Rule: Air Quality Implementation Plans: Preparation, Adoption, and Submittal&#8212 Air Quality Models Guidelines

Wednesday,
November 9, 2005
Part III
Environmental
Protection Agency
40 CFR Part 51
Revision to the Guideline on Air Quality
Models: Adoption of a Preferred General
Purpose (Flat and Complex Terrain)
Dispersion Model and Other Revisions;
Final Rule
VerDate Aug<31>2005 16:55 Nov 08, 2005 Jkt 205001 PO 00000 Frm 00001 Fmt 4717 Sfmt 4717 E:\FR\FM\09NOR3.SGM 09NOR3
68218 Federal Register / Vol. 70, No. 216 / Wednesday, November 9, 2005 / Rules and Regulations
ENVIRONMENTAL PROTECTION telephone (919) 541–5562. SCRAM) at: www.epa.gov/scram001.

AGENCY (Fox.Tyler@epa.gov). You may find codes and documentation
SUPPLEMENTARY INFORMATION: for models referenced in today’s action
40 CFR Part 51 on the SCRAM Web site. We have also
Outline uploaded various support documents
[AH–FRL–7990–9] I. General Information (e.g., evaluation reports).
II. Background
RIN 2060–AK60 III. Public Hearing on the April 2000 II. Background
proposal The Guideline is used by EPA, States,
Revision to the Guideline on Air IV. Discussion of Public Comments and
Quality Models: Adoption of a and industry to prepare and review new
Issues from our April 21, 2000 Proposal
Preferred General Purpose (Flat and A. AERMOD and PRIME
source permits and State
Complex Terrain) Dispersion Model B. Appropriate for Proposed Use Implementation Plan revisions. The
and Other Revisions C. Implementation Issues/Additional Guideline is intended to ensure
Guidance consistent air quality analyses for
AGENCY: Environmental Protection D. AERMOD revision and reanalyses in activities regulated at 40 CFR 51.112,
Agency (EPA). 2003 51.117, 51.150, 51.160, 51.166, and
ACTION: Final rule. 1. Performance analysis for AERMOD 52.21. We originally published the
(02222) Guideline in April 1978 and it was
SUMMARY: EPA’s Guideline on Air a. Non-downwash cases: AERMOD (99351) incorporated by reference in the
vs. AERMOD (02222)
Quality Models (‘‘Guideline’’) addresses regulations for the Prevention of
b. Downwash cases
the regulatory application of air quality 2. Analysis of regulatory design Significant Deterioration (PSD) of Air
models for assessing criteria pollutants concentrations for AERMOD (02222) Quality in June 1978. We revised the
under the Clean Air Act. In today’s a. Non-downwash cases Guideline in 1986, and updated it with
action we promulgate several additions b. Downwash cases supplement A in 1987, supplement B in
and changes to the Guideline. We c. Complex terrain July 1993, and supplement C in August
recommend a new dispersion model— E. Emission and Dispersion Modeling 1995. We published the Guideline as
AERMOD—for adoption in appendix A System (EDMS) appendix W to 40 CFR part 51 when we
V. Discussion of Public Comments and Issues issued supplement B. We republished
of the Guideline. AERMOD replaces the
from our September 8, 2003 Notice of
Industrial Source Complex (ISC3) Data Availability
the Guideline in August 1996 (61 FR
model, applies to complex terrain, and VI. Final action 41838) to adopt the CFR system for
incorporates a new downwash VII. Final editorial changes to appendix W labeling paragraphs. On April 21, 2000
algorithm—PRIME. We remove an VIII. Statutory and Executive Order Reviews we issued a Notice of Proposed
existing model—the Emissions Rulemaking (NPR) in the Federal
I. General Information Register (65 FR 21506), which was the
Dispersion Modeling System (EDMS)—
from appendix A. We also make various A. How Can I Get Copies of Related original proposal for today’s
editorial changes to update and Information? promulgation.
reorganize information. EPA established an official public III. Public Hearing on the April 2000
DATES: This rule is effective December 9, docket for this action under Docket No. Proposal
2005. As proposed, beginning November A–99–05. The official public docket is We held the 7th Conference on Air
9, 2006, the new model—AERMOD— the collection of materials that is Quality Modeling (7th conference) in
should be used for appropriate available for public viewing at the Air Washington, DC on June 28–29, 2000.
application as replacement for ISC3. Docket in the EPA Docket Center, (EPA/ As required by Section 320 of the Clean
During the one-year period following DC) EPA West (MC 6102T), 1301 Air Act, these conferences take place
this promulgation, protocols for Constitution Ave., NW., Washington, approximately every three years to
modeling analyses based on ISC3 which DC 20004. The EPA Docket Center standardize modeling procedures, with
are submitted in a timely manner may Public Reading Room (B102) is open special attention given to appropriate
be approved at the discretion of the from 8:30 a.m. to 4:30 p.m., Monday modeling practices for carrying out
appropriate Reviewing Authority. through Friday, excluding legal programs PSD (42 U.S.C. 7620). This
Applicants are therefore encouraged to holidays. The telephone number for the conference served as the forum for
consult with the Reviewing Authority as Reading Room is (202) 566–1744, and receiving public comments on the
soon as possible to assure acceptance the telephone number for the Air Docket Guideline revisions proposed in April
during this period. is (202) 566–1742. An electronic image 2000. The 7th conference featured
ADDRESSES: All documents relevant to of this docket may be accessed via presentations in several key modeling
this rule have been placed in Docket No. Internet at www.epa.gov/eDocket, where areas that support the revisions
A–99–05 at the following address: Air Docket No. A–99–05 is indexed as promulgated today. A presentation by
Docket in the EPA Docket Center, (EPA/ OAR–2003–0201. Materials related to the American Meteorological Society
DC) EPA West (MC 6102T), 1301 our Notice of Data Availability (AMS)/EPA Regulatory Model
Constitution Ave., NW., Washington, (published September 8, 2003) and Improvement Committee (AERMIC)
DC 20004. This docket is available for public comments received pursuant to covered the enhanced Gaussian
public inspection and copying between the notice were placed in eDocket OAR– dispersion model with boundary layer
8 a.m. and 5:30 p.m., Monday through 2003–0201.1 parameterization: AERMOD.2 Also at
Friday, at the address above. Our Air Quality Modeling Group the 7th conference, the Electric Power
FOR FURTHER INFORMATION CONTACT: maintain an Internet website (Support Research Institute (EPRI) presented
Tyler J. Fox, Air Quality Modeling Center for Regulatory Air Models— evaluation results from the recent
Group (MD–D243–01), Office of Air 1 http://cascade.epa.gov/RightSite/
research efforts to better define and
Quality Planning and Standards, U.S. dk_public_collection_detail.htm? characterize dispersion around
Environmental Protection Agency, ObjectType=dk_docket_collection&cid=OAR-2003-
Research Triangle Park, NC 27711; 0201&ShowList=items&Action=view. 2 AMS/EPA Regulatory MODel.
Federal Register / Vol. 70, No. 216 / Wednesday, November 9, 2005 / Rules and Regulations 68219
buildings (downwash effects). These to public concerns, we believed that the considered and discussed all significant
efforts were part of a program called the revised AERMOD merited another comments. Whenever the comments
Plume RIse Model Enhancements public examination of performance revealed any new information or
(PRIME). At the time, PRIME was results. Also, since the April 2000 NPR, suggested any alternative solutions, we
integrated within ISC3ST (ISC–PRIME) the Federal Aviation Administration considered this prior to taking final
and the results presented were within (FAA) decided to configure EDMS 3.1 to action.
the ISC3 context. As discussed in incorporate the AERMOD dispersion The remainder of this preamble
today’s rule, the PRIME algorithm has model. FAA presented this strategy at section discusses the primary issues
now been fully integrated into the 7th conference and performance encountered by the Agency during the
AERMOD. evaluations at two airports were to be public comment period associated with
We proposed an update to the available before final promulgation. the April 2000 proposal. This overview
Emissions and Dispersion Modeling This was in response to public concern also serves in part to explain the
System (EDMS 3.1), which is used for over lack of EDMS evaluation. changes to the Guideline in today’s
assessing air quality impacts from On April 15, 2003 we published a action, and the main technical and
airports. A representative of the Federal Notice of Final Rulemaking (NFR; 68 FR policy concerns addressed by the
Aviation Administration (FAA) 18440) that adopted CALPUFF in Agency.
presented a further upgrade to EDMS appendix A of the Guideline. We also A. AERMOD and PRIME
4.0 that would include AERMOD and made various editorial changes to
forthcoming performance evaluations update and reorganize information, and AERMOD is a best state-of-the-
for two airports. removed obsolete models. We practice Gaussian plume dispersion
The presentations were followed by a announced that action on AERMOD and model whose formulation is based on
critical review/discussion of AERMOD the Emissions and Dispersion Model planetary boundary layer principles.
and available performance evaluations, (EDMS) for assessing airport impacts AERMOD provides better
facilitated jointly by the Air & Waste was being deferred, and would be characterization of plume dispersion
Management Association’s AB–3 reconsidered in a separate action when than does ISC3. At the 7th conference,
Committee and the American new information became available for AERMIC members presented
Meteorological Society’s Committee of these models. developmental and evaluation results of
Meteorological Aspects of Air Pollution. This deferred action took the form of AERMOD. Comprehensive comments
For the new models and modeling a Notice of Data Availability (NDA), were submitted on the AERMOD code
techniques proposed in April 2000, we which was published on September 8, and formulation document and on the
asked the public to address the 2003 (68 FR 52934). In this notice, we AERMET draft User’s Guide (AERMET
following questions: made clear that the purpose of the NDA is the meteorological preprocessor for
• Has the scientific merit of the was to furnish pertinent technical AERMOD).
models presented been established? details related to model changes since As identified in the April 2000
• Are the models’ accuracy the April 2000 NPR. New performance Federal Register proposal, applications
sufficiently documented? for which AERMOD was suited include
data and evaluation of design
• Are the proposed regulatory uses of assessment of plume impacts from
concentration using the revised
individual models for specific stationary sources in simple,
AERMOD are contained in reports cited
applications appropriate and intermediate, and complex terrain, for
later in this preamble (see section V). In
reasonable? other than downwash and deposition
our April 2003 NFR, we stated that
• Do significant implementation applications. We invited comments on
results of EDMS 4.0 performance (with
issues remain or is additional guidance whether technical concerns had been
AERMOD) had recently become
needed? reasonably addressed and whether
available. In the NDA we clarified that
• Are there serious resource these results would not be provided
AERMOD is appropriate for its intended
constraints imposed by modeling applications. Since AERMOD lacks a
because of FAA’s decision to withdraw general (all-terrain) screening tool, we
systems presented? EDMS from the Guideline’s appendix A,
• What additional analyses or invited comment on the practicality of
and we affirmed our support for this using SCREEN3 as an interim tool for
information are needed?
removal. We solicited public comments AERMOD. We also sought comments on
We placed a transcript of the 7th
on the new data and information related minor changes to the list of acceptable
conference proceedings and a copy of
to AERMOD. screening techniques for complex
all written comments, many of which
address the above questions, in Docket IV. Discussion of Public Comments and terrain.
No. A–99–05. The comments on Issues From Our April 21, 2000 PRIME was designed to incorporate
AERMOD were reviewed and nearly Proposal the latest scientific algorithms for
every commenter urged us to integrate evaluating building downwash. At the
All comments submitted to Docket
aerodynamic downwash into AERMOD time of the proposal, the PRIME
No. A–99–05 are filed in Category IV–
(i.e., not to require two models for some algorithm for simulating aerodynamic
D.3 We summarized these comments,
analyses). The only comments calling downwash was not incorporated into
developed detailed responses, and
for further actions were associated with AERMOD. For testing purposes, PRIME
documented conclusions on appropriate
the need for documentation, evaluation was implemented within ISC3ST (short-
actions in a Response-to-Comments
and review of the suggested downwash term average version of the Industrial
document.4 In this document, we
enhancement to AERMOD. Source Complex), which AERMOD was
As a result of American 3 Additional comments received since we
proposed to replace. This special model,
Meteorological Society (AMS)/EPA published the final rule on April 15, 2003 called ISC–PRIME, was proposed for
Regulatory Model Improvement (discussed in the previous section) are filed in
Committee’s (AERMIC) efforts to revise category IV–E. This category includes comments Quality Modeling; Washington, DC, June 28–29,
received pursuant to the Notice of Data Availability 2000 AND Notice of Data Availability—September
AERMOD, incorporating the PRIME we published in September 2003. 8, 2003 (Air Docket A–99–05, Item V–C–2). This
algorithm and making certain other 4 Summary of Public Comments and EPA document may also be examined from EPA’s
incidental modifications and to respond Responses: AERMOD; 7th Conference on Air SCRAM Web site at www.epa.gov/scram001.
aerodynamic downwash and dry and the reasoning and references for the simultaneous use of two models (ISC–
deposition. We sought comment on the model assumptions.9 PRIME and AERMOD) for those sources
technical viability of AERMOD and Although some comments asked for with potential downwash concerns.
ISC–PRIME for its intended more detailed documentation and Commenters urged the Agency to
applications. review, there were no comments which eliminate the need to use two models
questioned the technical credibility of for evaluating the same source. In
Scientific merit and accuracy.
the PRIME model. In fact, almost every response to this request, AERMIC
Regarding the scientific merits of
commenter asked for PRIME to be developed a version of AERMOD that
AERMOD, substantial support was
incorporated into AERMOD. As incorporates PRIME: AERMOD (02222)
expressed in public comments that
summarized above, we believe that the and initiated an analysis to insure that
AERMOD represents sound and scientific merit of PRIME has been concentration estimates by AERMOD
significant advances over ISC3ST. The established via (1) model evaluation and (02222) are equivalent to ISC–PRIME
scientific merits of this approach have documentation, (2) peer review within predictions in areas affected by
been documented both through the submittal process to a technical downwash before it replaces ISC–
scientific peer review and performance journal, and (3) via the public review PRIME. Careful thought was given to the
evaluations. The formulation of process. way that PRIME was incorporated into
AERMOD has been subjected to an Based on the external peer review of AERMOD, with the goal of making the
extensive, independent peer review.5 the evaluation report and the public merge seamless. While discontinuities
Findings of the peer review panel review comments, we have concluded from the concatenation of these two sets
suggest that AERMOD’s scientific basis that: (1) AERMOD’s accuracy is of algorithms were of concern, we
is ‘‘state-of-the-science.’’ Additionally, adequately documented; (2) AERMOD’s mitigated this situation wherever
the model formulations used in accuracy is an improvement over possible (see part D of this preamble,
AERMOD and the performance ISC3ST’s ability to predict measured and the Response to Comments
evaluations have been accepted for concentrations; and (3) AERMOD is an document 4). With regard to testing the
publication in two refereed journals.6 7 acceptable regulatory air dispersion performance of AERMOD (02222), we
Finally, the adequacy of AERMOD’s model replacement for ISC3ST. have carefully confirmed that the
complex terrain approach for regulatory Some commenters have identified AERMOD (02222)’s air quality
applications is seen most directly in its what they perceived to be weaknesses in concentration predictions in the wake
performance. AERMOD’s complex the evaluation and performance of ISC– region reasonably compare to those
terrain component has been evaluated PRIME,10 and some concerns were predictions from ISC–PRIME. In fact,
extensively by comparing model- raised about the scope of the PRIME the results indicate that AERMOD
estimated regulatory design values and evaluation. However, as shown by the (02222)’s performance matches the
concentration frequency distributions overwhelming number of requests for performance of ISC–PRIME, and are
with observations. These comparisons the incorporation of PRIME into presented in an updated evaluation
have demonstrated AERMOD’s AERMOD, commenters were convinced report 11 and analysis of regulatory
superiority to ISC3ST and CTDMPLUS that the accuracy of PRIME, as design concentrations.12 We discuss
(Complex Terrain Dispersion Model implemented within the ISC3ST AERMOD (02222) performance in detail
PLUS unstable algorithms) in estimating framework, was reasonably documented in part D.
those flat and complex terrain impacts and found acceptable for regulatory Because the technical basis for the
of greatest regulatory importance.8 For applications. Although some PRIME algorithms and the AERMOD
incidental and unique situations commenters requested more formulations have been independently
involving a well-defined hill or ridge evaluations, practical limitations on the peer-reviewed, we believe that further
and where a detailed dispersion number of valid, available data sets peer review of the new model
analysis of the spatial pattern of plume prevented the inclusion of every source (AERMOD 02222) is not necessary. The
impacts is of interest, CTDMPLUS in the type and setting in the evaluation. All scientific formulation of the PRIME
Guideline’s appendix A remains the data bases that were reasonably algorithms has not been changed.
available. available were used in the development However, the coding for the interface
and evaluation of the model, and those between PRIME and the accompanying
Public comments also supported our
data bases were sufficient to establish dispersion model had to be modified
conclusion about the scientific merits of
the basis for the evaluation. Based on somewhat to accommodate the different
PRIME. A detailed article in a peer-
our review of the documentation and ways that ISC3ST and AERMOD
reviewed journal has been published
the public comments, we conclude that simulate the atmosphere. The main
which contains all the basic equations
the accuracy of PRIME is sufficiently public concern was the interaction
with clear definitions of the variables,
documented and find it acceptable for between the two models and whether
5 U.S. Environmental Protection Agency, 2002.
use in a dispersion model recommended the behavior would be appropriate for
Compendium of Reports from the Peer Review in the Guideline. all reasonable source settings. This
Process for AERMOD. February 2002. Available at B. Appropriate for Proposed Use concern was addressed through the
www.epa.gov/scram001/. extensive testing conducted within the
6 Cimorelli, A. et al., 2005. AERMOD: A Responding to a question posed in our performance evaluation 11 and analysis
Dispersion Model for Industrial Source April 2000 proposal, the majority of
Applications. Part I: General Model Formulation of design concentrations.12 Both sets of
and Boundary Layer Characterization. Journal of
commenters questioned the
Applied Meteorology, 44(5): 682–693. reasonableness of requiring 11 Environmental Protection Agency, 2003.
7 Perry, S. et al., 2005. AERMOD: A Dispersion AERMOD: Latest Features and Evaluation Results.
Model for Industrial Source Applications. Part II: 9 Schulman, L.L. et al., 2000. Development and Publication No. EPA–454/R–03–003. Available at
Model Performance against 17 Field Study Evaluation of the PRIME Plum Rise and Building www.epa.gov/scram001/.
Databases. Journal of Applied Meteorology, 44(5): Downwash Model. JAWMA 50: 378–390. 12 Environmental Protection Agency, 2003.
