculated simply by taking 1 minus (reduction formula).
Thus, the constraining standard is that which results in the maximum degree
of reduction (or, conversely, the minimum S-B ratio). The text in the
P
guideline will be changed accordingly.
The methodology for spatial and temporal calculation of PSD increments
is consistent with EPA's interpretation of the Clean Air Act and definition
of increment and baseline concentrations in the PSD regulations. The
methodology is also consistent with the manner in which the total concentra-
tion is calculated for comparison with ambient standards and is consistent
with the method used to calculate incremental concentrations for Level II
emissions trades. The application of screening models has been addressed
elsewhere in the guideline. In addition, screening models are used for
individual sources, not when multiple increment-consuming sources are
involved.
Paragraph 2 in this section of the guideline has been changed to
reflect the commenters clarification.
11-6
-------
APPENDIX A. SUMMARIES.OF PREFERRED AIR QUALITY MODELS
A.O INTRODUCTION
Comment Summary
One commenter suggested that EPA identify what changes or revisions
have been made to the models listed in Appendix A. Another disagreed with
the statement in the introductory paragraph and said that very little cost
.information is provided, and very little information on accuracy, other than
what is identified in Section 10, is presented. One commenter suggested
that EPA provide a "hardwired" optional version of each recommended model as
part of the next UNAMAP distribution. Another recommended that UNAMAP models
be modified to allow for computation of both geometric and arithmetic annual
averages to correspond to the appropriate NAAQS. (TEXA, MES, MMES)
EPA Response
Changes and additions to models listed in Appendix A will be clearly
indicated in supplements to user's guides for these models. It is unclear.
what cost information the second commenter is referring to. The reference
to cost in the introductory paragraph of Appendix A refers to the cost of
obtaining the model and is given wherever the model developer provided that
information. The accuracy of the various models is indicated by reference
to the model validation studies listed in section "n" of each model summary
description. Any further documentation would be redundant.
EPA is planning to develop a hardwired regulatory option in each
recommended EPA model as part of the next UNAMAP distribution (Version 6).
The suggestion to allow computing geometric averages will not be implemented
since this value can be approximated by using assumptions about the frequency
distribution of the concentrations.
A.2 CALINE3
Comment Summary
One commenter stated that the listing of the TEXIN model publication
(Messina, et al.) in the CALINE3 model description section is confusing.
A-l
-------
Further, the Texas work is based on a modified version of CALINE3 which is
not generally available. A separate section dealing with the TEXIN model
was suggested. (DOT)
EPA Response
EPA will reword the description of the CALINE3 model and delete the
reference to TEXIN (please refer to comment responses in Section 6.2.2
dealing with "Models for Carbon Monoxide").
A.4 RAM
Comment Summary
One commenter recommended that an option be included to allow the user
to position a polar grid of receptors around a specific coordinate as is
allowed in the MPTER model. Another noted that the model description
refers incorrectly to the RAM model's application to rolling terrain and
fails to mention the output of annual average concentrations.(MMES, NYEC)
EPA Response
EPA believes that the present treatment of receptor locations is adequate
because receptors can be placed at any location chosen by the user. EPA does
not plan to implement the commenters suggestion due to the substantial modi-
fication of the model code that would be required. Also, EPA will eliminate
the reference to rolling terrain in the RAM model description, since RAM
only treats flat terrain, and will add a note to the model description that
the model prints out annual average concentrations.