694–708. 10 Electric Power Research Institute, 1997. Results Comparison of Regulatory Design Concentrations:
8 Paine R. J. et al., 1998. Evaluation Results for of the Independent Evaluation of ISCST3 and ISC– AERMOD versus ISC3ST, CTDMPLUS, and ISC–
AERMOD, Draft Report. Docket No. A–99–05; II–A– PRIME. Final Report, TR–2460026, November 1997. PRIME. Final Report. Publication No. EPA–454/R–
05. Available at www.epa.gov./scram001/. Available at www.epa.gov/scram001/. 03–002. Available at www.epa.gov/scram001/.
analyses indicate that the new model is the rural unstable dispersion settings Survey (USGS), (3) appropriately use
performing acceptably well and the (only the rural, stable dispersion setting Digital Elevation Model (DEM) data
results are similar to those obtained considered meander in AERMOD with 2 different datums (NAD27 and
from the earlier performance (99351)—this change created a NAD83); (4) accept all 7 digits of the
evaluation 8 10 and analysis of regulatory consistent treatment of air dispersion in North UTM coordinate, and (5) do more
design concentrations (i.e., for AERMOD all dispersion settings); error-checking in the raw data (mostly
(99351)).13 • Making some changes to the basic checking for missing values, but not for
While dry deposition is treated in meander algorithms (improved harsh terrain changes in adjacent
ISC3ST, time and resources did not scientific formulation); and points). All of these recommendations
allow its incorporation in AERMOD • Repairing miscellaneous coding have been implemented.
(99351). Since no recommendation for errors. In response to comments about the
deposition is made for regulatory As we mentioned earlier, the version selection of the domain affecting the
applications, we did not consider that of AERMOD that is being promulgated results of the maximum concentrations
the absence of this capability today—AERMOD (02222)—has been in complex terrain and the way
compromises the suitability of subjected to further performance AERMAP estimates the effective hill
AERMOD for its intended purposes. evaluation 11 and analysis of design height scale (hC), the algorithms within
Nevertheless, a number of commenters concentrations.12 AERMAP and AERMOD have been
requested that deposition algorithms be C. Implementation Issues/Additional adjusted so that the hill height is less
added to AERMOD, and we developed Guidance sensitive to the arbitrary selection of the
an update to AERMOD (02222) that domain. This adjustment has been
offers dry and wet deposition for both Other than miscellaneous suggestions
evaluated against the entire set of
gases and particles as an option. for certain enhancements for AERMOD
(99351) such as a Fortran90 compilation evaluation data. The correction was
The version of AERMOD under found to substantially reduce the effect
review at the 7th Conference was of the source code, creation of
allocatable arrays, and development of a of the domain size upon the
AERMOD (99351) and, as mentioned computation of controlling hill heights
above, AERMIC has made a number of Windows graphical user interface, no
significant implementation obstacles for each receptor. Application of this
changes to AERMOD (99351) following change to the evaluation databases did
this conference. These changes were were identified in public comments.
For AERMET (meteorological not materially affect the evaluation
initiated in response to public results.
comments and, after the release of a new preprocessor for AERMOD), we have
implemented some enhancements that In general, public comments that
draft version of the model, in response requested additional guidance were
to the recommendations from the beta commenters suggested. For site-specific
applications, several commenters cited either obviated by revisions to AERMOD
testers. Changes made to AERMOD (99351) and its related preprocessors or
include the following: AERMOD’s requirements for NWS cloud
deemed unnecessary. In the latter case,
• Adding the PRIME algorithms to the cover data. In response, we revised the
AERMET to incorporate the bulk the reasons were explained in the
model (response to public comments);
• Modifying the complex terrain Richardson number methodology. This Response-to-Comments document.4
approach uses temperature differences Some public comments suggested
algorithms to make AERMOD less
near the surface of the earth, which can additional testing of AERMOD (99351).
sensitive to the selection of the domain
of the study area (response to public be routinely monitored, and eliminates In fact, after the model revisions that
comments); the need for the cloud cover data at were described earlier were completed,
• Modifying the urban dispersion for night. We made a number of other AERMOD (02222) was subjected to
low-level emission sources, such as area revisions in response to public additional testing.11 12 These new
sources, to produce a more realistic comments, enabling AERMET to: (1) analyses will be discussed in part D.
urban dispersion and, as a part of this Use the old and the new Forecasting With respect to a screening version of
change, changing the minimum layer Systems Laboratory formats, (2) use the AERMOD, a tool called AERSCREEN is
depth used to calculate the effective Hourly U.S. Weather Observations/ being developed with a beta version
dispersion parameters for all dispersion Automated Surface Observing Stations expected to be publicly available in Fall
settings (scientific formulation (HUSWO/ASOS) data, (3) use site- 2005. SCREEN3 is the current screening
correction which was requested by beta specific solar radiation and temperature model in the Guideline, and since
testers); and gradient data to eliminate the need for SCREEN3 has been successfully applied
• Upgrading AERMOD to include all cloud cover data, (4) appropriately for a number of years, we believe that
the newest features that exist in the handle meteorological data from above SCREEN3 produces an acceptable
latest version of ISC3ST such as the arctic circle, and (5) accept a wider degree of conservatism for regulatory
Fortran90 compliance and allocatable range of reasonable friction velocities applications and may be used until
arrays, EVENTS processing and the and reduce the number of warning AERSCREEN or a similar technique
TOXICS option (response to public messages. As mentioned earlier, we becomes available and tested for general
comments). added a meander component to the application.
In the follow-up quality control treatment of stable and unstable urban D. AERMOD Revision and Reanalyses
checking of the model and the source conditions to consistently treat meander Published In 2003
code, additional changes were identified phenomena for all cases.
as necessary and the following revisions AERMAP (the terrain preprocessor for 1. Performance Analysis for AERMOD
were made: AERMOD) has been upgraded in (02222)
• Adding meander treatment to: (1) response to public comments calling for We have tested the performance of
Stable and unstable urban cases, and (2) it to: (1) Treat complex terrain receptors AERMOD (02222) by applying all of the
13 Peters, W.D. et al., 1999. Comparison of
without a dependance on the selected original data sets used to support the
Regulatory Design Concentrations: AERMOD vs.
domain, (2) accommodate the Spatial version proposed in April, 2000:
ISCST3 and CTDMPLUS, Draft Report. Docket No. Data Transfer Standard (SDTS) data AERMOD (99351) 8 and ISC–PRIME.10
A–99–05; II–A–15. available from the U.S. Geological These data sets include: 5 complex
terrain data sets, 7 building downwash Evaluation of AERMOD (02222) pollutants, stack heights and averaging
data sets, and 5 simple terrain data sets With the changes to AERMOD (99351) times where the proposed (ISC–PRIME)
(see appendix A of the Response-to- as outlined above, how has the model performance could be compared
Comments document 4). This performance of the AERMOD been to the performance of AERMOD (02222)
performance analysis, which is a check affected? The performance of the current with PRIME incorporated. There was an
of the model’s maximum concentration version of AERMOD is about the same equal number of non-downwash cases
predictions against observed data, or slightly better than the April 2000 where AERMOD performed better than
includes a comparison of the current version when a comparison is made ISC–PRIME and where ISC–PRIME
version of the new model (AERMOD over all the available data sets. There
performed better than AERMOD. There
02222) with ISC3ST or ISC–PRIME for was only one case where there was a
were examples of AERMOD (02222)
downwash conditions. The results and significant difference between the two
showing better and poorer performance
conclusions of the performance analyses models’ performance, and AERMOD
when compared to the performance
are presented in 2 sections: Non- clearly performed better than ISC–
results of AERMOD (99351). However,
downwash and downwash source PRIME in this case. In all other cases,
for those cases where AERMOD
scenarios. the difference in the performance,
(02222)’s performance was degraded,
whether an improvement or a
a. Non-Downwash Cases the degradation was small. On the other
degradation, was small. This
For the user community to obtain a side, there were more examples where
comparison indicated that AERMOD
full understanding of the impacts of AERMOD (02222) more closely (02222) performs very similarly, if not
today’s proposal for the non-downwash predicted measured concentrations. The somewhat better, when compared to
source scenarios (flat and complex performance improvements were also ISC–PRIME for downwash cases.
terrain), our performance evaluation of rather small but, in general, were
AERMOD (02222) must be discussed somewhat larger than the size of the 2. Analysis of Regulatory Design
with respect to the old model, ISC3ST, performance degradations. There also Concentrations for AERMOD (02222)
and with respect to AERMOD (99351). were a number of cases where the Although not a performance tool, the
Based on the evaluation, we have performance remained unchanged analysis of design concentrations
concluded that AERMOD (02222) between the 2 models. Thus, overall, (‘‘consequence’’ analysis) is designed to
significantly outperforms ISC3ST and there was a slight improvement in test model stability and continuity, and
that AERMOD (02222)’s performance is AERMOD’s performance and, to help the user community understand
even better than that of AERMOD consequently, we believe that AERMOD the differences to be expected between
(99351). (02222) significantly outperforms air dispersion models. The
ISC3ST for non-downwash source consequences, or changes in the
Evaluation of AERMOD (99351) scenarios. regulatory concentrations predicted
Comparative performance statistics For AERMOD (02222) with the 5 data when using the new model (AERMOD
were calculated for both ISC3ST and bases examined for simple terrain, the 02222) versus ISC3ST, cover 96 source
AERMOD (99351) using data sets in ratios of modeled/observed Robust High scenarios and at least 3 averaging
non-downwash conditions. This Concentration ranged from 0.77 to 1.11 periods per source scenario, and are
analysis looked at combinations of test (1-hr average), 0.98 to 1.24 (3-hr evaluated and summarized here. The
sites (flat and complex terrain), average), 0.94 to 0.97 (24-hr average) purpose is to provide the user
pollutants, and concentration averaging and 0.30 to 0.97 (annual average). These community with a sense of potential
times. Comparisons indicated very ratios reflect better performance than changes in their air dispersion analyses
significant improvements in ISC3ST for all cases. when applying the new model over a
performance when applying AERMOD For AERMOD (02222) with the 5 data broad range of source types and settings.
(99351). In all but 1 of the total of 20 bases examined for complex terrain, The consequence analysis, in which
cases in which AERMOD (99351) could these ratios ranged from 1.03 to 1.12 (3- AERMOD was run for hundreds of
be compared to ISC3ST, AERMOD hr average), 0.67 to 1.78 (24-hr average) source scenarios, also provides a check
performed as well as (but generally and 0.54 to 1.59 (annual average). At for model stability (abnormal halting of
better than) ISC3ST, that is, AERMOD Tracy—the only site for which there are model executions when using valid
predicted maximum concentrations that 1-hr data—AERMOD performed control files and input data) and for
were closer to the measured maximum considerably better (ratio = 1.04) than spurious results (unusually high or low
concentrations. In the most dramatic either ISC3ST or CTDMPLUS. At three concentration predictions which are
case (i.e., Lovett; 24-hr) in which of the other four sites, AERMOD unexplained). The results are placed
AERMOD performed better than generally performed much better than into 3 categories: non-downwash source
ISC3ST, AERMOD’s maximum either ISC3ST or (where applicable) scenarios in flat, simple terrain;
concentration predictions were about alternative models for the 3-hr and 24- downwash source scenarios in flat
the same as the measured hr averaging times; results were terrain; and, complex terrain source
concentrations while the ISC3ST’s comparable for Clifty Creek (for the 3- settings. The focus of this discussion is
predicted maximum concentrations hr averaging times, AERMOD (02222) on how design concentrations change
were about 9 times higher than the predictions were only about 5% higher from those predicted by ISC3ST when
measured concentrations. In the one than ISC3ST’s—down from 25% for applying the latest version of AERMOD
case (i.e., Clifty Creek; 3-hr) where AERMOD (99351) as described earlier). versus applying the earlier version of
ISC3ST performed better than AERMOD At the two sites where annual peak AERMOD (99351).
(99351), ISC3ST’s concentration comparisons are available, AERMOD
predictions matched the observed data performed much better than either a. Non-Downwash Cases
and the AERMOD concentration ISC3ST or alternative models. For the non-downwash situations,
predictions were about 25% higher than there were 48 cases covering a variety of
the observed data. These results were b. Downwash Cases source types (point, area, and volume
reported in the supporting For the downwash data sets, there sources), stack heights, terrain types
documentation for AERMOD (99351). were combinations of test sites, (flat and simple), and dispersion
settings (urban and rural). For each case from 2.24 for ISC–PRIME/ISC3ST to were almost unchanged. There were no
in the consequence analysis, we 1.87 for AERMOD (02222)/ISC3ST and cases in either consequence analysis
calculated the ratio between AERMOD’s the minimum value of the concentration where AERMOD (02222 & 99351)
regulatory concentration predictions ratios range from 0.34 for ISC–PRIME/ predicted higher concentrations than
and ISC3ST’s regulatory concentration ISC3ST to 0.38 for AERMOD (02222)/ those predicted by ISC3ST. Thus, in
predictions. The average ratio of ISC3ST. These results show relatively general, the consequences of moving
AERMOD to ISC3ST-predicted close agreement between the two PRIME from ISC3ST to AERMOD (02222) rather
concentrations changed from 1.14 when models. (See Table 4–6 in reference 12.) than to AERMOD (99351) in complex
applying AERMOD (99351) to 0.96 ISC3ST does not predict cavity terrain were essentially the same. (See
when applying AERMOD (02222).14 concentrations but comparisons can be Table 4–9 in reference 12.)
Thus, in general, AERMOD (02222) made between AERMOD and ISC– E. Emission and Dispersion Modeling
tends to predict concentrations closer to PRIME. The average AERMOD (02222) System (EDMS)
ISC3ST than does version 99351 predicted 1-hour cavity concentration is
proposed in April 2000. Also, the about the same (112%) as the average The Emissions and Dispersion
variation of the differences between ISC–PRIME 1-hour cavity concentration. Modeling System (EDMS) was
ISC3ST and AERMOD has decreased In the extremes, the AERMOD (02222)- developed jointly by the Federal
with AERMOD (02222). Comparing the predicted cavity concentrations ranged Aviation Administration (FAA) and the
earlier consequence analysis to the from about 40% higher to 15% lower U.S. Air Force in the late 1970s and first
latest study with AERMOD (02222), we than the corresponding ISC–PRIME released in 1985 to assess the air quality
saw a 25% reduction in the number of cavity concentration predictions. Thus, of proposed airport development
cases where the AERMOD-predicted in general, where downwash is a projects. EDMS has an emissions
concentrations differed by over a factor significant factor, AERMOD (02222) and preprocessor and its dispersion module
of two from ISC3ST’s predictions. ISC–PRIME predict similar maximum estimates concentrations for various
concentrations. (See Table 4–8 in averaging times for the following
b. Downwash Cases pollutants: CO, HC, NOX, SOX, and
reference 12.)
For the downwash analysis, there Although the same downwash suspended particles (e.g., PM–10). The
were 20 cases covering a range of stack algorithms are used in both models, first published application of EDMS was
heights, locations of stacks relative to there are differences in the melding of in December 1986 for Stapleton
the building, dispersion settings, and PRIME with the core model, and International Airport (FAA–EE–11–A/
building shapes. As before, we differences in the way that these models REV2).
calculated the ratio regulatory simulate the atmosphere.15 The In 1988, version 4a4 revised the
concentration predictions from downwash algorithm implementation dispersion module to include an
AERMOD (02222 with PRIME) and therefore could not be exactly the same. integral dispersion submodel: GIMM
compared them as ratios to those from (Graphical Input Microcomputer
ISC3ST for each case. For additional c. Complex Terrain Model). This version was proposed for
information, we also included ratios During the testing of AERMOD after adoption in the Guideline’s appendix A
with ISC–PRIME that was also proposed modifications were made to the in February 1991 (56 FR 5900). This
in April 2000. complex terrain algorithm (see version was included in appendix A in
Calculated over all the 20 cases, and discussion of hill height scale (hC) in B. July 1993 (58 FR 38816) and
for all averaging times considered, the Appropriate for Proposed Use in this recommended for limited applications
average ISC–PRIME to ISC3ST preamble), a small error was found in for assessments of localized airport
concentration ratio is about 0.86, the original complex terrain code while impacts on air quality. FAA later
whereas for AERMOD (PRIME) to conducting the consequence analysis. updated EDMS to Version 3.0.
ISC3ST, it is 0.82. The maximum value This error was subsequently repaired. In response to the growing needs of
of the concentration ratios range from Final testing indicated that the revised air quality analysts and changes in
2.24 for ISC–PRIME/ISC3ST to 3.67 for complex terrain code produced regulations (e.g., conformity
AERMOD (PRIME)/ISC3ST. Similarly, reasonable results for the consequence requirements from the Clean Air Act
the minimum value of the concentration analysis, as described below. Amendment of 1990), FAA updated
ratio range from 0.04 for ISC–PRIME/ The analysis of predicted design EDMS to version 3.1, which is based on
ISC3ST to 0.08 for AERMOD (PRIME)/ concentrations included a suite of the CALINE3 16 and PAL2 dispersion
ISC3ST. (See Table 4–5 in reference 12.) complex terrain settings. There were 28 kernels. In our April 2000 NPR we
Although results above for the two cases covering a variety of stack heights, proposed to adopt the version 3.1
models that use PRIME—AERMOD stack gas buoyancy values, types of update to EDMS. However, this update
(02222) and ISC–PRIME—show hills, and distances between source and had not been subjected to performance
differences, we find that building terrain. The ratios between the evaluation and no studies of EDMS’
downwash is not a significant factor in AERMOD (02222 & 99351)—predicted performance have been cited in
determining the maximum maximum concentrations and the appendix A of the Guideline. Comment
concentrations in some of the cases, i.e., ISC3ST maximum concentrations were was invited on whether this
the PRIME algorithms do not predict a calculated for all cases for a series of compromises the viability of EDMS 3.1
building cavity concentration. Of those averaging times. When comparing as a recommended or preferred model
cases where downwash was important, AERMOD (99351) to ISC3ST and then and how this deficiency can be
the average concentration ratios of ISC– AERMOD (02222) to ISC3ST, the corrected.
PRIME/ISC3ST and AERMOD (02222)/ average maximum concentration ratio, Several commenters expressed
ISC3ST are about 1. The maximum the highest ratios and the lowest ratios concern about EDMS 3.1 as a
value of the concentration ratios range recommended model in appendix A.
15 AERMOD uses more complex techniques to Indeed, there were concerns that EDMS
14 A ratio of 1.00 indicates that the two models estimate temperature profiles which, in turn, affect
are predicting the same concentrations. See Table the calculation of the plume rise. Plume rise may 16 Currently listed in appendix A of the
4.1 in reference 12. affect the cavity and downwash concentrations. Guideline.
3.1 had not been as well validated as comments, developed responses, and methodology used to evaluate AERMOD
other models, nor subjected to peer documented conclusions on appropriate (one that emphasizes Robust High
review, as required by the Guideline’s actions for today’s notice. Whenever the Concentration), claiming it is ill-suited
subsection 3.1.1. One of these comments revealed any new to the way dispersion models estimate
commenters suggested that EDMS 3.1 information or suggested any alternative ambient concentrations. We
should be presented only as one of solutions, we considered them in our acknowledged that other methods are
several alternative models. final action and made corrections or available that are designed to reflect the
At the 7th Conference, FAA proposed enhancements where appropriate. underlying physics and formulations of
for appendix A adoption an even newer, In the remainder of this preamble dispersion models, and may be more
enhanced version of EDMS—version section we highlight the main issues robust in their mechanisms to account
4.0, which incorporates the AERMOD raised by the commenters who reviewed
for the stochastic nature of the
dispersion kernel (without alteration). the NDA, and summarize our responses.
atmosphere. In fact, we cited several
In this system, the latest version of These comments broadly fall into two
categories: technical/operational, and recent cases from the literature in which
AERMOD would be employed as a
standalone component of EDMS. This administrative. such methods were applied in
dispersion kernel was to replace PAL2 The technical/operational comments evaluations that included AERMOD. We
and CALINE3 currently in EDMS 3.1. were varied. One commenter thought also explained that the approach taken
There were no public comments specific EPA’s sensitivity studies for simulating by AERMIC was based on existing
to FAA’s proposed AERMOD-based area sources were too limited, and noted guidance in section 9 of Appendix W,
enhancements to EDMS announced after that AERMOD, when used to simulate and expressed a commitment to explore
our April 2000 NPR. an area source adjacent to gently sloping other methods in the future, including
In response to written comments on terrain, produced ground-level an update to section 9. We believe
our April 2000 NPR, at the 7th concentrations not unlike those from however that the evaluation
Conference (transcript) FAA promised a ISC3ST. In response we explained methodology used was reasonable for its
complete evaluation process that would qualitatively how AERMOD interprets intended purpose—examining a large
include sensitivity testing, intermodel this situation and cautioned that array of concentrations for a wide
comparison, and analysis of EDMS reviewing authorities should be variety of source types—and confers a
predictions against field observations. consulted in such scenarios for measure of consistency given its past
The intermodel comparisons were guidance on switch settings. Other use. Other commenters expressed
proposed for the UK’s Atmospheric commenters believed that AERMOD disappointment that AERMOD wasn’t
Dispersion Modeling System (ADMS).17 exhibited unrealistic treatment of compared to state-of-the-science models
As we explained in our September 8, complex terrain elements and offered as advised in its peer review report. In
2003 Notice of Data Availability, FAA supporting data. In response, AERMIC response, we cited a substantial list of
has decided to withdraw EDMS from concluded that AERMOD does exhibit studies in which AERMOD has, in fact,
the Guideline’s appendix A. We stated terrain amplification factors on the
been compared to some of these models,
that no new information was therefore windward side of isolated hills, where
e.g., HPDM and ADMS (in various
provided in that notice, and we affirmed impacts are expected to be greatest.
Commenters also presented evidence combinations). On the whole, as we
support for EDMS’ removal from noted in our response, AERMOD
appendix A. This removal, which we that the PRIME algorithm in AERMOD
misbehaves in its treatment of building typically performed as well as HPDM
promulgate today, obviates the need for
wake and wind incidence. Another and ADMS, and all of them generally
EDMS’ documentation and evaluation at
model was cited as having better skill in performed better than ISC3ST. Still
this time.
this regard. In response, we others expressed disappointment that
V. Discussion of Public Comments on acknowledged this but established that the evaluation input data weren’t posted
Our September 8, 2003 Notice of Data AERMOD’s capability was acceptable on our Web site until January 22, 2004—
Availability for handling the majority of building three months after the close of the
As mentioned in section III, after geometries encountered (see Response- comment period. We acknowledge that
AERMOD was revised pursuant to to-Comments document 4 for more the input data were not posted when the
comments received on the April 21, details). NDA was published. However, the
2000 proposal, a Notice of Data A number of commenters addressed actual evaluation input data for
Availability (NDA) was issued on administrative or procedural matters. AERMOD had not been requested
September 8, 2003 to explain the Some believed that the transition period previously, and we did not believe they
modifications and to reveal AERMOD’s for implementation—one year—is too were required as a basis for reviewing
new evaluation data. Public comments short. We explained in response that the reports we released. Moreover, since
were solicited for 30 days and posted one year is consistent with past practice the posting, we are unaware of any
electronically in eDocket OAR–2003– and is adequate for most users and belated adverse comments from anyone
0201.1 (As mentioned in section IV, reviewing authorities given our previous attempting to access and use the data.
additional comments received since we experience with new models and the
fact that AERMOD has been in the We believe we have carefully
published the final rule on April 15,
public domain for several years. Some considered and responded to public
2003 are filed in Docket A–99–05;
were disappointed that the review comments and concerns regarding
category IV–E.) We summarized these
comments and developed detailed period (30 days) for the NDA was too AERMOD. We have also made efforts to
responses; these appear as appendix C short. We believe that the period was update appendix W to better reflect
to the Response-to-Comments adequate to review the two reports that current practice in model solicitation,
document.4 In appendix C, we presented updated information on the evaluation and selection. We also have
considered and discussed all significant performance and practical consequences made other technical revisions so the
of the model as revised. Regarding the guidance conforms with the latest form
17 Cambridge Environmental Research evaluation/comparison regime used for of the PM–10 National Ambient Air
Consultants; http://www.cerc.co.uk/. AERMOD, others objected to the Quality Standard.
VI. Final Action 2004. The latest version of AERMOD Since AERMOD (02222) was released,
In this section we explain the changes may now be used for deposition an updated version was posted on our
to the Guideline in today’s action in analysis in special situations. Web site on March 22, 2004: AERMOD
Since AERMOD treats dispersion in (04079). The version we are releasing
terms of the main technical and policy
complex terrain, we have merged pursuant to today’s promulgation,
concerns addressed by the Agency in its
sections 4 and 5 of appendix W, as however, is AERMOD (04300). This
response to public comments (sections
proposed in the April 2000 NPR. And version, consonant with AERMOD
IV & V). Air quality modeling involves
while AERMOD produces acceptable (02222) in its formulations, addresses
estimating ambient concentrations using
regulatory design concentrations in the following minor code issues:
scientific methodologies selected from a
complex terrain, it does not replace • The area source algorithm in simple
range of possible methods, and should
CTDMPLUS for detailed or receptor- and complex terrain required a
utilize the most advanced practical
oriented complex terrain analysis, as we correction to the way the dividing
technology that is available at a
have made clear in Guideline section streamline height is calculated.
reasonable cost to users, keeping in
4.2.2. CTDMPLUS remains available for • In PRIME, incorrect turbulence
mind the intended uses of the modeling
use in complex terrain. parameters were being passed to one of
and ensuring transparency to the public.
We have implemented the majority of the numerical plume rise routines, and
With these changes, we believe that the
suggestions to improve the AERMET, this has been corrected.
Guideline continues to reflect recent • A limit has been placed on plume
AERMAP, and AERMOD source code to
advances in the field and balance these cooling within PRIME to avoid
reflect all the latest features that have
important considerations. Today’s supercooling, which had been causing
been available in ISC3ST and that are
action amends Appendix W of 40 CFR runtime instability.
available in the latest versions of
part 51 as detailed below: • A correction has been made to
Fortran compilers. Also, the latest
AERMOD formats for meteorological and terrain avoid AERMOD’s termination under
input data are now accepted by the new certain situations with capped stacks
Based on the supporting information (i.e., where the routine was attempting
contained in the docket, and reflected in versions of AERMET and AERMAP. Our
guidance, documentation and users’ to take a square root of a negative
peer review and public comments, we number). Our testing has demonstrated
find that the AERMOD modeling system guides have been modified in response
to a number of detailed comments. only very minor impacts from these
and PRIME are based on sound corrections on the evaluation results or
With respect to AERMOD (02222)’s
scientific principles and provide the consequence analysis.
performance, we have concluded that:
significant improvements over the (1) AERMOD (99351), the version AERMOD (04300) has other draft
current regulatory model, ISC3ST. proposed in April 2000, performs portions of code that represent options
AERMOD characterizes plume significantly better than ISC3ST, and not required for regulatory applications.
dispersion better than ISC3ST. The AERMOD (02222) performs slightly These include:
accuracy of the AERMOD system is better than AERMOD (99351) in non- • Dry and wet deposition for both
generally well-documented and superior downwash settings in both simple and gases and particles;
to that of ISC3ST. We are adopting the complex terrain; • The ozone limiting method (OLM),
model based on its performance and (2) The performance evaluation referenced in section 5.2.4 (Models for
other factors. indicates that AERMOD (02222) Nitrogen Dioxide—Annual Average) of
Public comments on the April 2000 performs slightly better than ISC–PRIME the Guideline for treating NOX
proposal expressed significant concern for downwash cases. conversion; and
about the need to use two models With respect to changes in AERMOD’s • The Plume Volume Molar Ratio
(AERMOD and ISC–PRIME) to simulate regulatory design concentrations Method (PVMRM) for treating NOX
just one source when downwash posed compared to those for ISC3ST, we have conversion.
a potential impact. In response to this concluded that: • The bulk Richardson number
concern we incorporated PRIME into • For non-downwash settings, approach (discussed earlier) for using
AERMOD and documented satisfactory AERMOD (02222), on average, tends to near-surface temperature difference has
tests of the algorithm. AERMOD, with predict concentrations closer to ISC3ST, been corrected in AERMOD (04300).
the inclusion of PRIME, is now and with somewhat smaller variations, Based on the technical information
appropriate and practical for regulatory than the April 2000 proposal of contained in the docket for this rule,
applications. AERMOD; and with consideration of the
The state-of-the-science for modeling • Where downwash is a significant performance analysis in combination
atmospheric deposition continues to factor in the air dispersion analysis, with the analysis of design
evolve, the best techniques are currently AERMOD (02222) predicts maximum concentrations, we believe that
being assessed, and their results are concentrations that are very similar to AERMOD is appropriate for regulatory
being compared with observations. ISC–PRIME’s predictions; use and we are revising the Guideline to
Consequently, as we now say in • For those source scenarios where adopt it as a refined model today.
Guideline paragraph 4.2.2(c), the maximum 1-hour cavity concentrations In implementing the changes to the
approach taken for any regulatory are calculated, the average AERMOD Guideline, we recognize that there may
purpose should be coordinated with the (02222)-predicted cavity concentration arise occasions in which the application
appropriate reviewing authority. We tends to be about the same as the of a new model can result in the
agreed with the public comments average ISC–PRIME cavity discovery by a permit applicant of
calling for the addition of state-of-the- concentrations; and previously unknown violations of
science deposition algorithms, and • In complex terrain, the NAAQS or PSD increments due to
developed a modification to AERMOD consequences of using AERMOD emissions from existing nearby sources.
(02222) for beta testing. This model, (02222) instead of ISC3ST remained This potential has been acknowledged
AERMOD (04079) was posted on our essentially unchanged in general, previously and is addressed in existing
Web site http://www.epa.gov/scram001/ although they varied based on EPA guidance (‘‘Air Quality Analysis for
tt25.htm#aermoddep on March 19, individual circumstances. Prevention of Significant Deterioration
(PSD),’’ Gerald A. Emison, July 5, 1988). the interim, as appropriate, we will Section 6
To summarize briefly, the guidance consider the use of either ISC3 or As proposed, we renumbered this to
identifies three possible outcomes of AERMOD in air toxic risk assessment become section 5. In subsection 5.1, we
modeling by a permit applicant and applications. reference the Plume Volume Molar
details actions that should be taken in Ratio Method (PVMRM) for point
EDMS
response to each: sources of NOX, and mention that it is
1. Where dispersion modeling shows FAA has completed development of
currently being tested to determine
no violation of a NAAQS or PSD the new EDMS4.0 to incorporate
suitability as a refined method.
increment in the impact area of the AERMOD. The result is a conforming
proposed source, a permit may be enhancement that offers a stronger Section 7
issued and no further action is required. scientific basis for air quality modeling. As proposed, we renumbered this to
2. Where dispersion modeling FAA has made this model available on become section 6. We updated the
predicts a violation of a NAAQS or PSD its Web site, which we cite in an reference to the Emissions and
increment within the impact area but it updated Guideline paragraph 7.2.4(c). Dispersion Modeling System (EDMS).
is determined that the proposed source As described earlier in this preamble,
will not have a significant impact (i.e., the summary description for EDMS will Section 8
will not be above de minimis levels) at be removed from appendix A. As proposed, we revised section 8
the point and time of the modeled (renumbered to section 7) to provide
violation, then the permit may be issued VII. Final Editorial Changes to
Appendix W guidance for using AERMET
immediately, but the State must take (AERMOD’s meteorological
appropriate actions to remedy the Today’s update of the Guideline takes preprocessor).
violations within a timely manner. the form of many revisions, and some of • In subsection 7.2.4, we introduce
3. Where dispersion modeling the text is unaltered. Therefore, as a the atmospheric stability
predicts a violation of a NAAQS or PSD purely practical matter, we have chosen characterization for AERMOD.
increment within the impact area and it to publish the new version of the entire • In subsection 7.2.5, we describe the
is determined that the proposed source text of appendix W and its appendix A. plume rise approaches used by
will have a significant impact at the Guidance and editorial changes AERMOD.
point and time of the modeled violation, associated with the resolution of the
then the permit may not be issued until issues discussed in the previous section Section 9
the source owner or operator eliminates are adopted in the appropriate sections As proposed, we renumbered section
or reduces that impact below of the Guideline, as follows: 9 to become section 8. We added
significance levels through additional paragraphs 8.3.1.2(e) and 8.3.1.2(f) to
controls or emissions offsets. Once it Preface
clarify use of site specific
does so, then the permit may be issued You will note some minor revisions of meteorological data for driving
even if the violation persists after the appendix W to reflect current EPA CALMET in the separate circumstances
source owner or operator eliminates its practice. of long range transport and for complex
contribution, but the State must take Section 4 terrain applications.
further appropriate actions at nearby
sources to eliminate the violations As mentioned earlier, we revised Section 10
within a timely manner. section 4 to present AERMOD as a As proposed, we revised section 10
In previous promulgations, we have refined regulatory modeling technique (renumbered section 9) to include
traditionally allowed a one-year for particular applications. AERMOD. In May 1999, the D.C. Court
transition (‘‘grandfather’’) period for Section 5 of Appeals vacated the PM–10 standard
new refined techniques. Accordingly, we promulgated in 1997, and this
for appropriate applications, AERMOD As mentioned above, we merged
standard has since been removed from
may be substituted for ISC3 during the pertinent guidance in section 5
the CFR (69 FR 45592; July 30, 2004).
one-year period following the (Modeling in Complex Terrain) with
Paragraph 10.2.3.2(a) has been corrected
promulgation of today’s notice. that in section 4. With the anticipated
to be consistent with the current
Beginning one year after promulgation widespread use of AERMOD for all
(original) PM–10 standard, which is
of today’s notice, (1) applications of terrain types, there is no longer any
based on expected exceedances.
ISC3 with approved protocols may be utility in the previous differentiation
accepted (see DATES section) and (2) between simple and complex terrain for Section 11
AERMOD should be used for model selection. To further simplify, the As proposed, we renumbered section
appropriate applications as a list of acceptable, yet equivalent, 11 to become section 10.
replacement for ISC3. screening techniques for complex
We separately issue guidance for use terrain was removed. CTSCREEN and Sections 12 & 13
of modeling for facility-specific and guidance for its use are retained; We renumbered section 12 to become
community-scale air toxics risk CTSCREEN remains acceptable for all section 11, and section 13 (References)
assessments through the Air Toxics Risk terrain above stack top. The screening to become section 12. We revised
Assessment Reference Library.18 We techniques whose descriptions we renumbered section 12 by adding some
recognize that the tools and approaches removed, i.e., Valley (as implemented in references, deleting obsolete/superseded
recommended therein will eventually SCREEN3), COMPLEX I (as ones, and resequencing. You will note
reflect the improved formulations of the implemented in ISC3ST), and RTDM that the peer scientific review for
AERMOD modeling system and we remain available for use in applicable AERMOD and latest evaluation
expect to appropriately incorporate cases where established/accepted references have been included.
them as expeditiously as practicable. In procedures are used. Consultation with
the appropriate reviewing authority is Appendix A
18 http://www.epa.gov/ttn/fera/risk still advised for application of these We added AERMOD (with the PRIME
_atra_main.html. screening models. downwash algorithm integrated) to
appendix A. We removed EDMS from needed to review instructions; develop, D. Unfunded Mandates Reform Act of
appendix A. We also updated the acquire, install, and utilize technology 1995
description for CALPUFF, and made and systems for the purposes of Title II of the Unfunded Mandates
minor updates to some of the other collecting, validating, and verifying Reform Act of 1995 (UMRA), Public
model descriptions. information, processing and Law 104–4, establishes requirements for
maintaining information, and disclosing Federal agencies to assess the effects of
Availability of Related Information
and providing information; adjust the their regulatory actions on State, local,
Our Air Quality Modeling Group existing ways to comply with any and tribal governments and the private
maintains an Internet Web site (Support previously applicable instructions and sector. Under section 202 of the UMRA,
Center for Regulatory Air Models— requirements; train personnel to be able EPA generally must prepare a written
SCRAM) at: http://www.epa.gov/ to respond to a collection of statement, including a cost-benefit
scram001. You may find codes and information; search data sources; analysis, for proposed and final rules
documentation for models referenced in complete and review the collection of with ‘‘Federal mandates’’ that may
today’s action on the SCRAM Web site. information; and transmit or otherwise result in expenditures to State, local,
In addition, we have uploaded various disclose the information. and tribal governments, in the aggregate,
support documents (e.g., evaluation An agency may not conduct or or to the private sector, of $100 million
reports). sponsor, and a person is not required to or more in any one year. Before
VIII. Statutory and Executive Order respond to a collection of information promulgating an EPA rule for which a
Reviews unless it displays a currently valid OMB written statement is needed, section 205
control number. The OMB control of the UMRA generally requires EPA to
A. Executive Order 12866: Regulatory numbers for EPA’s regulations in 40 identify and consider a reasonable
Planning and Review CFR are listed in 40 CFR part 9. number of regulatory alternatives and
Under Executive Order 12866 [58 FR C. Regulatory Flexibility Act (RFA) adopt the least costly, most cost-
51735 (October 4, 1993)], the Agency effective or least burdensome alternative
must determine whether the regulatory The RFA generally requires an agency that achieves the objectives of the rule.
action is ‘‘significant’’ and therefore to prepare a regulatory flexibility The provisions of section 205 do not
subject to review by the Office of analysis of any rule subject to notice apply when they are inconsistent with
Management and Budget (OMB) and the and comment rulemaking requirements applicable law. Moreover, section 205
requirements of the Executive Order. under the Administrative Procedure Act allows EPA to adopt an alternative other
The Order defines ‘‘significant or any other statute unless the agency than the least costly, most cost-effective
regulatory action’’ as one that is likely certifies that the rule will not have a or least burdensome alternative if the
to result in a rule that may: significant economic impact on a Administrator publishes with the final
(1) Have an annual effect on the substantial number of small entities. rule an explanation why that alternative
economy of $100 million or more or Small entities include small businesses, was not adopted. Before EPA establishes
adversely affect in a material way the small organizations, and small any regulatory requirements that may
economy, a sector of the economy, governmental jurisdictions. significantly or uniquely affect small
productivity, competition, jobs, the For purposes of assessing the impact governments, including tribal
environment, public health or safety, or of today’s rule on small entities, small governments, it must have developed
State, local, or tribal governments or entities are defined as: (1) A small under section 203 of the UMRA a small
communities; business that meets the RFA default government agency plan.