A.5 ISC
Comment Summary
One commenter recommended that no receptor elevations be input into ISC
while the effects of building downwash are being simulated. It is further
recommended that ISC be explicitly restricted to flat terrain because it
is overly conservative in terrain less than stack height. Another suggested
that ISC be modified 1) to accept receptor locations with both coordinates
for each receptor input together, 2) to treat building orientation relative
to the wind direction, and 3) to automatically truncate terrain to stack
A-2
-------
height as recommended by EPA for certain situations. The last three suggest-
ions are labor saving changes, and do not affect the resulting concentration
estimates. (WDNR)
EPA Responses
Since the Huber and Snyder downwash algorithm used in ISC only increases
plume size and does not lower plume center!ine height, EPA believes that it
is reasonable to allow treatment of elevated receptors. EPA does not agree
with the commenters claim that the ISC model is overly conservative and the
commenter provided no data to support this claim. The American Petroleum
Institute has asked EPA to allow the use of a modified ISC downwash algorithm
because the present algorithm underpredicts concentration for sources with
short stacks. EPA will implement two of the recommended modifications in the
next revision of the UNAMAP models (Version 6). EPA will propose in a
Federal Register notice changes to the ISC model which will allow orientation
of the building with respect to the wind to be accounted for in building
downwash calculations.
A.6 MPTER
Comment Summary
One commenter suggested that the MPTER output be formatted to display
the input data with more significant digits, corresponding to the input
data format. Another suggestion was that the user be permitted to specify
pollutants in addition to S02 and particulates to improve readability of the
model printout. Also, it was suggested that MPTER gets a 366 day matrix like
CRSTER and ISCST so that the user could select the days to be modeled from
a year of data Another commenter noted that MPTER prints out annual concen-
trations and that this should be noted in the model description. (MMES,
NYEC)
EPA Responses
EPA will change the MPTER output format as suggested by the commenter.
EPA believes the suggestion to specify additional pollutant names is not
A-3
-------
v/orthwhile. MPTER can currently be run for any continuous period from one
hour to one year and no modification to allow an individual day to be
specified is necessary. EPA does not plan to add the 366 day matrix because
this will require a substantial coding modification to the model. EPA will
correct the model description to reflect that MPTER prints out annual
concentrations as the last comment suggested.
A-4
-------
APPENDIX B. SUMMARIES OF ALTERNATIVE AIR QUALITY MODELS
B.3 APRAC-3
Comment Summary
One commenter noted that APRAC-3 cannot use data output from the Urban
Transportation Planning System (UTPS) travel forecasting models currently
used by many states, the comment also noted the lack of evaluation studies
and suggested EPA consider deleting APRAC-3 as an alternative model. (FHA)
EPA Responses
EPA agrees that there is a continuing strong need for coordination in
modeling and evaluation activities among transportation and air quality
planning efforts at the State, local and federal levels. Where a metropo-
litan area or a State is using the UTPS or other transportation modeling
package, and is seeking to perform air quality analyses, EPA is more than
willing to work with the appropriate agencies and FHWA to resolve any
specific analytical difficulties on a case-by-case basis. The inclusion of
a model in Appendix B of the guideline does not constitute a recommendation
for its use, nor does it imply that the model has been evaluated. If a
user wishes to substitute a model from Appendix B for a recommended model,
the request for the substitution to be permitted must be supported with
appropriate validation studies, as described in Section 3.2 of the guideline.
B.9 IMPACT (Sklarew)
This model has been withdrawn at the request of the model developer.
B.I 2 MESOPLUME
This model has been withdrawn at the request of the model developer.
B.I8 Multi-Source (SCSTER) Model
Comment Summary
Several corrections to the description of the SCSTER model were supplied
by its developer. (SCS)
B-l
-------
EPA Response
These corrections will be made to the model description.
B.24 RTDM (Version 3.00)
This model has been withdrawn at the request of the model developer.
B-2
-------
REFERENCES
1. Environmental Protection Agency, 1984. Summary of Comments and Responses
on the October 1980 Proposed Revisions to the Guideline on Air Quality
Models. U.S. Environmental Protection Agency, Research Triangle Park
NC. (Docket A-80-46, Reference No. II-G-5).
2. Environmental Protection Agency, 1978. Guideline on Air Quality Models.
EPA Publication No. EPA 450/2-78-027. U.S. Environmental Protection
Agency, Research Triangle Park, NC. (NTIS No. PB 288-783).
3. Environmental Protection Agency, 1982. Regional Workshops on Air Quality
Modeling: A Summary Report. EPA Publication No. EPA 450/4-82-015. U.S.
Environmental Protection Agency, Research Triangle Park, NC. (NTIS No.