(2) Create a serious inconsistency or definitions for small business (based on The plan must provide for notifying
otherwise interfere with an action taken Small Business Administration size potentially affected small governments,
or planned by another agency; standards), as described in 13 CFR enabling officials of affected small
(3) Materially alter the budgetary 121.201; (2) a small governmental governments to have meaningful and
impact of entitlements, grants, user fees, jurisdiction that is a government of a timely input in the development of EPA
or loan programs of the rights and city, county, town, school district or regulatory proposals with significant
obligations of recipients thereof; or special district with a population of less Federal intergovernmental mandates,
(4) Raise novel legal or policy issues than 50,000; and (3) a small and informing, educating, and advising
arising out of legal mandates, the organization that is any not-for-profit small governments on compliance with
President’s priorities, or the principles enterprise which is independently the regulatory requirements.
set forth in the Executive Order. owned and operated and is not Today’s rule recommends a new
It has been determined that this rule dominant in its field. modeling system, AERMOD, to replace
is not a ‘‘significant regulatory action’’ After considering the economic ISC3ST as an analytical tool for use in
under the terms of Executive Order impacts of today’s final rule on small SIP revisions and for calculating PSD
12866 and is therefore not subject to EO entities, I certify that this action will not increment consumption. AERMOD has
12866 review. have a significant economic impact on been used for these purposes on a case-
a substantial number of small entities. by-case basis (per Guideline subsection
B. Paperwork Reduction Act As this rule merely updates existing 3.2.2) for several years. Since the two
This final rule does not contain any technical requirements for air quality modeling systems are comparable in
information collection requirements modeling analyses mandated by various scope and purpose, use of AERMOD
subject to review by OMB under the CAA programs (e.g., prevention of itself does not involve any significant
Paperwork Reduction Act, 44 U.S.C. significant deterioration, new source increase in costs. Moreover, modeling
3501 et seq. review, State Implementation Plan costs (which include those for input
Burden means the total time, effort, or revisions) and imposes no new data acquisition) are typically among
financial resources expended by persons regulatory burdens, there will be no the implementation costs that are
to generate, maintain, retain, or disclose additional impact on small entities considered as part of the programs (i.e.,
or provide information to or for a regarding reporting, recordkeeping, and PSD) that establish and periodically
Federal agency. This includes the time compliance requirements. revise requirements for compliance.
Any incremental modeling costs calculating PSD increment test methods, sampling procedures, and
attributable to today’s rule do not consumption, and does not impose any business practices) that are developed or
approach the $100 million threshold additional requirements for the adopted by voluntary consensus
prescribed by UMRA. EPA has regulated community, including Indian standards bodies. The NTTAA directs
determined that this rule contains no Tribal Governments. Thus, Executive EPA to provide Congress, through OMB,
regulatory requirements that might Order 13175 does not apply to this rule. explanations when the Agency decides
significantly or uniquely affect small Today’s final rule does not not to use available and applicable
governments. This rule therefore significantly or uniquely affect the voluntary consensus standards.
contains no Federal mandates (under communities of Indian tribal
the regulatory provisions of Title II of governments. Accordingly, the This action does not involve technical
the UMRA) for State, local, or tribal requirements of section 3(b) of standards. Therefore, EPA did not
governments or the private sector. Executive Order 13175 do not apply to consider the use of any voluntary
this rule. consensus standards.
E. Executive Order 13132: Federalism
Executive Order 13132, entitled G. Executive Order 13045: Protection of J. Congressional Review Act of 1998
‘‘Federalism’’ (64 FR 43255, August 10, Children From Environmental Health
and Safety Risks The Congressional Review Act, 5
1999), requires EPA to develop an U.S.C. 801 et seq., as added by the Small
accountable process to ensure Executive Order 13045 applies to any Business Regulatory Enforcement
‘‘meaningful and timely input by State rule that EPA determines (1) to be
Fairness Act of 1996, generally provides
and local officials in the development of ‘‘economically significant’’ as defined
that before a rule may take effect, the
regulatory policies that have federalism under Executive Order 12866, and (2)
implications.’’ ‘‘Policies that have agency promulgating the rule must
the environmental health or safety risk
federalism implications’’ is defined in addressed by the rule has a submit a rule report, which includes a
the Executive Order to include disproportionate effect on children. If copy of the rule, to each House of the
regulations that have ‘‘substantial direct the regulatory action meets both the Congress and to the Comptroller General
effects on the States, on the relationship criteria, the Agency must evaluate the of the United States. EPA will submit a
between the national government and environmental health or safety effects of report containing this rule and other
the States, or on the distribution of the planned rule on children; and required information to the U.S. Senate,
power and responsibilities among the explain why the planned regulation is the U.S. House of Representatives, and
various levels of government.’’ preferable to other potentially effective the Comptroller General of the United
This final rule does not have and reasonably feasible alternatives States prior to publication of the rule in
federalism implications. It will not have considered by the Agency. the Federal Register. A Major rule
substantial direct effects on the States, This final rule is not subject to cannot take effect until 60 days after it
on the relationship between the national Executive Order 13045, entitled is published in the Federal Register.
government and the States, or on the ‘‘Protection of Children from This action is not a ‘‘major rule’’ as
distribution of power and Environmental Health Risks and Safety defined by 5 U.S.C. 804(2), and will be
responsibilities among the various Risks’’ (62 FR 19885, April 23, 1997) effective 30 days from the publication
levels of government, as specified in because it does not impose an date of this notice.
Executive Order 13132. This rule does economically significant regulatory
not create a mandate on State, local or action as defined by Executive Order List of Subjects in 40 CFR Part 51
tribal governments. The rule does not 12866 and the action does not involve
impose any enforceable duties on these decisions on environmental health or Environmental protection,
entities (see D. Unfunded Mandates safety risks that may disproportionately Administrative practice and procedure,
Reform Act of 1995, above). The rule affect children. Air pollution control, Carbon monoxide,
would add better, more accurate Intergovernmental relations, Nitrogen
techniques for air dispersion modeling H. Executive Order 13211: Actions That oxides, Ozone, Particulate Matter,
analyses and does not impose any Significantly Affect Energy Supply, Reporting and recordkeeping
additional requirements for any of the Distribution, or Use requirements, Sulfur oxides.
affected parties covered under Executive This rule is not subject to Executive Dated: October 21, 2005.
Order 13132. Thus, Executive Order Order 13211, ‘‘Actions Concerning
Stephen L. Johnson,
13132 does not apply to this rule. Regulations That Significantly Affect
Energy Supply, Distribution, or Use’’ (66 Administrator.
F. Executive Order 13175: Consultation
FR 28355 (May 22, 2001)) because it is ■ Part 51, chapter I, title 40 of the Code
and Coordination With Indian Tribal
not a significant regulatory action under of Federal Regulations is amended as
Governments
Executive Order 12866. follows:
Executive Order 13175, entitled
‘‘Consultation and Coordination with I. National Technology Transfer and
Advancement Act of 1995 PART 51—REQUIREMENTS FOR
Indian Tribal Governments’’ (65 FR
67249, November 9, 2000), requires EPA Section 12(d) of the National PREPARATION, ADOPTION, AND
to develop an accountable process to Technology Transfer and Advancement SUBMITTAL OF IMPLEMENTATION
ensure ‘‘meaningful and timely input by Act of 1995 (‘‘NTTAA’’), Public Law PLANS
tribal officials in the development of 104–113, section 12(d) (15 U.S.C. 272
regulatory policies that have tribal note) directs EPA to use voluntary ■ 1. The authority citation for part 51
implications.’’ This final rule does not consensus standards in its regulatory continues to read as follows:
have tribal implications, as specified in activities unless to do so would be Authority: 23 U.S.C. 100; 42 U.S.C. 7401–
Executive Order 13175. As stated above inconsistent with applicable law or 7671q.
(see D. Unfunded Mandates Reform Act otherwise impractical. Voluntary
of 1995, above), the rule does not consensus standards are technical ■ 2. Appendix W to Part 51 revised to
impose any new requirements for standards (e.g., materials specifications, read as follows:
Appendix W to Part 51—Guideline on 4.0 Stationary-Source Models 10.2.3 Emission Limits

Air Quality Models 4.1 Discussion 11.0 Bibliography
4.2 Recommendations
Preface 12.0 References
4.2.1 Screening Techniques
a. Industry and control agencies have long 4.2.1.1 Simple Terrain Appendix A to Appendix W of 40 CFR Part
expressed a need for consistency in the 4.2.1.2 Complex Terrain 51—Summaries of Preferred Air Quality
application of air quality models for 4.2.2 Refined Analytical Techniques Models
regulatory purposes. In the 1977 Clean Air
5.0 Models for Ozone, Particulate Matter,
Act, Congress mandated such consistency LIST OF TABLES
Carbon Monoxide, Nitrogen Dioxide, and
and encouraged the standardization of model
applications. The Guideline on Air Quality Lead
5.1 Discussion Table No. Title
Models (hereafter, Guideline) was first
published in April 1978 to satisfy these 5.2 Recommendations
5.2.1 Models for Ozone 4–1a ............ Neutral/Stable Meteorological
requirements by specifying models and
5.2.2 Models for Particulate Matter Matrix for CTSCREEN.
providing guidance for their use. The
5.2.2.1 PM–2.5 4–1b ............ Unstable/Convective Meteoro-
Guideline provides a common basis for
estimating the air quality concentrations of 5.2.2.2 PM–10 logical Matrix for
criteria pollutants used in assessing control 5.2.3 Models for Carbon Monoxide CTSCREEN.
strategies and developing emission limits. 5.2.4 Models for Nitrogen Dioxide 8–1 .............. Model Emission Input Data for
b. The continuing development of new air (Annual Average) Point Sources.
quality models in response to regulatory 5.2.5 Models for Lead 8–2 .............. Point Source Model Emission
requirements and the expanded requirements Input Data for NAAQS Com-
6.0 Other Model Requirements pliance in PSD Demonstra-
for models to cover even more complex
problems have emphasized the need for 6.1 Discussion tions.
periodic review and update of guidance on 6.2 Recommendations 8–3 .............. Averaging Times for Site Spe-
these techniques. Historically, three primary 6.2.1 Visibility cific Wind and Turbulence
activities have provided direct input to 6.2.2 Good Engineering Practice Stack Measurements.
revisions of the Guideline. The first is a series Height
of annual EPA workshops conducted for the 6.2.3 Long Range Transport (LRT) (i.e., 1.0 Introduction
purpose of ensuring consistency and beyond 50 km)
6.2.4 Modeling Guidance for Other a. The Guideline recommends air quality
providing clarification in the application of modeling techniques that should be applied
models. The second activity was the Governmental Programs
to State Implementation Plan (SIP) revisions
solicitation and review of new models from 7.0 General Modeling Considerations for existing sources and to new source
the technical and user community. In the 7.1 Discussion reviews (NSR), including prevention of
March 27, 1980 Federal Register, a procedure 7.2 Recommendations significant deterioration (PSD).1 2 3
was outlined for the submittal to EPA of 7.2.1 Design Concentrations Applicable only to criteria air pollutants, it
privately developed models. After extensive 7.2.2 Critical Receptor Sites is intended for use by EPA Regional Offices
evaluation and scientific review, these 7.2.3 Dispersion Coefficients in judging the adequacy of modeling analyses
models, as well as those made available by 7.2.4 Stability Categories performed by EPA, State and local agencies
EPA, have been considered for recognition in 7.2.5 Plume Rise and by industry. The guidance is appropriate
the Guideline. The third activity is the 7.2.6 Chemical Transformation for use by other Federal agencies and by State
extensive on-going research efforts by EPA 7.2.7 Gravitational Settling and agencies with air quality and land
and others in air quality and meteorological Deposition management responsibilities. The Guideline
modeling. 7.2.8 Complex Winds serves to identify, for all interested parties,
c. Based primarily on these three activities, 7.2.9 Calibration of Models
new sections and topics have been included those techniques and data bases EPA
as needed. EPA does not make changes to the 8.0 Model Input Data considers acceptable. The Guideline is not
guidance on a predetermined schedule, but 8.1 Source Data intended to be a compendium of modeling
rather on an as-needed basis. EPA believes 8.1.1 Discussion techniques. Rather, it should serve as a
that revisions of the Guideline should be 8.1.2 Recommendations common measure of acceptable technical
timely and responsive to user needs and 8.2 Background Concentrations analysis when supported by sound scientific
should involve public participation to the 8.2.1 Discussion judgment.
greatest possible extent. All future changes to 8.2.2 Recommendations (Isolated Single b. Due to limitations in the spatial and
the guidance will be proposed and finalized Source) temporal coverage of air quality
in the Federal Register. Information on the 8.2.3 Recommendations (Multi-Source measurements, monitoring data normally are
current status of modeling guidance can Areas) not sufficient as the sole basis for
always be obtained from EPA’s Regional 8.3 Meteorological Input Data demonstrating the adequacy of emission
Offices. 8.3.1 Length of Record of Meteorological limits for existing sources. Also, the impacts
Data of new sources that do not yet exist can only
Table of Contents 8.3.2 National Weather Service Data be determined through modeling. Thus,
List of Tables 8.3.3 Site Specific Data models, while uniquely filling one program
8.3.4 Treatment of Near-calms and Calms need, have become a primary analytical tool
1.0 Introduction in most air quality assessments. Air quality
9.0 Accuracy and Uncertainty of Models
2.0 Overview of Model Use measurements can be used in a
9.1 Discussion complementary manner to dispersion
2.1 Suitability of Models 9.1.1 Overview of Model Uncertainty models, with due regard for the strengths and
2.2 Levels of Sophistication of Models 9.1.2 Studies of Model Accuracy weaknesses of both analysis techniques.
2.3 Availability of Models 9.1.3 Use of Uncertainty in Decision- Measurements are particularly useful in
3.0 Recommended Air Quality Models Making assessing the accuracy of model estimates.
9.1.4 Evaluation of Models
3.1 Preferred Modeling Techniques The use of air quality measurements alone
9.2 Recommendations
3.1.1 Discussion however could be preferable, as detailed in
3.1.2 Recommendations 10.0 Regulatory Application of Models a later section of this document, when
3.2 Use of Alternative Models 10.1 Discussion models are found to be unacceptable and
3.2.1 Discussion 10.2 Recommendations monitoring data with sufficient spatial and
3.2.2 Recommendations 10.2.1 Analysis Requirements temporal coverage are available.
3.3 Availability of Supplementary Modeling 10.2.2 Use of Measured Data in Lieu of c. It would be advantageous to categorize
Guidance Model Estimates the various regulatory programs and to apply
a designated model to each proposed source promote the use of more accurate air quality inventory, meteorological data, and air
needing analysis under a given program. models and data bases. The workshops serve quality data. Appropriate data should be
However, the diversity of the nation’s to provide further explanations of Guideline available before any attempt is made to apply
topography and climate, and variations in requirements to the Regional Offices and a model. A model that requires detailed,
source configurations and operating workshop reports are issued with this precise, input data should not be used when
characteristics dictate against a strict clarifying information. In addition, findings such data are unavailable. However,
modeling ‘‘cookbook’’. There is no one model from ongoing research programs, new model assuming the data are adequate, the greater
capable of properly addressing all development, or results from model the detail with which a model considers the
conceivable situations even within a broad evaluations and applications are spatial and temporal variations in emissions
category such as point sources. continuously evaluated. Based on this and meteorological conditions, the greater
Meteorological phenomena associated with information changes in the guidance may be the ability to evaluate the source impact and
threats to air quality standards are rarely indicated. to distinguish the effects of various control
amenable to a single mathematical treatment; g. All changes to the Guideline must follow strategies.
thus, case-by-case analysis and judgment are rulemaking requirements since the Guideline b. Air quality models have been applied
frequently required. As modeling efforts is codified in Appendix W of Part 51. EPA with the most accuracy, or the least degree
become more complex, it is increasingly will promulgate proposed and final rules in of uncertainty, to simulations of long term
important that they be directed by highly the Federal Register to amend this averages in areas with relatively simple
competent individuals with a broad range of Appendix. Ample opportunity for public topography. Areas subject to major
experience and knowledge in air quality comment will be provided for each proposed topographic influences experience
meteorology. Further, they should be change and public hearings scheduled if meteorological complexities that are
coordinated closely with specialists in requested. extremely difficult to simulate. Although
emissions characteristics, air monitoring and h. A wide range of topics on modeling and models are available for such circumstances,
data processing. The judgment of data bases are discussed in the Guideline. they are frequently site specific and resource
experienced meteorologists and analysts is Section 2 gives an overview of models and intensive. In the absence of a model capable
essential. their appropriate use. Section 3 provides of simulating such complexities, only a
d. The model that most accurately specific guidance on the use of ‘‘preferred’’ preliminary approximation may be feasible
estimates concentrations in the area of air quality models and on the selection of until such time as better models and data
interest is always sought. However, it is clear alternative techniques. Sections 4 through 7 bases become available.
from the needs expressed by the States and provide recommendations on modeling c. Models are highly specialized tools.
EPA Regional Offices, by many industries techniques for application to simple-terrain Competent and experienced personnel are an
and trade associations, and also by the stationary source problems, complex terrain essential prerequisite to the successful
deliberations of Congress, that consistency in problems, and mobile source problems. application of simulation models. The need
the selection and application of models and Specific modeling requirements for selected for specialists is critical when the more
data bases should also be sought, even in regulatory issues are also addressed. Section sophisticated models are used or the area
case-by-case analyses. Consistency ensures 8 discusses issues common to many being investigated has complicated
that air quality control agencies and the modeling analyses, including acceptable meteorological or topographic features. A
general public have a common basis for model components. Section 9 makes model applied improperly, or with
estimating pollutant concentrations, recommendations for data inputs to models inappropriate data, can lead to serious
assessing control strategies and specifying including source, meteorological and misjudgements regarding the source impact
emission limits. Such consistency is not, background air quality data. Section 10 or the effectiveness of a control strategy.
however, promoted at the expense of model covers the uncertainty in model estimates d. The resource demands generated by use
and data base accuracy. The Guideline and how that information can be useful to the of air quality models vary widely depending
provides a consistent basis for selection of regulatory decision-maker. The last chapter on the specific application. The resources
the most accurate models and data bases for summarizes how estimates and required depend on the nature of the model
use in air quality assessments. measurements of air quality are used in and its complexity, the detail of the data
e. Recommendations are made in the assessing source impact and in evaluating base, the difficulty of the application, and the
Guideline concerning air quality models, data control strategies. amount and level of expertise required. The
bases, requirements for concentration i. Appendix W to 40 CFR Part 51 itself costs of manpower and computational
estimates, the use of measured data in lieu contains an appendix: Appendix A. Thus, facilities may also be important factors in the
of model estimates, and model evaluation when reference is made to ‘‘Appendix A’’ in selection and use of a model for a specific
procedures. Models are identified for some this document, it refers to Appendix A to analysis. However, it should be recognized
specific applications. The guidance provided Appendix W to 40 CFR Part 51. Appendix A that under some sets of physical
here should be followed in air quality contains summaries of refined air quality circumstances and accuracy requirements, no
analyses relative to State Implementation models that are ‘‘preferred’’ for specific present model may be appropriate. Thus,
Plans and in supporting analyses required by applications; both EPA models and models consideration of these factors should lead to
EPA, State and local agency air programs. developed by others are included. selection of an appropriate model.
EPA may approve the use of another
technique that can be demonstrated to be 2.0 Overview of Model Use 2.2 Levels of Sophistication of Models
more appropriate than those recommended a. Before attempting to implement the a. There are two levels of sophistication of
in this guide. This is discussed at greater guidance contained in this document, the models. The first level consists of relatively
length in Section 3. In all cases, the model reader should be aware of certain general simple estimation techniques that generally
applied to a given situation should be the one information concerning air quality models use preset, worst-case meteorological
that provides the most accurate and their use. Such information is provided conditions to provide conservative estimates
representation of atmospheric transport, in this section. of the air quality impact of a specific source,
dispersion, and chemical transformations in or source category. These are called screening
the area of interest. However, to ensure 2.1 Suitability of Models techniques or screening models. The purpose
consistency, deviations from this guide a. The extent to which a specific air quality of such techniques is to eliminate the need
should be carefully documented and fully model is suitable for the evaluation of source of more detailed modeling for those sources
supported. impact depends upon several factors. These that clearly will not cause or contribute to
f. From time to time situations arise include: (1) The meteorological and ambient concentrations in excess of either
requiring clarification of the intent of the topographic complexities of the area; (2) the the National Ambient Air Quality Standards
guidance on a specific topic. Periodic level of detail and accuracy needed for the (NAAQS) 4 or the allowable prevention of
workshops are held with the headquarters, analysis; (3) the technical competence of significant deterioration (PSD) concentration
Regional Office, State, and local agency those undertaking such simulation modeling; increments.2 3 If a screening technique
modeling representatives to ensure (4) the resources available; and (5) the detail indicates that the concentration contributed
consistency in modeling guidance and to and accuracy of the data base, i.e., emissions by the source exceeds the PSD increment or
the increment remaining to just meet the appropriate reviewing authority, you should i. The model must be written in a common
NAAQS, then the second level of more contact the Regional modeling contact programming language, and the executable(s)
sophisticated models should be applied. (http://www.epa.gov/scram001/ must run on a common computer platform.
b. The second level consists of those tt28.htm#regionalmodelingcontacts) in the ii. The model must be documented in a
analytical techniques that provide more appropriate EPA Regional Office, whose user’s guide which identifies the
detailed treatment of physical and chemical jurisdiction generally includes the physical mathematics of the model, data requirements
atmospheric processes, require more detailed location of the source in question and its and program operating characteristics at a
and precise input data, and provide more expected impacts. level of detail comparable to that available
specialized concentration estimates. As a c. In all regulatory analyses, especially if for other recommended models in Appendix
result they provide a more refined and, at other-than-preferred models are selected for A.
least theoretically, a more accurate estimate use, early discussions among Regional Office iii. The model must be accompanied by a
of source impact and the effectiveness of staff, State and local control agencies, complete test data set including input
control strategies. These are referred to as industry representatives, and where parameters and output results. The test data
refined models. appropriate, the Federal Land Manager, are must be packaged with the model in
c. The use of screening techniques invaluable and are encouraged. Agreement computer-readable form.
followed, as appropriate, by a more refined on the data base(s) to be used, modeling iv. The model must be useful to typical
analysis is always desirable. However there techniques to be applied and the overall users, e.g., State air pollution control
are situations where the screening techniques technical approach, prior to the actual agencies, for specific air quality control
are practically and technically the only analyses, helps avoid misunderstandings problems. Such users should be able to
viable option for estimating source impact. In concerning the final results and may reduce operate the computer program(s) from
such cases, an attempt should be made to the later need for additional analyses. The available documentation.
acquire or improve the necessary data bases use of an air quality analysis checklist, such v. The model documentation must include
and to develop appropriate analytical as is posted on EPA’s Internet SCRAM Web a comparison with air quality data (and/or
techniques. site (subsection 2.3), and the preparation of tracer measurements) or with other well-
a written protocol help to keep established analytical techniques.
2.3 Availability of Models misunderstandings at a minimum. vi. The developer must be willing to make
a. For most of the screening and refined d. It should not be construed that the the model and source code available to users
models discussed in the Guideline, codes, preferred models identified here are to be at reasonable cost or make them available for
associated documentation and other useful permanently used to the exclusion of all public access through the Internet or
information are available for download from others or that they are the only models National Technical Information Service: The
EPA’s Support Center for Regulatory Air available for relating emissions to air quality. model and its code cannot be proprietary.
Modeling (SCRAM) Internet Web site at The model that most accurately estimates c. The evaluation process includes a
http://www.epa.gov/scram001. A list of concentrations in the area of interest is determination of technical merit, in
alternate models that can be used with case- always sought. However, designation of accordance with the above six items
by-case justification (subsection 3.2) and an specific models is needed to promote including the practicality of the model for
example air quality analysis checklist are consistency in model selection and use in ongoing regulatory programs. Each
also posted on this Web site. This is a site application. model will also be subjected to a
with which modelers should become e. The 1980 solicitation of new or different performance evaluation for an appropriate
familiar. models from the technical community 6 and data base and to a peer scientific review.
the program whereby these models were Models for wide use (not just an isolated
3.0 Recommended Air Quality Models evaluated, established a means by which new case) that are found to perform better will be
a. This section recommends the approach models are identified, reviewed and made proposed for inclusion as preferred models in
to be taken in determining refined modeling available in the Guideline. There is a pressing future Guideline revisions.
techniques for use in regulatory air quality need for the development of models for a d. No further evaluation of a preferred
programs. The status of models developed by wide range of regulatory applications. model is required for a particular application
EPA, as well as those submitted to EPA for Refined models that more realistically if the EPA recommendations for regulatory
review and possible inclusion in this simulate the physical and chemical process use specified for the model in the Guideline
guidance, is discussed. The section also in the atmosphere and that more reliably are followed. Alternative models to those
addresses the selection of models for estimate pollutant concentrations are needed. listed in Appendix A should generally be
individual cases and provides compared with measured air quality data
3.1 Preferred Modeling Techniques when they are used for regulatory
recommendations for situations where the
preferred models are not applicable. Two 3.1.1 Discussion applications consistent with
additional sources of modeling guidance are a. EPA has developed models suitable for recommendations in subsection 3.2.
the Model Clearinghouse 5 and periodic regulatory application. Other models have 3.1.2 Recommendations
Regional/State/Local Modelers workshops. been submitted by private developers for a. Appendix A identifies refined models
b. In this guidance, when approval is possible inclusion in the Guideline. Refined that are preferred for use in regulatory
required for a particular modeling technique models which are preferred and applications. If a model is required for a
or analytical procedure, we often refer to the recommended by EPA have undergone particular application, the user should select
‘‘appropriate reviewing authority’’. In some evaluation exercises 7 8 9 10 that include a model from that appendix. These models
EPA regions, authority for NSR and PSD statistical measures of model performance in may be used without a formal demonstration
permitting and related activities has been comparison with measured air quality data as of applicability as long as they are used as
delegated to State and even local agencies. In suggested by the American Meteorological indicated in each model summary of
these cases, such agencies are Society 11 and, where possible, peer scientific Appendix A. Further recommendations for
‘‘representatives’’ of the respective regions. reviews.12 13 14 the application of these models to specific
Even in these circumstances, the Regional b. When a single model is found to perform source problems are found in subsequent
Office retains the ultimate authority in better than others, it is recommended for sections of the Guideline.
decisions and approvals. Therefore, as application as a preferred model and listed b. If changes are made to a preferred model
discussed above and depending on the in Appendix A. If no one model is found to without affecting the concentration estimates,
circumstances, the appropriate reviewing clearly perform better through the evaluation the preferred status of the model is
authority may be the Regional Office, Federal exercise, then the preferred model listed in unchanged. Examples of modifications that
Land Manager(s), State agency(ies), or Appendix A may be selected on the basis of do not affect concentrations are those made
perhaps local agency(ies). In cases where other factors such as past use, public to enable use of a different computer
review and approval comes solely from the familiarity, cost or resource requirements, platform or those that affect only the format
Regional Office (sometimes stated as and availability. Accordingly, dispersion or averaging time of the model results.
‘‘Regional Administrator’’), this will be models listed in Appendix A meet these However, when any changes are made, the
stipulated. If there is any question as to the conditions: Regional Administrator should require a test
case example to demonstrate that the better for the given application than a workshops with headquarters, Regional
concentration estimates are not affected. comparable model in Appendix A; or (3) if Office, State, and local agency modeling
c. A preferred model should be operated the preferred model is less appropriate for representatives.
with the options listed in Appendix A as the specific application, or there is no b. The Regional Office should always be
‘‘Recommendations for Regulatory Use.’’ If preferred model. Any one of these three consulted for information and guidance
other options are exercised, the model is no separate conditions may make use of an concerning modeling methods and
longer ‘‘preferred.’’ Any other modification to alternative model acceptable. Some known interpretations of modeling guidance, and to
a preferred model that would result in a alternative models that are applicable for ensure that the air quality model user has
change in the concentration estimates selected situations are listed on EPA’s available the latest most up-to-date policy
likewise alters its status as a preferred model. SCRAM Internet Web site (subsection 2.3). and procedures. As appropriate, the Regional
Use of the model must then be justified on However, inclusion there does not confer any Office may request assistance from the Model
a case-by-case basis. unique status relative to other alternative Clearinghouse after an initial evaluation and
models that are being or will be developed decision has been reached concerning the
3.2 Use of Alternative Models in the future. application of a model, analytical technique
3.2.1 Discussion c. Equivalency, condition (1) in paragraph or data base in a particular regulatory action.
a. Selection of the best techniques for each (b) of this subsection, is established by
4.0 Traditional Stationary Source Models
individual air quality analysis is always demonstrating that the maximum or highest,
encouraged, but the selection should be done second highest concentrations are within 2 4.1 Discussion
in a consistent manner. A simple listing of percent of the estimates obtained from the a. Guidance in this section applies to
models in this Guideline cannot alone preferred model. The option to show modeling analyses for which the
achieve that consistency nor can it equivalency is intended as a simple predominant meteorological conditions that
necessarily provide the best model for all demonstration of acceptability for an control the design concentration are steady
possible situations. An EPA reference 15 alternative model that is so nearly identical state and for which the transport distances
provides a statistical technique for evaluating (or contains options that can make it are nominally 50km or less. The models
model performance for predicting peak identical) to a preferred model that it can be recommended in this section are generally
concentration values, as might be observed at treated for practical purposes as the preferred used in the air quality impact analysis of
individual monitoring locations. This model. Two percent was selected as the basis stationary sources for most criteria
protocol is available to assist in developing for equivalency since it is a rough pollutants. The averaging time of the
a consistent approach when justifying the use approximation of the fraction that PSD Class concentration estimates produced by these
of other-than-preferred modeling techniques I increments are of the NAAQS for SO2, i.e., models ranges from 1 hour to an annual
recommended in the Guideline. The the difference in concentrations that is average.
procedures in this protocol provide a general judged to be significant. However, b. Simple terrain, as used here, is
framework for objective decision-making on notwithstanding this demonstration, models considered to be an area where terrain
the acceptability of an alternative model for that are not equivalent may be used when features are all lower in elevation than the
a given regulatory application. These one of the two other conditions described in top of the stack of the source(s) in question.
objective procedures may be used for paragraphs (d) and (e) of this subsection are Complex terrain is defined as terrain
conducting both the technical evaluation of satisfied. exceeding the height of the stack being
the model and the field test or performance d. For condition (2) in paragraph (b) of this modeled.
evaluation. An ASTM reference 16 provides a subsection, established procedures and c. In the early 1980s, model evaluation
general philosophy for developing and techniques 15 16 for determining the exercises were conducted to determine the
implementing advanced statistical acceptability of a model for an individual ‘‘best, most appropriate point source model’’
evaluations of atmospheric dispersion case based on superior performance should for use in simple terrain.12 No one model was
models, and provides an example statistical be followed, as appropriate. Preparation and found to be clearly superior and, based on
technique to illustrate the application of this implementation of an evaluation protocol past use, public familiarity, and availability,
philosophy. which is acceptable to both control agencies ISC (predecessor to ISC3 17) became the
b. This section discusses the use of and regulated industry is an important recommended model for a wide range of
alternate modeling techniques and defines element in such an evaluation. regulatory applications. Other refined models
three situations when alternative models may e. Finally, for condition (3) in paragraph (b) which also employed the same basic
be used. of this subsection, an alternative refined Gaussian kernel as in ISC, i.e., BLP, CALINE3
model may be used provided that: and OCD, were developed for specialized
3.2.2 Recommendations
i. The model has received a scientific peer applications (Appendix A). Performance
a. Determination of acceptability of a review; evaluations were also made for these models,
model is a Regional Office responsibility. ii. The model can be demonstrated to be which are identified below.
Where the Regional Administrator finds that applicable to the problem on a theoretical d. Encouraged by the development of
an alternative model is more appropriate basis; pragmatic methods for better characterization
than a preferred model, that model may be iii. The data bases which are necessary to of plume dispersion 18 19 20 21 the AMS/EPA
used subject to the recommendations of this perform the analysis are available and Regulatory Model Improvement Committee
subsection. This finding will normally result adequate; (AERMIC) developed AERMOD.22 AERMOD
from a determination that (1) a preferred air iv. Appropriate performance evaluations of employs best state-of-practice
quality model is not appropriate for the the model have shown that the model is not parameterizations for characterizing the
particular application; or (2) a more biased toward underestimates; and meteorological influences and dispersion.
appropriate model or analytical procedure is v. A protocol on methods and procedures The model utilizes a probability density
available and applicable. to be followed has been established. function (pdf) and the superposition of
b. An alternative model should be several Gaussian plumes to characterize the
evaluated from both a theoretical and a 3.3 Availability of Supplementary Modeling distinctly non-Gaussian nature of the vertical
performance perspective before it is selected Guidance pollutant distribution for elevated plumes
for use. There are three separate conditions a. The Regional Administrator has the during convective conditions; otherwise the
under which such a model may normally be authority to select models that are distribution is Gaussian. Also, nighttime
approved for use: (1) If a demonstration can appropriate for use in a given situation. urban boundary layers (and plumes within
be made that the model produces However, there is a need for assistance and them) have the turbulence enhanced by
concentration estimates equivalent to the guidance in the selection process so that AERMOD to simulate the influence of the
estimates obtained using a preferred model; fairness and consistency in modeling urban heat island. AERMOD has been
(2) if a statistical performance evaluation has decisions is fostered among the various evaluated using a variety of data sets and has
been conducted using measured air quality Regional Offices and the States. To satisfy been found to perform better than ISC3 for
data and the results of that evaluation that need, EPA established the Model many applications, and as well or better than
indicate the alternative model performs Clearinghouse 5 and also holds periodic CTDMPLUS for several complex terrain data
sets (Section A.1; subsection n). The current height, and wind directions for both neutral/ the terrain both as a single feature and as
version of AERMOD has been modified to stable conditions and unstable convective multiple hills to determine design
include an algorithm for dry and wet conditions. Table 4–1 contains the matrix of concentrations.
deposition for both gases and particles. Note meteorological variables that is used for each d. Other screening techniques 17 25 29 may
that when deposition is invoked, mass in the CTSCREEN analysis. There are 96 be acceptable for complex terrain cases
plume is depleted. Availability of this combinations, including exceptions, for each where established procedures are used. The
version is described in Section A.1, and is wind direction for the neutral/stable case, user is encouraged to confer with the
subject to applicable guidance published in and 108 combinations for the unstable case. appropriate reviewing authority if any
the Guideline. The specification of wind direction, however, unresolvable problems are encountered, e.g.,
e. A new building downwash algorithm 23 is handled internally, based on the source applicability, meteorological data, receptor
was developed and tested within AERMOD. and terrain geometry. Although CTSCREEN siting, or terrain contour processing issues.
The PRIME algorithm has been evaluated is designed to address a single source
using a variety of data sets and has been scenario, there are a number of options that 4.2.2 Refined Analytical Techniques
found to perform better than the downwash can be selected on a case-by-case basis to a. A brief description of each preferred
algorithm that is in ISC3, and has been address multi-source situations. However, model for refined applications is found in
shown to perform acceptably in tests within the appropriate reviewing authority should Appendix A. Also listed in that appendix are
AERMOD (Section A.1; subsection n). be consulted, and concurrence obtained, on availability, the model input requirements,
the protocol for modeling multiple sources the standard options that should be selected
4.2 Recommendations
with CTSCREEN to ensure that the worst case when running the program, and output
4.2.1 Screening Techniques is identified and assessed. The maximum options.
4.2.1.1 Simple Terrain concentration output from CTSCREEN b. For a wide range of regulatory
represents a worst-case 1-hour concentration. applications in all types of terrain, the
a. Where a preliminary or conservative Time-scaling factors of 0.7 for 3-hour, 0.15
estimate is desired, point source screening recommended model is AERMOD. This
for 24-hour and 0.03 for annual concentration recommendation is based on extensive
techniques are an acceptable approach to air averages are applied internally by
quality analyses. EPA has published developmental and performance evaluation
CTSCREEN to the highest 1-hour
guidance for screening procedures.24 25 (Section A.1; subsection n). Differentiation of
concentration calculated by the model.
b. All screening procedures should be simple versus complex terrain is unnecessary
b. Placement of receptors requires very
adjusted to the site and problem at hand. with AERMOD. In complex terrain, AERMOD
careful attention when modeling in complex
Close attention should be paid to whether the terrain. Often the highest concentrations are employs the well-known dividing-streamline
area should be classified urban or rural in predicted to occur under very stable concept in a simplified simulation of the
accordance with Section 7.2.3. The conditions, when the plume is near, or effects of plume-terrain interactions.
climatology of the area should be studied to impinges on, the terrain. The plume under c. If aerodynamic building downwash is
help define the worst-case meteorological such conditions may be quite narrow in the important for the modeling analysis, e.g.,
conditions. Agreement should be reached vertical, so that even relatively small changes paragraph 6.2.2(b), then the recommended
between the model user and the appropriate in a receptor’s location may substantially model is AERMOD. The state-of-the-science
reviewing authority on the choice of the affect the predicted concentration. Receptors for modeling atmospheric deposition is
screening model for each analysis, and on the within about a kilometer of the source may evolving and the best techniques are
input data as well as the ultimate use of the be even more sensitive to location. Thus, a currently being assessed and their results are
results. dense array of receptors may be required in being compared with observations.
4.2.1.2 Complex Terrain some cases. In order to avoid excessively Consequently, while deposition treatment is
a. CTSCREEN 26 can be used to obtain large computer runs due to such a large array available in AERMOD, the approach taken for
conservative, yet realistic, worst-case of receptors, it is often desirable to model the any purpose should be coordinated with the
estimates for receptors located on terrain area twice. The first model run would use a appropriate reviewing authority. Line sources
above stack height. CTSCREEN accounts for moderate number of receptors carefully can be simulated with AERMOD if point or
the three-dimensional nature of plume and located over the area of interest. The second volume sources are appropriately combined.
terrain interaction and requires detailed model run would use a more dense array of If buoyant plume rise from line sources is
terrain data representative of the modeling receptors in areas showing potential for high important for the modeling analysis, the
domain. The model description and user’s concentrations, as indicated by the results of recommended model is BLP. For other
instructions are contained in the user’s the first model run. special modeling applications, CALINE3 (or
guide.26 The terrain data must be digitized in c. As mentioned above, digitized contour CAL3QHCR on a case-by-case basis), OCD,
the same manner as for CTDMPLUS and a data must be preprocessed 27 to provide hill and EDMS are available as described in
terrain processor is available.27 A discussion shape parameters in suitable input format. Sections 5 and 6.
of the model’s performance characteristics is The user then supplies receptors either d. If the modeling application involves a
provided in a technical paper.28 CTSCREEN through an interactive program that is part of well defined hill or ridge and a detailed
is designed to execute a fixed matrix of the model or directly, by using a text editor; dispersion analysis of the spatial pattern of
meteorological values for wind speed (u), using both methods to select receptors will plume impacts is of interest, CTDMPLUS,
standard deviation of horizontal and vertical generally be necessary to assure that the listed in Appendix A, is available.
wind speeds (sv, sw), vertical potential maximum concentrations are estimated by CDTMPLUS provides greater resolution of
temperature gradient (dq/dz), friction either model. In cases where a terrain feature concentrations about the contour of the hill
velocity (u*), Monin-Obukhov length (L), may ‘‘appear to the plume’’ as smaller, feature than does AERMOD through a
mixing height (zi) as a function of terrain multiple hills, it may be necessary to model different plume-terrain interaction algorithm.