PB 83-150573).
4. Environmental Protection Agency, 1984. Interim Procedures for Evaluating
Air Quality Models (Revised). EPA Publication No. EPA 450/4-84-023. U.S.
Environmental Protection Agency, Research Triangle Park, NC. (NTIS No.
PB 85-106060).
5. Environmental Protection Agency, 1981 . Model Clearinghouse: Operational
Plan. Staff Report. U.S. Environmental Protection Agency, Research
Triangle Park, NC. (Docket Reference No. II-G-6).
6. Environmental Protection Agency, 1980. Guideline on Air Quality Models.
Federal Register, 45(61): 20157-20158.
7. Environmental Protection Agency, 1983. User's Network for Applied
Modeling of Air Pollution (UNAMAP), Version 5 (Computer Programs on Tape).
National Technical Information Service, Springfield, VA. (NTIS No. PB
83-244368) .
8. Londergan, R. J., D. H. Minott, D. J. Wackter, T. Kincaid and D. Bonitata,
1982. Evaluation of Rural Air Quality Simulation Models. EPA Publication
No. EPA-450/4-83-003. U.S. Environmental Protection Agency, Research
Triangle Park, NC. (NTIS No. PB 83-182758).
9. Londergan, R. J., D. H. Minott, D. J. Wackter and R. R. Fizz, 1983.
Evaluation of Urban Air Quality Simulation Models. EPA Publication No.
EPA-450/4-83-020. U.S. Environmental Protection Agency, Research Triangle
Park, NC. (NTIS No. PB 84-241173).
10. Londergan, R. J. and D. J. Wackter, 1984. Evaluation of Complex Terrain
Air Quality Simulation Models. EPA Publication No. EPA-450/4-84-017. .
U.S. Environmental Protection Agency, Research Triangle Park, NC. (NTIS
No. PB 85-119485)
American Meteorological Society, 1983. Synthesis of the Rural Model
Reviews EPA Publication No. EPA-600/3-83-108. U.S. Environmental
tect ion 'Agency, Research Triangle Park, NC. (NTIS No. PB 84-121037).
11
Pro-
R-l
-------
12. American Meteorological Society, 1984. Review of the Attributes and
Performance of Six Urban Diffusion Models. EPA Publication No. EPA 600/S3-
84-089. U.S. Environmental Protection Agency, Research Triangle Park, NC,
(NTIS No. PB 84-236850).
13. White, F. D., Ed., and J. K. S. Ching, R. L. Dennis and W. H. Snyder, 1985.
Summary of Complex Terrain Model Evaluation, EPA Publication No. EPA 600/
3-85-060. U.S. Environmental Protection Agency, Research Triangle Park, NC.
(NTIS No. PB 85-236891).
14. Cox, W. M., and J. A. Tikvart, 1985. Assessing the Performance Level of
Air Quality Models, Paper presented at the 15th International Technical
Meeting on Air Pollution and Its Application, NATO/CCMS, St. Louis, MO.
15. Cox, W. M., G. K. Moss, E. Baldridge, and J. A. Tikvart, 1985. Evaluation
of Rural Air Quality .Simulation Models, Addendum B: Graphical Display of
Model Performance Using the Clifty Creek Data Base. EPA Publication No.
EPA-450/4-83-003b. U.S. Environmental Protection Agency, Research Triangle
Park, NC.
16. Cox, W. M., and G. K. Moss, 1985. Evaluation of Rural Air Quality
Simulation Models, Addendum A: Muskingum River Data Base. EPA Publica-
tion No. EPA-450/4-83-003a. U.S. Environmental Protection Agency,
Research Triangle Park, NC. (NTIS No. PB 85-244135)
>
17. Bowne, N. E. and R. J. Londergan, 1983. Overview, Results and Conclusions
for the EPRI Plume Model Validation and Development Project: Plains Site.
EPRI EA-3074. Electric Power Research Institute, Palo Alto, CA.