TABLE 4–1A.—NEUTRAL/STABLE METEOROLOGICAL MATRIX FOR CTSCREEN
Variable Specific values
U (m/s) ............................................................................................. 1.0 2.0 3.0 4.0 5.0

sv (m/s) ............................................................................................ 0.3 0.75
sw (m/s) ............................................................................................ 0.08 0.15 0.30 0.75
Dq/Dz (K/m) ...................................................................................... 0.01 0.02 0.035
WD ................................................................................................... (Wind direction is optimized internally for each meteorological combination.)
Exceptions: (2) If sw = 0.75 m/s and U ≥ 3.0 m/s, then (4) sw ≤ sv

(1) If U ≤ 2 m/s and sv ≤ 0.3 m/s, then include Dq/Dz is limited to ≤ 0.01 K/m.
sw = 0.04 m/s. (3) If U ≥ 4 m/s, then sw ≥ 0.15 m/s.
TABLE 4–1B.—UNSTABLE/CONVECTIVE METEOROLOGICAL MATRIX FOR CTSCREEN
Variable Specific values
U (m/s) ............................................................................................... 1.0 2.0 3.0 4.0 5.0

U* (m/s) .............................................................................................. 0.1 0.3 0.5
L (m) ................................................................................................... ¥10 ¥50 ¥90
Dq/Dz (K/m) ........................................................................................ 0.030 (potential temperature gradient above Zi)
Zi (m) .................................................................................................. 0.5h 1.0h 1.5h (h = terrain height)
5.0 Models for Ozone, Particulate Matter, transport and dispersion of ozone and its geographical coverage for a modeling
Carbon Monoxide, Nitrogen Dioxide, and precursors. Other approaches, such as application.33
Lead Lagrangian or observational models may be g. The NAAQS for PM–10 was
used to guide choice of appropriate strategies promulgated in July 1987 (40 CFR 50.6). A
5.1 Discussion to consider with a photochemical grid model. SIP development guide 34 is available to
a. This section identifies modeling These other approaches may be sufficient to assist in PM–10 analyses and control strategy
approaches or models appropriate for address ozone in an area where observed development. EPA promulgated regulations
addressing ozone (O3) a, carbon monoxide concentrations are near the NAAQS or only for PSD increments measured as PM–10 in a
(CO), nitrogen dioxide (NO2), particulates slightly above it. Such a decision needs to be notice published on June 3, 1993 (40 CFR
(PM–2.5 a and PM–10), and lead. These made on a case-by-case basis in concert with 51.166(c)). As an aid to assessing the impact
pollutants are often associated with the Regional Office. on ambient air quality of particulate matter
emissions from numerous sources. Generally, d. A control agency with jurisdiction over generated from prescribed burning activities,
mobile sources contribute significantly to one or more areas with significant ozone a reference 35 is available.
emissions of these pollutants or their problems should review available ambient air h. Models for assessing the impacts of
precursors. For cases where it is of interest quality data to assess whether the problem is particulate matter may involve dispersion
to estimate concentrations of CO or NO2 near likely to be significantly impacted by models or receptor models, or a combination
a single or small group of stationary sources, regional transport.32 Choice of a modeling (depending on the circumstances). Receptor
refer to Section 4. (Modeling approaches for approach depends on the outcome of this models focus on the behavior of the ambient
SO2 are discussed in Section 4.) review. In cases where transport is environment at the point of impact as
b. Several of the pollutants mentioned in considered significant, use of a nested opposed to source-oriented dispersion
the preceding paragraph are closely related to regional model may be the preferred models, which focus on the transport,
each other in that they share common approach. If the observed problem is believed diffusion, and transformation that begin at
sources of emissions and/or are subject to to be primarily of local origin, use of a model the source and continue to the receptor site.
chemical transformations of similar with a single horizontal grid resolution and Receptor models attempt to identify and
precursors.30 31 For example, strategies geographical coverage that is less than that of apportion sources by relating known sample
designed to reduce ozone could have an a regional model may suffice. compositions at receptors to measured or
effect on the secondary component of PM–2.5 e. The fine particulate matter NAAQS, inferred compositions of source emissions.
and vice versa. Thus, it makes sense to use promulgated on July 18, 1997, includes When complete and accurate emission
models which take into account the chemical particles with an aerodynamic diameter inventories or meteorological
coupling between O3 and PM–2.5, when nominally less than or equal to 2.5 characterization are unavailable, or unknown
feasible. This should promote consistency micrometers (PM–2.5). Models for PM–2.5 pollutant sources exist, receptor modeling
among methods used to evaluate strategies are needed to assess adequacy of a proposed may be necessary.
for reducing different pollutants as well as strategy for meeting annual and/or 24-hour i. Models for assessing the impact of CO
consistency among the strategies themselves. NAAQS for PM–2.5. PM–2.5 is a mixture emissions are needed for a number of
Regulatory requirements for the different consisting of several diverse components. different purposes. Examples include
pollutants are likely to be due at different Because chemical/physical properties and evaluating effects of point sources, congested
times. Thus, the following paragraphs origins of each component differ, it may be intersections and highways, as well as the
identify appropriate modeling approaches for appropriate to use either a single model cumulative effect of numerous sources of CO
pollutants individually. capable of addressing several of the in an urban area.
c. The NAAQS for ozone was revised on important components or to model primary j. Models for assessing the impact of
July 18, 1997 and is now based on an 8-hour and secondary components using different sources on ambient NO2 concentrations are
averaging period. Models for ozone are models. Effects of a control strategy on PM– primarily needed to meet new source review
needed primarily to guide choice of strategies 2.5 is estimated from the sum of the effects requirements, such as addressing the effect of
to correct an observed ozone problem in an on the components composing PM–2.5. a proposed source on PSD increments for
area not attaining the NAAQS for ozone. Use Model users may refer to guidance 33 for annual concentrations of NO2. Impact of an
of photochemical grid models is the further details concerning appropriate individual source on ambient NO2 depends,
recommended means for identifying modeling approaches. in part, on the chemical environment into
strategies needed to correct high ozone f. A control agency with jurisdiction over which the source’s plume is to be emitted.
concentrations in such areas. Such models
one or more areas with PM–2.5 problems There are several approaches for estimating
need to consider emissions of volatile organic
should review available ambient air quality effects of an individual source on ambient
compounds (VOC), nitrogen oxides (NOX)
data to assess which components of PM–2.5 NO2. One approach is through use of a
and carbon monoxide (CO), as well as means
are likely to be major contributors to the plume-in-grid algorithm imbedded within a
for generating meteorological data governing
problem. If it is determined that regional photochemical grid model. However, because
transport of secondary particulates, such as of the rigor and complexity involved, and
a Modeling for attainment demonstrations for O
3
sulfates or nitrates, is likely to contribute because this approach may not be capable of
and PM–2.5 should be conducted in time to meet
required SIP submission dates as provided for in
significantly to the problem, use of a regional defining sub-grid concentration gradients, the
the respective implementation rules. Information on model may be the preferred approach. plume-in-grid approach may be impractical
implementation of the 8-hr O3 and PM–2.5 Otherwise, coverage may be limited to a for estimating effects on an annual PSD
standards is available at: http://www.epa.gov/ttn/ domain that is urban scale or less. Special increment. A second approach which does
naags/. care should be taken to select appropriate not have this limitation and accommodates
distance-dependent conversion ratios—the a highly complex and resource-intensive should be approved by the Regional Office on
Plume Volume Molar Ratio Method exercise. Control agencies with jurisdiction a case-by-case basis. Analyses involving
(PVMRM) 36—is currently being tested to over areas with secondary PM–2.5 problems model calculations for stagnation conditions
determine suitability as a refined method. A are encouraged to use models which integrate should also be justified on a case-by-case
third (screening) approach is to develop site chemical and physical processes important basis (subsection 7.2.8).
specific (domain-wide) conversion factors in the formation, decay and transport of these e. Fugitive dust usually refers to dust put
based on measurements. If it is not possible species (e.g., Models-3/CMAQ 38 or into the atmosphere by the wind blowing
to develop site specific conversion factors REMSAD 41). Primary components can be over plowed fields, dirt roads or desert or
and use of the plume-in-grid algorithm is also simulated using less resource-intensive sandy areas with little or no vegetation.
not feasible, other screening procedures may techniques. Suitability of a modeling
Reentrained dust is that which is put into the
be considered. approach or mix of modeling approaches for
air by reason of vehicles driving over dirt
k. In January 1999 (40 CFR Part 58, a given application requires technical
judgement,33 as well as professional roads (or dirty roads) and dusty areas. Such
Appendix D), EPA gave notice that concern
experience in choice of models, use of the sources can be characterized as line, area or
about ambient lead impacts was being shifted
away from roadways and toward a focus on model(s) in an attainment test, development volume sources. Emission rates may be based
stationary point sources. EPA has also issued of emissions and meteorological inputs to the on site specific data or values from the
guidance on siting ambient monitors in the model and selection of days to model. general literature. Fugitive emissions include
vicinity of such sources.37 For lead, the SIP b. Choice of Analysis Techniques to the emissions resulting from the industrial
should contain an air quality analysis to Complement Air Quality Simulation Models. process that are not captured and vented
determine the maximum quarterly lead Receptor models may be used to corroborate through a stack but may be released from
concentration resulting from major lead point predictions obtained with one or more air various locations within the complex. In
sources, such as smelters, gasoline additive quality simulation models. They may also be some unique cases a model developed
plants, etc. General guidance for lead SIP potentially useful in helping to define specifically for the situation may be needed.
development is also available.38 specific source categories contributing to Due to the difficult nature of characterizing
major components of PM–2.5.33 and modeling fugitive dust and fugitive
5.2 Recommendations c. Estimating the Impact of Individual emissions, it is recommended that the
5.2.1 Models for Ozone Sources. Choice of methods used to assess proposed procedure be cleared by the
a. Choice of Models for Multi-source the impact of an individual source depends Regional Office for each specific situation
Applications. Simulation of ozone formation on the nature of the source and its emissions. before the modeling exercise is begun.
and transport is a highly complex and Thus, model users should consult with the
Regional Office to determine the most 5.2.3 Models for Carbon Monoxide
resource intensive exercise. Control agencies
with jurisdiction over areas with ozone suitable approach on a case-by-case basis a. Guidance is available for analyzing CO
problems are encouraged to use (subsection 3.2.2). impacts at roadway intersections.48 The
photochemical grid models, such as the 5.2.2.2 PM–10 recommended screening model for such
Models-3/Community Multi-scale Air analyses is CAL3QHC.49 50 This model
a. Screening techniques like those combines CALINE3 (listed in Appendix A)
Quality (CMAQ) modeling system,39 to identified in subsection 4.2.1 are applicable
evaluate the relationship between precursor with a traffic model to calculate delays and
to PM–10. Conservative assumptions which
species and ozone. Judgement on the queues that occur at signalized intersections.
do not allow removal or transformation are
suitability of a model for a given application The screening approach is described in
suggested for screening. Thus, it is
should consider factors that include use of reference 48; a refined approach may be
recommended that subjectively determined
the model in an attainment test, development values for ‘‘half-life’’ or pollutant decay not considered on a case-by-case basis with
of emissions and meteorological inputs to the be used as a surrogate for particle removal. CAL3QHCR.51 The latest version of the
model and choice of episodes to model.32 Proportional models (rollback/forward) may MOBILE (mobile source emission factor)
Similar models for the 8-hour NAAQS and not be applied for screening analysis, unless model should be used for emissions input to
for the 1-hour NAAQS are appropriate. such techniques are used in conjunction with intersection models.
b. Choice of Models to Complement receptor modeling.34 b. For analyses of highways characterized
Photochemical Grid Models. As previously b. Refined models such as those discussed by uninterrupted traffic flows, CALINE3 is
noted, observational models, Lagrangian in subsection 4.2.2 are recommended for recommended, with emissions input from the
models, or the refined version of the Ozone PM–10. However, where possible, particle latest version of the MOBILE model. A
Isopleth Plotting Program (OZIPR) 40 may be size, gas-to-particle formation, and their scientific review article for line source
used to help guide choice of strategies to effect on ambient concentrations may be models is available.52
simulate with a photochemical grid model considered. For point sources of small c. For urban area wide analyses of CO, an
and to corroborate results obtained with a particles and for source-specific analyses of Eulerian grid model should be used.
grid model. Receptor models have also been complicated sources, use the appropriate Information on SIP development and
used to apportion sources of ozone recommended steady-state plume dispersion requirements for using such models can be
precursors (e.g., VOC) in urban domains. EPA model (subsection 4.2.2). found in several references.48 53 54 55
has issued guidance 32 in selecting c. Receptor models have proven useful for d. Where point sources of CO are of
appropriate techniques. helping validate emission inventories and for concern, they should be treated using the
c. Estimating the Impact of Individual corroborating source-specific impacts
Sources. Choice of methods used to assess screening and refined techniques described
estimated by dispersion models. The in Section 4.
the impact of an individual source depends Chemical Mass Balance (CMB) model is
on the nature of the source and its emissions. useful for apportioning impacts from 5.2.4 Models for Nitrogen Dioxide (Annual
Thus, model users should consult with the localized sources.42 43 44 Other receptor Average)
Regional Office to determine the most models, e.g., the Positive Matrix a. A tiered screening approach is
suitable approach on a case-by-case basis Factorization (PMF) model 45 and Unmix,46 recommended to obtain annual average
(subsection 3.2.2). which don’t share some of CMB’s constraints, estimates of NO2 from point sources for New
5.2.2 Models for Particulate Matter have also been applied. In regulatory Source Review analysis, including PSD, and
applications, dispersion models have been for SIP planning purposes. This multi-tiered
5.2.2.1 PM–2.5 used in conjunction with receptor models to approach is conceptually shown in Figure 5–
a. Choice of Models for Multi-source attribute source (or source category) 1 and described in paragraphs b through d of
Applications. Simulation of phenomena contributions. Guidance is available for PM– this subsection:
resulting in high ambient PM–2.5 can be a 10 sampling and analysis applicable to
multi-faceted and complex problem resulting receptor modeling.47 Figure 5–1
from PM–2.5’s existence as an aerosol d. Under certain conditions, recommended Multi-tiered screening approach for
mixture. Treating secondary components of dispersion models may not be reliable. In Estimating Annual NO2 Concentrations from
PM–2.5, such as sulfates and nitrates, can be such circumstances, the modeling approach Point Sources
b. For Tier 1 (the initial screen), use an located within the same distance of the square based on the portion of highway link
appropriate model in subsection 4.2.2 to source as the source-to-monitor distance. within each grid square. If localized areas of
estimate the maximum annual average e. In urban areas (subsection 7.2.3), a high concentrations are likely, then mobile
concentration and assume a total conversion proportional model may be used as a sources should be modeled as line sources
of NO to NO2. If the concentration exceeds preliminary assessment to evaluate control using an appropriate steady-state plume
the NAAQS and/or PSD increments for NO2, strategies to meet the NAAQS for multiple dispersion model (e.g., CAL3QHCR;
proceed to the 2nd level screen. minor sources, i.e., minor point, area and subsection 5.2.3).
c. For Tier 2 (2nd level) screening analysis, mobile sources of NOX; concentrations g. More refined techniques to handle
multiply the Tier 1 estimate(s) by an resulting from major point sources should be special circumstances may be considered on
empirically derived NO2/NOX value of 0.75 estimated separately as discussed above, then a case-by-case basis and agreement with the
(annual national default).56 The reviewing added to the impact of the minor sources. An appropriate reviewing authority (paragraph
agency may establish an alternative default acceptable screening technique for urban 3.0(b)) should be obtained. Such techniques
NO2/NOX ratio based on ambient annual complexes is to assume that all NOX is should consider individual quantities of NO
average NO2 and annual average NOX data emitted in the form of NO2 and to use a and NO2 emissions, atmospheric transport
representative of area wide quasi-equilibrium model from Appendix A for nonreactive and dispersion, and atmospheric
conditions. Alternative default NO2/NOX pollutants to estimate NO2 concentrations. A transformation of NO to NO2. Where they are
ratios should be based on data satisfying more accurate estimate can be obtained by: available, site specific data on the conversion
quality assurance procedures that ensure data (1) Calculating the annual average of NO to NO2 may be used. Photochemical
accuracy for both NO2 and NOX within the concentrations of NOX with an urban model, dispersion models, if used for other
typical range of measured values. In areas and (2) converting these estimates to NO2 pollutants in the area, may also be applied
with relatively low NOX concentrations, the concentrations using an empirically derived to the NOX problem.
quality assurance procedures used to annual NO2/NOX ratio. A value of 0.75 is
5.2.5 Models for Lead
determine compliance with the NO2 national recommended for this ratio. However, a
ambient air quality standard may not be spatially averaged alternative default annual a. For major lead point sources, such as
adequate. In addition, default NO2/NOX NO2/NOX ratio may be determined from an smelters, which contribute fugitive emissions
ratios, including the 0.75 national default existing air quality monitoring network and and for which deposition is important,
value, can underestimate long range NO2 used in lieu of the 0.75 value if it is professional judgement should be used, and
impacts and should be used with caution in determined to be representative of prevailing there should be coordination with the
long range transport scenarios. ratios in the urban area by the reviewing appropriate reviewing authority (paragraph
d. For Tier 3 (3rd level) analysis, a detailed agency. To ensure use of appropriate locally 3.0(b)). To model an entire major urban area
screening method may be selected on a case- derived annual average NO2/NOX ratios, or to model areas without significant sources
by-case basis. For point source modeling, monitoring data under consideration should of lead emissions, as a minimum a
detailed screening techniques such as the be limited to those collected at monitors proportional (rollback) model may be used
Ozone Limiting Method 57 may also be meeting siting criteria defined in 40 CFR Part for air quality analysis. The rollback
considered. Also, a site specific NO2/NOX 58, Appendix D as representative of philosophy assumes that measured pollutant
ratio may be used as a detailed screening ‘‘neighborhood’’, ‘‘urban’’, or ‘‘regional’’ concentrations are proportional to emissions.
method if it meets the same restrictions as scales. Furthermore, the highest annual However, urban or other dispersion models
described for alternative default NO2/NOX spatially averaged NO2/NOX ratio from the are encouraged in these circumstances where
ratios. Ambient NOX monitors used to most recent 3 years of complete data should the use of such models is feasible.
develop a site specific ratio should be sited be used to foster conservatism in estimated b. In modeling the effect of traditional line
to obtain the NO2 and NOX concentrations impacts. sources (such as a specific roadway or
under quasi-equilibrium conditions. Data f. To demonstrate compliance with NO2 highway) on lead air quality, dispersion
obtained from monitors sited at the PSD increments in urban areas, emissions models applied for other pollutants can be
maximum NOX impact site, as may be from major and minor sources should be used. Dispersion models such as CALINE3
required in a PSD pre-construction included in the modeling analysis. Point and and CAL3QHCR have been used for modeling
monitoring program, likely reflect area source emissions should be modeled as carbon monoxide emissions from highways
transitional NOX conditions. Therefore, NOX discussed above. If mobile source emissions and intersections (subsection 5.2.3). Where
data from maximum impact sites may not be do not contribute to localized areas of high there is a point source in the middle of a
suitable for determining a site specific NO2/ ambient NO2 concentrations, they should be substantial road network, the lead
NOX ratio that is applicable for the entire modeled as area sources. When modeled as concentrations that result from the road
modeling analysis. A site specific ratio area sources, mobile source emissions should network should be treated as background
derived from maximum impact data can only be assumed uniform over the entire highway (subsection 8.2); the point source and any
ER09NO05.001</GPH>
be used to estimate NO2 impacts at receptors link and allocated to each area source grid nearby major roadways should be modeled
separately using the appropriate no apparent bias toward over or under authority (paragraph 3.0(b)) and the affected
recommended steady-state plume dispersion prediction, so long as the transport distance Federal Land Manager (FLM). FLMs are
model (subsection 4.2.2). was limited to less than 300km.60 responsible for determining whether there is
an adverse effect by a plume on a Class I area.