18. Moore, G. E., T. E. Stoeckenius and D. A. Stewart, 1982. A Survey of
Statistical Measures of Model Performance and Accuracy for Several Air
Quality Models. EPA Publication No. EPA-450/4-83-001. U.S. Environmental
Protection Agency, Research Triangle Park, NC. (NTIS No. PB 83-260810).
19. Consolidated Edison Company, 1985. Written Comments on the November 1984
Draft Revision of the Guideline on Air Quality Models, p. 8. (Docket
, Reference No. IV-D-40).
20. Fox, D. G., 1981. Judging Air Quality Model Performance. Bulletin of the
American Meteorological Society, 62(5) :599-609.
21. Environmental Protection Agency, 1985. Interim Procedures for Evaluating
Air Quality Models: Experience with Implementation. EPA Publication No.
EPA-450/4-85-006. U.S. Environmental Protection Agency, Research Triangle
Park, NC. (NTIS No. PB 85-242477).
22. Environmental Protection Agency, 1981. Guideline for Fluid Modeling of
Atmospheric Diffusion. EPA Publication No. EPA-600/8-81-009. U.S.
Environmental Protection Agency, Research Triangle Park, NC. (NTIS No.
PB 81-201410).
R-2
-------
23,
24.
25.
26.
27.
Environmental Protection Agency, 198S. Guideline for Determination of
Good Engineering Practice Stack Height (Technical Support Document For
the Stack Height Regulations), Revised. EPA Publication No. EPA 450/4-
80-023R. U.S. Environmental Protection Agency, Research Triangle
Park, NC. (NTIS No. PB 85-225241).
»
Environmental Protection Agency, 1981. Guideline for Use of Fluid Modeling
to Determine Good Engineering Practice (GEP) Stack Height. EPA Publica-
tion No. EPA-450/4-81-003. U.S. Environmental Protection Agency, Research
Triangle Park, NC. (NTIS No. PB 82-145327)
American Meteorological Society, 1981. Air Quality Modeling and the
Clean Air Act: Recommendations to EPA on Dispersion Modeling for Regula-
tory Applications. Boston, MA.
Sethu Raman, S., 1985. Analysis and Evaluation of Statistical Coastal
Fumigation Models. Final Report Submitted to Office of Air Quality
Planning and Standards, Environmental Protection Agency, Research
Triangle Park, NC.
Schiermeier, 1983.
Preprints, 6th
Lavery, R. F., B. R. Greene, B. A. Egan, and F. A.
The EPA Complex Terrain Model Development Program. ., »r. , „„
Symposium on Turbulence and Diffusion, March 15-22, 1983, Boston,
Massachusetts. American Meteorological Society, Boston.
28. Schiermeier, F. A., T. F. Lavery, D. G. Strimaitis, A. Venkatram, B. R.
Green and B. A. Egan, 1983. EPA Model Development for Stable Impingement
on Elevated Terrain Obstacles. Proceedings, 14th International Technical
Meeting on Air Pollution Modeling and Its Applications, Copenhagen, Denmark
29 Environmental Protection Agency, 1984. Guideline for Using the Carbon
Bond Mechanism in City - Specific EKMA. EPA Publication No. EPA 450/4-
84-005. (NTIS No. PB 84-198910).
?n YnrkP M A ?t al Evaluation of Short-Term N02 Plume Models for Point
Souses EPA Publication No. EPA-600/4-81-079. U. S. Environmental
Protection Agency, Research Triangle Park, NC. (NTIS No. PB 82-234329)
31. Environmental Protection Agency, 1985 Compilation of Air Pollutant
Emission Factors Fourth Edition. EPA Publication No. AP-42. U.S.
Environmenta? ?rotation Agency, Research Triangle Park, NC (NTIS No.
PB-275525).
32. Environmental Protection Agency, 1984. PM10 SIP Development Guideline.
D?aft for Public Comment. U.S. Environmental Protection Agency,
Research Triangle Park, NC.
33< r_..,- *,i D.m-ortinn Aaencv, 1985. Control Technology for Sources
Contra?; No. ea-oHaSi: U.S. Environmental Protection Agency,
Research Triangle Park, NC.