6.0 Other Model Requirements 6.2 Recommendations e. CALPUFF (Section A.3) may be applied
6.1 Discussion 6.2.1 Visibility when assessment is needed of reasonably
a. Visibility in important natural areas (e.g., attributable haze impairment or atmospheric
a. This section covers those cases where
Federal Class I areas) is protected under a deposition due to one or a small group of
specific techniques have been developed for
number of provisions of the Clean Air Act, sources. This situation may involve more
special regulatory programs. Most of the
including Sections 169A and 169B sources and larger modeling domains than
programs have, or will have when fully
(addressing impacts primarily from existing that to which VISCREEN ideally may be
developed, separate guidance documents that
sources) and Section 165 (new source applied. The procedures and analyses should
cover the program and a discussion of the
review). Visibility impairment is caused by be determined in consultation with the
tools that are needed. The following
light scattering and light absorption appropriate reviewing authority (paragraph
paragraphs reference those guidance
associated with particles and gases in the 3.0(b)) and the affected FLM(s).
documents, when they are available. No
atmosphere. In most areas of the country, f. Regional scale models are used by EPA
attempt has been made to provide a
light scattering by PM–2.5 is the most to develop and evaluate national policy and
comprehensive discussion of each topic since
significant component of visibility assist State and local control agencies. Two
the reference documents were designed to do
impairment. The key components of PM–2.5 such models which can be used to assess
that. This section will undergo periodic
contributing to visibility impairment include visibility impacts from source emissions are
revision as new programs are added and new
sulfates, nitrates, organic carbon, elemental Models-3/CMAQ 38 and REMSAD.41 Model
techniques are developed.
carbon, and crustal material. users should consult with the appropriate
b. Other Federal agencies have also
b. The visibility regulations as promulgated reviewing authority (paragraph 3.0(b)), which
developed specific modeling approaches for
in December 1980 (40 CFR 51.300–307) in this instance would include FLMs.
their own regulatory or other requirements.58
Although such regulatory requirements and require States to mitigate visibility 6.2.2 Good Engineering Practice Stack
manuals may have come about because of impairment, in any of the 156 mandatory Height
EPA rules or standards, the implementation Federal Class I areas, that is found to be a. The use of stack height credit in excess
of such regulations and the use of the ‘‘reasonably attributable’’ to a single source of Good Engineering Practice (GEP) stack
modeling techniques is under the jurisdiction or a small group of sources. In 1985, EPA height or credit resulting from any other
of the agency issuing the manual or directive. promulgated Federal Implementation Plans dispersion technique is prohibited in the
c. The need to estimate impacts at (FIPs) for several States without approved development of emission limitations by 40
distances greater than 50km (the nominal visibility provisions in their SIPs. The CFR 51.118 and 40 CFR 51.164. The
distance to which EPA considers most IMPROVE (Interagency Monitoring for definitions of GEP stack height and
steady-state Gaussian plume models are Protected Visual Environments) monitoring dispersion technique are contained in 40 CFR
applicable) is an important one especially network, a cooperative effort between EPA, 51.100. Methods and procedures for making
when considering the effects from secondary the States, and Federal land management the appropriate stack height calculations,
pollutants. Unfortunately, models originally agencies, was established to implement the determining stack height credits and an
available to EPA had not undergone monitoring requirements in these FIPs. Data example of applying those techniques are
sufficient field evaluation to be has been collected by the IMPROVE network found in several references 64 65 66 67, which
recommended for general use. Data bases since 1988. provide a great deal of additional information
from field studies at mesoscale and long c. In 1999, EPA issued revisions to the for evaluating and describing building cavity
range transport distances were limited in 1980 regulations to address visibility and wake effects.
detail. This limitation was a result of the impairment in the form of regional haze, b. If stacks for new or existing major
expense to perform the field studies required which is caused by numerous, diverse sources are found to be less than the height
to verify and improve mesoscale and long sources (e.g., stationary, mobile, and area defined by EPA’s refined formula for
range transport models. Meteorological data sources) located across a broad region (40 determining GEP height, then air quality
adequate for generating three-dimensional CFR 51.308–309). The state of relevant impacts associated with cavity or wake
wind fields were particularly sparse. scientific knowledge has expanded effects due to the nearby building structures
Application of models to complicated terrain significantly since the Clean Air Act should be determined. The EPA refined
compounds the difficulty of making good Amendments of 1977. A number of studies formula height is defined as H + 1.5L (see
assessments of long range transport impacts. and reports 61 62 have concluded that long reference 66). Detailed downwash screening
EPA completed limited evaluation of several range transport (e.g., up to hundreds of procedures 24 for both the cavity and wake
long range transport (LRT) models against kilometers) of fine particulate matter plays a regions should be followed. If more refined
two sets of field data and evaluated results.59 significant role in visibility impairment concentration estimates are required, the
Based on the results, EPA concluded that across the country. Section 169A of the Act recommended steady-state plume dispersion
long range and mesoscale transport models requires states to develop SIPs containing model in subsection 4.2.2 contains
were limited for regulatory use to a case-by- long-term strategies for remedying existing algorithms for building wake calculations
case basis. However a more recent series of and preventing future visibility impairment and should be used.
comparisons has been completed for a new in 156 mandatory Class I federal areas. In
model, CALPUFF (Section A.3). Several of order to develop long-term strategies to 6.2.3 Long Range Transport (LRT) (i.e.,
these field studies involved three-to-four address regional haze, many States will need Beyond 50km)
hour releases of tracer gas sampled along arcs to conduct regional-scale modeling of fine a. Section 165(d) of the Clean Air Act
of receptors at distances greater than 50km particulate concentrations and associated requires that suspected adverse impacts on
downwind. In some cases, short-term visibility impairment (e.g., light extinction PSD Class I areas be determined. However,
concentration sampling was available, such and deciview metrics). 50km is the useful distance to which most
that the transport of the tracer puff as it d. To calculate the potential impact of a steady-state Gaussian plume models are
passed the arc could be monitored. plume of specified emissions for specific considered accurate for setting emission
Differences on the order of 10 to 20 degrees transport and dispersion conditions (‘‘plume limits. Since in many cases PSD analyses
were found between the location of the blight’’), a screening model, VISCREEN, and show that Class I areas may be threatened at
simulated and observed center of mass of the guidance are available.63 If a more distances greater than 50km from new
tracer puff. Most of the simulated centerline comprehensive analysis is required, a refined sources, some procedure is needed to (1)
concentration maxima along each arc were model should be selected . The model determine if an adverse impact will occur,
within a factor of two of those observed. It selection (VISCREEN vs. PLUVUE II or some and (2) identify the model to be used in
was concluded from these case studies that other refined model), procedures, and setting an emission limit if the Class I
the CALPUFF dispersion model had analyses should be determined in increments are threatened. In addition to the
performed in a reasonable manner, and had consultation with the appropriate reviewing situations just described, there are certain
applications containing a mixture of both b. The Offshore and Coastal Dispersion increments. For the 24-hour PM–10 NAAQS
long range and short range source-receptor (OCD) model, described in Appendix A, was (which is a probabilistic standard)—when
relationships in a large modeled domain (e.g., developed by the Minerals Management multiple years are modeled, they collectively
several industrialized areas located along a Service and is recommended for estimating represent a single period. Thus, if 5 years of
river or valley). Historically, these air quality impact from offshore sources on NWS data are modeled, then the highest
applications have presented considerable onshore, flat terrain areas. The OCD model is sixth highest concentration for the whole
difficulty to an analyst if impacts from not recommended for use in air quality period becomes the design value. And in
sources having transport distances greater impact assessments for onshore sources. general, when n years are modeled, the
than 50km significantly contributed to the Sources located on or just inland of a (n+1)th highest concentration over the n-year
design concentrations. To properly analyze shoreline where fumigation is expected period is the design value, since this
applications of this type, a modeling should be treated in accordance with represents an average or expected exceedance
approach is needed which has the capability subsection 7.2.8. rate of one per year.
of combining, in a consistent manner, c. The latest version of the Emissions and c. When sufficient and representative data
impacts involving both short and long range Dispersion Modeling System (EDMS), was exist for less than a 5-year period from a
transport. The CALPUFF modeling system, developed and is supported by the Federal nearby NWS site, or when site specific data
listed in Appendix A, has been designed to Aviation Administration (FAA), and is have been collected for less than a full
accommodate both the Class I area LRT appropriate for air quality assessment of continuous year, or when it has been
situation and the large modeling domain primary pollutant impacts at airports or air determined that the site specific data may not
situation. Given the judgement and bases. EDMS has adopted AERMOD for be temporally representative (subsection
refinement involved, conducting a LRT treating dispersion. Application of EDMS is 8.3.3), then the highest concentration
modeling assessment will require significant intended for estimating the collective impact estimate should be considered the design
consultation with the appropriate reviewing of changes in aircraft operations, point value. This is because the length of the data
authority (paragraph 3.0(b)) and the affected source, and mobile source emissions on record may be too short to assure that the
FLM(s). The FLM has an affirmative pollutant concentrations. It is not intended conditions producing worst-case estimates
responsibility to protect air quality related for PSD, SIP, or other regulatory air quality have been adequately sampled. The highest
values (AQRVs) that may be affected, and to analyses of point or mobile sources at or value is then a surrogate for the
provide the appropriate procedures and peripheral to airport property that are concentration that is not to be exceeded more
analysis techniques. Where there is no unrelated to airport operations. If changes in than once per year (the wording of the
increment violation, the ultimate decision on other than aircraft operations are associated deterministic standards). Also, the highest
whether a Class I area is adversely affected with analyses, a model recommended in concentration should be used whenever
is the responsibility of the appropriate Chapter 4 or 5 should be used. The latest selected worst-case conditions are input to a
reviewing authority (Section 165(d)(2)(C)(ii) version of EDMS may be obtained from FAA screening technique, as described in EPA
of the Clean Air Act), taking into at its Web site: http://www.aee.faa.gov/ guidance.24
consideration any information on the impacts emissions/edms/edmshome.htm. d. If the controlling concentration is an
on AQRVs provided by the FLM. According annual average value and multiple years of
to Section 165(d)(2)(C)(iii) of the Clean Air 7.0 General Modeling Considerations data (site specific or NWS) are used, then the
Act, if there is a Class I increment violation, design value is the highest of the annual
7.1 Discussion
the source must demonstrate to the averages calculated for the individual years.
a. This section contains recommendations If the controlling concentration is a quarterly
satisfaction of the FLM that the emissions
concerning a number of different issues not average and multiple years are used, then the
from the source will have no adverse impact
explicitly covered in other sections of this highest individual quarterly average should
on the AQRVs.
guide. The topics covered here are not be considered the design value.
b. If LRT is determined to be important,
specific to any one program or modeling area e. As long a period of record as possible
then refined estimates utilizing the CALPUFF
but are common to nearly all modeling should be used in making estimates to
modeling system should be obtained. A
screening approach 60 68 is also available for analyses for criteria pollutants. determine design values and PSD
use on a case-by-case basis that generally increments. If more than 1 year of site
7.2 Recommendations specific data is available, it should be used.
provides concentrations that are higher than
7.2.1 Design Concentrations (See Also 7.2.1.2 Design Concentrations for O3 and
those obtained using refined
Subsection 10.2.3.1) PM–2.5
characterizations of the meteorological
conditions. The meteorological input data 7.2.1.1 Design Concentrations for SO2, PM– a. Guidance and specific instructions for
requirements for developing the time and 10, CO, Pb, and NO2 the determination of the 1-hr and 8-hr design
space varying three-dimensional winds and a. An air quality analysis for SO2, PM–10, concentrations for ozone are provided in
dispersion meteorology for refined analyses CO, Pb, and NO2 is required to determine if Appendix H and I (respectively) of reference
are discussed in paragraph 8.3.1.2(d). the source will (1) cause a violation of the 4. Appendix H explains how to determine
Additional information on applying this when the expected number of days per
NAAQS, or (2) cause or contribute to air
model is contained in Appendix A. To calendar year with maximum hourly
quality deterioration greater than the
facilitate use of complex air quality and concentrations above the NAAQS is equal to
specified allowable PSD increment. For the
meteorological modeling systems, a written or less than 1. Appendix I explains the data
former, background concentration
protocol approved by the appropriate handling conventions and computations
(subsection 8.2) should be added to the
reviewing authority (paragraph 3.0(b)) and necessary for determining whether the 8-hour
estimated impact of the source to determine
the affected FLM(s) may be considered for primary and secondary NAAQS are met at an
the design concentration. For the latter, the
developing consensus in the methods and ambient monitoring site. For PM–2.5,
design concentration includes impact from
procedures to be followed. Appendix N of reference 4, and
all increment consuming sources.
6.2.4 Modeling Guidance for Other supplementary guidance,69 explain the data
b. If the air quality analyses are conducted
Governmental Programs handling conventions and computations
using the period of meteorological input data
necessary for determining when the annual
a. When using the models recommended or recommended in subsection 8.3.1.2 (e.g., 5 and 24-hour primary and secondary NAAQS
discussed in the Guideline in support of years of National Weather Service (NWS) are met. For all SIP revisions the user should
programmatic requirements not specifically data or at least 1 year of site specific data; check with the Regional Office to obtain the
covered by EPA regulations, the model user subsection 8.3.3), then the design most recent guidance documents and policy
should consult the appropriate Federal or concentration based on the highest, second- memoranda concerning the pollutant in
State agency to ensure the proper application highest short term concentration over the question. There are currently no PSD
and use of the models. For modeling entire receptor network for each year increments for O3 and PM–2.5.
associated with PSD permit applications that modeled or the highest long term average
involve a Class I area, the appropriate Federal (whichever is controlling) should be used to 7.2.2 Critical Receptor Sites
Land Manager should be consulted on all determine emission limitations to assess a. Receptor sites for refined modeling
modeling questions. compliance with the NAAQS and PSD should be utilized in sufficient detail to
estimate the highest concentrations and buoyant sources, e.g., those involving fuel is the recommended technique for this
possible violations of a NAAQS or a PSD combustion, are involved. situation and is used in preferred models for
increment. In designing a receptor network, 7.2.4 Stability Categories point sources.
the emphasis should be placed on receptor 7.2.6 Chemical Transformation
resolution and location, not total number of a. The Pasquill approach to classifying
receptors. The selection of receptor sites stability is commonly used in preferred a. The chemical transformation of SO2
should be a case-by-case determination models (Appendix A). The Pasquill method, emitted from point sources or single
taking into consideration the topography, the as modified by Turner 75, was developed for industrial plants in rural areas is generally
climatology, monitor sites, and the results of use with commonly observed meteorological assumed to be relatively unimportant to the
the initial screening procedure. data from the National Weather Service and estimation of maximum concentrations when
is based on cloud cover, insolation and wind travel time is limited to a few hours.
7.2.3 Dispersion Coefficients speed. However, in urban areas, where synergistic
a. Steady-state Gaussian plume models b. Procedures to determine Pasquill effects among pollutants are of considerable
used in most applications should employ stability categories from other than NWS data consequence, chemical transformation rates
dispersion coefficients consistent with those are found in subsection 8.3. Any other may be of concern. In urban area
contained in the preferred models in method to determine Pasquill stability applications, a half-life of 4 hours 75 may be
Appendix A. Factors such as averaging time, categories must be justified on a case-by-case applied to the analysis of SO2 emissions.
urban/rural surroundings (see paragraphs basis. Calculations of transformation coefficients
(b)—(f) of this subsection), and type of source c. For a given model application where from site specific studies can be used to
(point vs. line) may dictate the selection of stability categories are the basis for selecting define a ‘‘half-life’’ to be used in a steady-
specific coefficients. Coefficients used in dispersion coefficients, both sy and sz should state Gaussian plume model with any travel
some Appendix A models are identical to, or be determined from the same stability time, or in any application, if appropriate
at least based on, Pasquill-Gifford category. ‘‘Split sigmas’’ in that instance are documentation is provided. Such conversion
coefficients 70 in rural areas and McElroy- not recommended. Sector averaging, which factors for pollutant half-life should not be
Pooler 71 coefficients in urban areas. A key eliminates the sy term, is commonly used with screening analyses.
feature of AERMOD’s formulation is the use acceptable in complex terrain screening b. Use of models incorporating complex
of directly observed variables of the methods. chemical mechanisms should be considered
boundary layer to parameterize dispersion.22 d. AERMOD, also a preferred model in only on a case-by-case basis with proper
b. The selection of either rural or urban Appendix A, uses a planetary boundary layer demonstration of applicability. These are
dispersion coefficients in a specific scaling parameter to characterize stability.22 generally regional models not designed for
application should follow one of the This approach represents a departure from the evaluation of individual sources but used
procedures suggested by Irwin 72 and briefly the discrete, hourly stability categories primarily for region-wide evaluations.
described in paragraphs (c)—(f) of this estimated under the Pasquill-Gifford-Turner Visibility models also incorporate chemical
subsection. These include a land use scheme. transformation mechanisms which are an
classification procedure or a population 7.2.5 Plume Rise integral part of the visibility model itself and
based procedure to determine whether the should be used in visibility assessments.
character of an area is primarily urban or a. The plume rise methods of Briggs 76 77
rural. are incorporated in many of the preferred 7.2.7 Gravitational Settling and Deposition
c. Land Use Procedure: (1) Classify the models and are recommended for use in a. An ‘‘infinite half-life’’ should be used for
land use within the total area, Ao, many modeling applications. In AERMOD,22 estimates of particle concentrations when
circumscribed by a 3km radius circle about for the stable boundary layer, plume rise is steady-state Gaussian plume models
the source using the meteorological land use estimated using an iterative approach, similar containing only exponential decay terms for
typing scheme proposed by Auer 73; (2) if to that in the CTDMPLUS model. In the treating settling and deposition are used.