R-3
-------
34. Environmental Protection Agency, 1981. An Evaluation Study for the
Industrial Source Complex (ISC) Dispersion Model. EPA Publication
EPA-450/4-81-002, U.S. Environmental Protection Agency, Research
Triangle Park, NC. (NTIS No. PB 81-176539).
35. Environmental Protection Agency, 1985. Dispersion of Airborne Particulates
in Surface Coal Mines, Data Analysis. EPA Publication EPA 450/4-85-001,
U.S. Environmental Protection Agency, Research Triangle Park, NC. (NTIS
No. PB 85-18b411).
36. TRC Environmental Consultants, 1985. Continued Analysis and Derivation
of a Method to Model Pit Retention. Draft Final Report for U.S. Environ-
mental Protection Agency, Research Triangle Park, NC.
37. Engineering-Science, 1980. Field Validation of Atmospheric Dispersion
Models for Natural Gas Compressor Stations. Final Project Report
Prepared for Pipeline Research Committee of the American Gas Association
Engineering-Science, Arcadia, CA.
38. Turner, D. B. and J. S. Irwin, 1982. Extreme Value Statistics Related
to Performance of a Standard Air Quality Simulation Model Using Data at
Seven Power Plant Sites. Atmospheric Environment 16:1907-1914.
39. Turner, D. B. and J. S. Irwin, 1985. The Relation of Urban Model
Performance to Stability. Proceedings 14th International Technical Meet-
ing on Air Pollution Modeling and Its Applications, Copenhagen, Denmark.
40. Turner, D. B., J. S. Irwin, and A. D. Busse, 1985. Comparison of RAM
Model Estimates with 1976 St. Louis RAPS Measurements of Sulfur Dioxide.
Atmospheric Environment 19(2):247-253.
41. Department of the Interior, Minerals Management Services, 1985. Oil and
Gas and Sulfur Operations in the Outer Continental Shelf, 30 CFR Part 250.
Federal Register, 50(60):12248-12249.
42. Bjorklund, J. R. and J. F. Bowers, 1982. User's Instructions for the
SHORTZ and LONGZ Computer Programs. EPA Publication Mo. EPA-903/9-82-
004a,b, U.S. Environmental Protection Agency, Region III, Philadelphia,
PA.
43. Cramer, H. E., Company, Inc., 1982. Tests of the Briggs (1974) and
Cramer, et al. (1975) stack-tip downwash corrections and the Schulman
and Scire (T9"80) Building Downwash Correction using the Huber, et al.
(1980) Data Set. Technical Note, H. E. Cramer Company, Inc., SaTtTake
City, UT.
44. Briggs, G. A., 1974. Diffusion Estimates for Small Emissions, in ERL,
ARL, USAEC Report ATDL-106. U.S. Atomic Energy Commission, Oak Ridge,
I n«
45. Bringfeld, B., 1968. Plume Rise Measurements at Industrial Chimneys.
Atmospheric Environment, 2(6):575-598.
R-4
-------
46. Huber, A. H. and W. H. Snyder, 1976. Building Wake Effects on Short
Stack Effluents. Preprint Volume for the Third symposium on Atmospheric
Turbulence, Diffusion and Air Quality, Americam Meteorological Society,
Boston, MA, pp. 235-242.
47. Bowers, J. F., J. R. Bjorklund, and C. S. Cheney, 1979. Industrial Source
Complex (ISC) Dispersion Model User's Guide, Volume 1. EPA Publication No.
EPA-450/4-79-030. U.S. Environmental Protection Agency, Research Triangle
Park, NC. (NTIS No. PB 80-133044).
48. Schulman, L. L. and J. S. Scire, 1980. Bouyant Line and Point Source
(BLP) Dispersion Model User's Guide. ERT Document No. P-73Q43, Envir
mental Research & Technology, Inc., Concord, MA.
50.
51
52.
53.
54.
55.
ron-
49. Cramer, et al., 1975. Diffusion Model Calculations of Long-term and
Short-term Ground-level S02 Concentrations in Allegheny County, Penn-
sylvania. EPA Publication No. EPA-903/9-75-018. U. S. Environmental
Protection Agency, Region III, Philadelphia, PA. (NTIS No. PB 245262),
Environmental Protection Agency, 1982. Evaluation of Rural Air Quality
Simulation Models, Appendix G, Statistical Tables for PPSP. EPA Publica-
tion No. EPA 450/4-83-003. U.S. Environmental Protection Agency, Research
Triangle Park, NC. (NTIS No. PB 83-182758).
Petersen, W. B. and J. S. Irwin, 1985. Climatological Variability in
Modeled Maximum Concentrations. Atmospheric Environment, 19(5): 739-742.
Holzworth, G. C., 1972. Mixing Heights, Wind Speeds, and Potential for
Urban Air Pollution Throughout the Contiguous United States. Office of
Air Programs Pub. No. AP-101. U.S. Environmental Protection Agency,
Research Triangle Park, NC.
Strimaitis, D., G. Hoffnagle and A. Bass, 1981. On-Site Meteorological
Instrumentation Requirements to Characterize Diffusion From Point
Sources: Workshop Report. EPA Publication No. EPA 600/9-81-020. U.S.
Environmental Protection Agency, Research Triangle Park, NC. (NTIS No.
PB 81-247223).
Kaimal, J, C., J. E. Gaynor, P. L. Finkelstein, M.
• ' wind Measurements
Evaluation of
U.S. Department of Commerce,
E. Graves, and T. J.
by Four Doppler Sodars.
NOAA, Boulder Atmospheric
Lockhart, 1984. An
Report Number Five.
Observatory, CO.
Turner D B 1961. Relationships Between 24-Hour Mean Air Quality
Measurements and Meteorological Factors in Nashville, TN. JAPCA, 11: 483-489,
56 Environmental Protection Agency, 1981. Receptor Model Technical Series
VolumeT Overview of Receptor Model Applications to Particulate Source
Apportionment? EPA Publication No. EPA-450/4-81-016a. U.S. Environmental
P?o?ect°on Agency, Research Triangle Park, NC. (NTIS No. PB 82-139429).
R-5
-------
57. Environmental Protection Agency, 1981. Receptor Model Technical Series
Volume II: Chemical Mass Balance. EPA Publication No. EPA 450/4-81-016b.
U.S. Environmental Protection Agency, Research Triangle Park, NC. (NTIS
No. PB 82-187345).
58. Environmental Protection Agency, 1983. Receptor Model Technical Series
Volume III: User's Manual for Chemical Mass Balance Model. EPA Publica-
tion No. EPA 450/4-83-014. U.S. Environmental Protection Agency, Research
Triangle Park, NC. (NTIS No. PB 83-264622).
59. Environmental Protection Agency, 1983. Receptor Model Technical Series
Volume IV: Summary of Particle Identification Techniques. EPA Publica-
tions No. EPA 450/4-83-018. U.S. Environmental Protection Agency, Research
Triangle Park, NC. (NTIS No. PB 84-103340).
60. Environmental Protection Agency, 1983. Receptor Model Technical Series
Volume V: Source Apportionment Techniques and Considerations In Combin-
ing Their Use. EPA Publication No. EPA 450/4-84-020. U.S. Environmental
Protection Agency, Research Triangle Park, NC. (NTIS No. PB 84-241173).
R-6
-------
GLOSSARY OF COMMENTERS APPEARING IN DOCKET A-80-46
Abb. Commenter Docket Number
OKIG Ohio-Kentucky-Indiana Regional Council of Govt. IV-D-1
WDNR Wisconsin Department of Natural Resources IV-D-2
DOT Department of Transportation, Division of
Engineering Services IV-D-3
WCHD Wayne County Health Department Air Pollution
Control Division IV-D-4
ADHS Arizona Department of Health Services IV-D-5
CDH Colorado Department of Health IV-D-7
NDDH North Dakota State Department of Health IV-D-8
MMS Minerals Management Service IV-D-9
ISBH Indiana State Board of Health IV-D-10
GMC General Motors Corporation IV-D-11
NYDOT New York Department of Transportation IV-D-12
UARG Utility Air Regulatory Group ITV"?"IJ
EPNG El Paso Natural Gas Company IV-D- 5
CC Cabot Corporation Ij-D- 6
FHA US Department of Transportation, Federal iv-u-i/
Highway Administration
SOC Standard Oil Company (Indiana) },, ~5~ ix
CDOT California Department of Transportation IV-D-20
PSCN Public Service Company of New Mexico IV~P~™
SCS Southern Company Services \lnll
ADEM Alabama Department of Environmental Management IV-D-25
AISI American Iron and Steel Institute \l~n £
UARG Utility Air Regulatory. Group TunoQ
API American Petroleum Institute IV-D-28
AMC American Mining Congress T\/ n ™
CMA Chemical Manufacturers Association IV-D-30
MCC Magma Copper Company TJ n ?i
MMES Martin Marietta Environmental Systems IV-D-32
OEPA Ohio Environmental Protection Agency IV-D-33
^r^a^ent of Natural Resources 1^35
IPL Indianapolis Power and Light Company IV-D-36
PEPC Potomac Electric Power Company iv-u-j/
NYEC New York Department of Environmental Conservation V-D-38
PENE Pennsylvania Electric Company |V-u-jy
CONE Consolidated Edison Company of New York V-D-40
KC Kennecott Copper TV D 42
CHEV Chevron, USA < TV n d^
MSUS Middle South Utilities iv-u-ij
G-l
-------
Abb. Commenter Docket Number
DOE Department of Energy IV-G-7
TVA Tennessee Valley Authority IV-G-8
NYEC New York Department of Environmental Conservation IV-G-10
NCNR North Carolina Department of Natural Resources
and Comm-unity Development IV-G-11
FDER Florida Department of Environmental Regulations IV-G-12
TAC3 Texas Air Pollution Control Board IV-G-13
APCA Air Pollution Control Association IV-G-14
UARG Utility Air Regulatory Group IV-G-15
ERT Environmental Research & Technology IV-G-17
API American Power Institute IV-G-18
CMA Chemicals Manufacturers Association IV-G-19
DS Diamond Shamrock/Texas Chemical Council IV-G-20
PEPC Potomac Electric Power Company IV-6-21
MES Meteorological Evaluation Services, Inc. IV-G-22
KC Kennecott Copper IV-G-23
NRDC Natural Resources Defense Council IV-G-24
G-2
-------
Abb. Commenter Docket Number
ODEQ Oregon Department of Environmental Quality IV-H-1
TEGP Texas Eastern Gas Pipeline Company IV-H-2
WC Weyerhaeuser Company IV-H-3
CITG Citgo Petroleum Corporation IV-H-4
PHC Peabody Holding Company IV-H-5
BAAQ Bay Area Air Quality Management District IV-H-6
TEXA Texco Inc. . IV-H-7
ASRC Atmospheric Science Research Center IV-H-8
NCA National Coal Association IV-H-9
CITG Citgo Holding Company IV-H-10
UOF University of Florida IV-H-11
IEPA Illinois Environmental Protection Agency IV-H-12
JCPL Jersey Central Power & Light Company IV-H-13
DS Diamond Shamrock IV-H-14
PPL Pennsylvania Power & Light Company IV-H-15
KOCH Koch Refining Company IV-H-16
SRP Salt River Project IV-H-17
CARB California Air Resources Board IV-H-18
NYCP City of New York Dept. of Environmental Protection IV-H-19
TVA Tennessee Valley Authority IV-H-20
UARG Utility Air Regulatory Group IV-H-21
OEPA Ohio Environmental Protection Agency IV-H-22
ARCO ARCO Petroleum Products IV-H-24
CONE Consolidated Edison Company of New York IV-H-25
UARG Utility Air Regulatory Group IV-H-26
OEPA Ohio Environmental Protection Agency IV-H-27
6-3
------- |