land use types I1, I2, C1, R2, and R3 account convective boundary layer, plume rise is b. Gravitational settling and deposition
for 50 percent or more of Ao, use urban superposed on the displacements by random may be directly included in a model if either
dispersion coefficients; otherwise, use convective velocities.78 In AERMOD, plume is a significant factor. When particulate
appropriate rural dispersion coefficients. rise is computed using the methods of Briggs matter sources can be quantified and settling
d. Population Density Procedure: (1) excepting cases involving building and dry deposition are problems,
Compute the average population density, p̄ downwash, in which a numerical solution of professional judgement should be used, and
per square kilometer with Ao as defined the mass, energy, and momentum there should be coordination with the
above; (2) If p̄ is greater than 750 people/km2, conservation laws is performed.23 No explicit appropriate reviewing authority (paragraph
use urban dispersion coefficients; otherwise provisions in these models are made for 3.0(b)).
use appropriate rural dispersion coefficients. multistack plume rise enhancement or the
handling of such special plumes as flares; 7.2.8 Complex Winds
e. Of the two methods, the land use
procedure is considered more definitive. these problems should be considered on a a. Inhomogeneous Local Winds. In many
Population density should be used with case-by-case basis. parts of the United States, the ground is
caution and should not be applied to highly b. Gradual plume rise is generally neither flat nor is the ground cover (or land
industrialized areas where the population recommended where its use is appropriate: use) uniform. These geographical variations
density may be low and thus a rural (1) In AERMOD; (2) in complex terrain can generate local winds and circulations,
classification would be indicated, but the screening procedures to determine close-in and modify the prevailing ambient winds
area is sufficiently built-up so that the urban impacts and (3) when calculating the effects and circulations. Geographic effects are most
land use criteria would be satisfied. In this of building wakes. The building wake apparent when the ambient winds are light
case, the classification should already be algorithm in AERMOD incorporates and or calm.79 In general these geographically
‘‘urban’’ and urban dispersion parameters exercises the thermodynamically based induced wind circulation effects are named
should be used. gradual plume rise calculations as described after the source location of the winds, e.g.,
f. Sources located in an area defined as in (a) above. If the building wake is lake and sea breezes, and mountain and
urban should be modeled using urban calculated to affect the plume for any hour, valley winds. In very rugged hilly or
dispersion parameters. Sources located in gradual plume rise is also used in downwind mountainous terrain, along coastlines, or
areas defined as rural should be modeled dispersion calculations to the distance of near large land use variations, the
using the rural dispersion parameters. For final plume rise, after which final plume rise characterization of the winds is a balance of
analyses of whole urban complexes, the is used. Plumes captured by the near wake various forces, such that the assumptions of
entire area should be modeled as an urban are re-emitted to the far wake as a ground- steady-state straight-line transport both in
region if most of the sources are located in level volume source. time and space are inappropriate. In the
areas classified as urban. c. Stack tip downwash generally occurs special cases described, the CALPUFF
g. Buoyancy-induced dispersion (BID), as with poorly constructed stacks and when the modeling system (described in Appendix A)
identified by Pasquill 74, is included in the ratio of the stack exit velocity to wind speed may be applied on a case-by-case basis for air
preferred models and should be used where is small. An algorithm developed by Briggs 77 quality estimates in such complex non-
steady-state meteorological conditions. The 7.2.9 Calibration of Models established. As a minimum, the source
purpose of choosing a modeling system like a. Calibration of models is not common should be modeled using the design capacity
CALPUFF is to fully treat the time and space practice and is subject to much error and (100 percent load). If a source operates at
variations of meteorology effects on transport misunderstanding. There have been attempts greater than design capacity for periods that
and dispersion. The setup and application of by some to compare model estimates and could result in violations of the standards or
the model should be determined in measurements on an event-by-event basis PSD increments, this load) a should be
consultation with the appropriate reviewing and then to calibrate a model with results of modeled. Where the source operates at
authority (paragraph 3.0(b)) consistent with that comparison. This approach is severely substantially less than design capacity, and
limitations of paragraph 3.2.2(e). The limited by uncertainties in both source and the changes in the stack parameters
meteorological input data requirements for meteorological data and therefore it is associated with the operating conditions
developing the time and space varying three- difficult to precisely estimate the could lead to higher ground level
dimensional winds and dispersion concentrations, loads such as 50 percent and
concentration at an exact location for a
meteorology for these situations are 75 percent of capacity should also be
specific increment of time. Such
discussed in paragraphs 8.3.1.2(d) and modeled. A range of operating conditions
uncertainties make calibration of models of
should be considered in screening analyses;
8.3.1.2(f). Examples of inhomogeneous winds questionable benefit. Therefore, model
the load causing the highest concentration, in
include, but aren’t limited to, situations calibration is unacceptable.
addition to the design load, should be
described in the following paragraphs (i)— included in refined modeling. For a steam
8.0 Model Input Data
(iii): power plant, the following (b–h) is typical of
i. Inversion Breakup Fumigation. Inversion a. Data bases and related procedures for
estimating input parameters are an integral the kind of data on source characteristics and
breakup fumigation occurs when a plume (or operating conditions that may be needed.
multiple plumes) is emitted into a stable part of the modeling procedure. The most
Generally, input data requirements for air
layer of air and that layer is subsequently appropriate data available should always be
quality models necessitate the use of metric
mixed to the ground through convective selected for use in modeling analyses.
units; where English units are common for
transfer of heat from the surface or because Concentrations can vary widely depending
engineering usage, a conversion to metric is
of advection to less stable surroundings. on the source data or meteorological data
required.
Fumigation may cause excessively high used. Input data are a major source of
b. Plant layout. The connection scheme
concentrations but is usually rather short- uncertainties in any modeling analysis. This
between boilers and stacks, and the distance
lived at a given receptor. There are no section attempts to minimize the uncertainty
and direction between stacks, building
recommended refined techniques to model associated with data base selection and use parameters (length, width, height, location
this phenomenon. There are, however, by identifying requirements for data used in and orientation relative to stacks) for plant
screening procedures 24 that may be used to modeling. A checklist of input data structures which house boilers, control
requirements for modeling analyses is posted equipment, and surrounding buildings
approximate the concentrations.
on EPA’s Internet SCRAM Web site within a distance of approximately five stack
Considerable care should be exercised in
(subsection 2.3). More specific data heights.
using the results obtained from the screening
requirements and the format required for the c. Stack parameters. For all stacks, the
techniques.
individual models are described in detail in stack height and inside diameter (meters),
ii. Shoreline Fumigation. Fumigation can
the users’ guide for each model. and the temperature (K) and volume flow rate
be an important phenomenon on and near
the shoreline of bodies of water. This can 8.1 Source Data (actual cubic meters per second) or exit gas
affect both individual plumes and area-wide velocity (meters per second) for operation at
8.1.1 Discussion 100 percent, 75 percent and 50 percent load.
emissions. When fumigation conditions are
expected to occur from a source or sources a. Sources of pollutants can be classified as d. Boiler size. For all boilers, the associated
with tall stacks located on or just inland of point, line and area/volume sources. Point megawatts, 106 BTU/hr, and pounds of steam
a shoreline, this should be addressed in the sources are defined in terms of size and may per hour, and the design and/or actual fuel
vary between regulatory programs. The line consumption rate for 100 percent load for
air quality modeling analysis. The Shoreline
sources most frequently considered are coal (tons/hour), oil (barrels/hour), and
Dispersion Model (SDM) listed on EPA’s
roadways and streets along which there are natural gas (thousand cubic feet/hour).
Internet SCRAM Web site (subsection 2.3)
well-defined movements of motor vehicles, e. Boiler parameters. For all boilers, the
may be applied on a case-by-case basis when percent excess air used, the boiler type (e.g.,
air quality estimates under shoreline but they may be lines of roof vents or stacks
such as in aluminum refineries. Area and wet bottom, cyclone, etc.), and the type of
fumigation conditions are needed.80 firing (e.g., pulverized coal, front firing, etc.).
Information on the results of EPA’s volume sources are often collections of a
multitude of minor sources with individually f. Operating conditions. For all boilers, the
evaluation of this model together with other type, amount and pollutant contents of fuel,
small emissions that are impractical to
coastal fumigation models is available.81 the total hours of boiler operation and the
consider as separate point or line sources.
Selection of the appropriate model for boiler capacity factor during the year, and the
Large area sources are typically treated as a
applications where shoreline fumigation is of percent load for peak conditions.
grid network of square areas, with pollutant
concern should be determined in g. Pollution control equipment parameters.
emissions distributed uniformly within each
consultation with the appropriate reviewing For each boiler served and each pollutant
grid square.
authority (paragraph 3.0(b)). affected, the type of emission control
b. Emission factors are compiled in an EPA
iii. Stagnation. Stagnation conditions are equipment, the year of its installation, its
publication commonly known as AP–42 82;
characterized by calm or very low wind design efficiency and mass emission rate, the
an indication of the quality and amount of
speeds, and variable wind directions. These date of the last test and the tested efficiency,
data on which many of the factors are based
stagnant meteorological conditions may the number of hours of operation during the
is also provided. Other information
persist for several hours to several days. latest year, and the best engineering estimate
concerning emissions is available in EPA
During stagnation conditions, the dispersion of its projected efficiency if used in
publications relating to specific source conjunction with coal combustion; data for
of air pollutants, especially those from low- categories. The appropriate reviewing
level emissions sources, tends to be any anticipated modifications or additions.
authority (paragraph 3.0(b)) should be h. Data for new boilers or stacks. For all
minimized, potentially leading to relatively consulted to determine appropriate source
high ground-level concentrations. If point new boilers and stacks under construction
definitions and for guidance concerning the
sources are of interest, users should note the determination of emissions from and a Malfunctions which may result in excess
guidance provided for CALPUFF in techniques for modeling the various source
paragraph (a) of this subsection. Selection of emissions are not considered to be a normal
types. operating condition. They generally should not be
the appropriate model for applications where
8.1.2 Recommendations considered in determining allowable emissions.
stagnation is of concern should be However, if the excess emissions are the result of
determined in consultation with the a. For point source applications the load or poor maintenance, careless operation, or other
appropriate reviewing authority (paragraph operating condition that causes maximum preventable conditions, it may be necessary to
3.0(b)). ground-level concentrations should be consider them in determining source impact.
and for all planned modifications to existing also be consulted for other possible emission k. The impact of growth on emissions
boilers or stacks, the scheduled date of data that could be helpful. NAAQS should be considered in all modeling
completion, and the data or best estimates compliance demonstrations in a PSD analysis analyses covering existing sources. Increases
available for items (b) through (g) of this should follow the emission input data shown in emissions due to planned expansion or
subsection following completion of in Table 8–2. For purposes of emissions planned fuel switches should be identified.
construction or modification. trading, new source review and Increases in emissions at individual sources
i. In stationary point source applications demonstrations, refer to current EPA policy
for compliance with short term ambient that may be associated with a general
and guidance to establish input data. industrial/commercial/residential expansion
standards, SIP control strategies should be
j. Line source modeling of streets and in multi-source urban areas should also be
tested using the emission input shown on
highways requires data on the width of the
Table 8–1. When using a refined model, treated. For new sources the impact of
roadway and the median strip, the types and
sources should be modeled sequentially with growth on emissions should generally be
these loads for every hour of the year. To amounts of pollutant emissions, the number
considered for the period prior to the start-
evaluate SIPs for compliance with quarterly of lanes, the emissions from each lane and
the height of emissions. The location of the up date for the source. Such changes in
and annual standards, emission input data emissions should treat increased area source
shown in Table 8–1 should again be used. ends of the straight roadway segments should
be specified by appropriate grid coordinates. emissions, changes in existing point source
Emissions from area sources should generally
be based on annual average conditions. The Detailed information and data requirements emissions which were not subject to
source input information in each model for modeling mobile sources of pollution are preconstruction review, and emissions due to
user’s guide should be carefully consulted provided in the user’s manuals for each of sources with permits to construct that have
and the checklist (paragraph 8.0(a)) should the models applicable to mobile sources. not yet started operation.
TABLE 8–1.—MODEL EMISSION INPUT DATA FOR POINT SOURCES 1

Emission limit Operating level Operating factor
Averaging time × ×
(#/MMBtu) 2 (MMBtu/hr) 2 (e.g., hr/yr, hr/day)
Stationary Point Source(s) Subject to SIP Emission Limit(s) Evaluation for Compliance with Ambient Standards (Including Areawide
Demonstrations)
Annual & quarterly ..................... Maximum allowable emission Actual or design capacity Actual operating factor aver-
limit or federally enforceable (whichever is greater), or fed- aged over most recent 2
permit limit. erally enforceable permit con- years.3
dition.
Short term .................................. Maximum allowable emission Actual or design capacity Continuous operation, i.e., all
limit or federally enforceable (whichever is greater), or fed- hours of each time period
permit limit. erally enforceable permit con- under consideration (for all
dition.4 hours of the meteorological
data base).5
Nearby Source(s) 6 7
Same input requirements as for stationary point source(s) above.
Other Source(s) 7
If modeled (subsection 8.2.3), input data requirements are defined below.
Annual & quarterly ..................... Maximum allowable emission Annual level when actually op- Actual operating factor aver-
limit or federally enforceable erating, averaged over the aged over the most recent 2
permit limit.6 most recent 2 years.3 years.3
Short term .................................. Maximum allowable emission Annual level when actually op- Continuous operation, i.e., all
limit or federally enforceable erating, averaged over the hours of each time period
permit limit.6 most recent 2 years.3 under consideration (for all
hours of the meteorological
data base).5
1 The model input data requirements shown on this table apply to stationary source control strategies for STATE IMPLEMENTATION PLANS.
For purposes of emissions trading, new source review, or prevention of significant deterioration, other model input criteria may apply. Refer to
the policy and guidance for these programs to establish the input data.
2 Terminology applicable to fuel burning sources; analogous terminology (e.g., #/throughput) may be used for other types of sources.
3 Unless it is determined that this period is not representative.
4 Operating levels such as 50 percent and 75 percent of capacity should also be modeled to determine the load causing the highest concentra-
tion.
5 If operation does not occur for all hours of the time period of consideration (e.g., 3 or 24 hours) and the source operation is constrained by a
federally enforceable permit condition, an appropriate adjustment to the modeled emission rate may be made (e.g., if operation is only 8 a.m. to
4 p.m. each day, only these hours will be modeled with emissions from the source. Modeled emissions should not be averaged across non-oper-
ating time periods.)
6 See paragraph 8.2.3(c).
7 See paragraph 8.2.3(d).
TABLE 8–2.—POINT SOURCE MODEL EMISSION INPUT DATA FOR NAAQS COMPLIANCE IN PSD DEMONSTRATIONS
Emission limit Operating level Operating factor
Averaging time × ×
(#/MMBtu) 1 (MMBtu/hr) 1 (e.g., hr/yr, hr/day)
Proposed Major New or Modified Source
Annual & quarterly ..................... Maximum allowable emission Design capacity or federally en- Continuous operation (i.e., 8760
limit or federally enforceable forceable permit condition. hours).2
permit limit.
Short term (≤ 24 hours) ............. Maximum allowable emission Design capacity or federally en- Continuous operation, i.e., all
limit or federally enforceable forceable permit condition.3 hours of each time period
permit limit. under consideration (for all
data base).2
Nearby Source(s) 4 6
Annual & quarterly ..................... Maximum allowable emission Actual or design capacity Actual operating factor aver-
limit or federally enforceable (whichever is greater), or fed- aged over the most recent 2
permit limit.5 erally enforceable permit con- years.7 8
dition.
Short term (≤ 24 hours) ............. Maximum allowable emission Actual or design capacity Continuous operation, i.e., all
limit or federally enforceable (whichever is greater), or fed- hours of each time period
permit limit.5 erally enforceable permit con- under consideration (for all
dition.3 hours of the meteorological
data base).2
Other Source(s) 6 9
Annual & quarterly ..................... Maximum allowable emission Annual level when actually op- Actual operating factor aver-
limit or federally enforceable erating, averaged over the aged over the most recent 2
permit limit.5 most recent 2 years.7 years.7 8
Short term (≤ 24 hours) ............. Maximum allowable emission Annual level when actually op- Continuous operation, i.e., all
limit or federally enforceable erating, averaged over the hours of each time period
permit limit.5 most recent 2 years.7 under consideration (for all
data base).2
1 Terminology applicable to fuel burning sources; analogous terminology (e.g., #/throughput) may be used for other types of sources.
2 Ifoperation does not occur for all hours of the time period of consideration (e.g., 3 or 24 hours) and the source operation is constrained by a
federally enforceable permit condition, an appropriate adjustment to the modeled emission rate may be made (e.g., if operation is only 8 a.m. to
4 p.m. each day, only these hours will be modeled with emissions from the source. Modeled emissions should not be averaged across non-oper-
ating time periods.
3 Operating levels such as 50 percent and 75 percent of capacity should also be modeled to determine the load causing the highest concentra-
tion.
4 Includes existing facility to which modification is proposed if the emissions from the existing facility will not be affected by the modification.
Otherwise use the same parameters as for major modification.
5 See paragraph 8.2.3(c).
6 See paragraph 8.2.3(d).
7 Unless it is determined that this period is not representative.
8 For those permitted sources not in operation or that have not established an appropriate factor, continuous operation (i.e., 8760) should be
used.
9 Generally, the ambient impacts from non-nearby (background) sources can be represented by air quality data unless adequate data do not
exist.
8.2 Background Concentrations c. If the source is not isolated, it may be background concentration for the averaging
8.2.1 Discussion necessary to use a multi-source model to times of concern. Determine the mean
establish the impact of nearby sources. Since background concentration at each monitor by
a. Background concentrations are an sources don’t typically operate at their excluding values when the source in
essential part of the total air quality maximum allowable capacity (which may question is impacting the monitor. The mean
concentration to be considered in include the use of ‘‘dirtier’’ fuels), modeling annual background is the average of the
determining source impacts. Background air is necessary to express the potential annual concentrations so determined at each
quality includes pollutant concentrations due contribution of background sources, and this
to: (1) Natural sources; (2) nearby sources monitor. For shorter averaging periods, the
impact would not be captured via meteorological conditions accompanying the
other than the one(s) currently under monitoring. Background concentrations
consideration; and (3) unidentified sources. concentrations of concern should be
should be determined for each critical identified. Concentrations for meteorological
b. Typically, air quality data should be (concentration) averaging time.
used to establish background concentrations conditions of concern, at monitors not
in the vicinity of the source(s) under 8.2.2 Recommendations (Isolated Single impacted by the source in question, should
consideration. The monitoring network used Source) be averaged for each separate averaging time
for background determinations should a. Two options (paragraph (b) or (c) of this to determine the average background value.
conform to the same quality assurance and section) are available to determine the Monitoring sites inside a 90° sector
other requirements as those networks background concentration near isolated downwind of the source may be used to
established for PSD purposes.83 An sources. determine the area of impact. One hour
appropriate data validation procedure should b. Use air quality data collected in the concentrations may be added and averaged to
be applied to the data prior to use. vicinity of the source to determine the determine longer averaging periods.
c. If there are no monitors located in the with the sources that they back up would not assessing representativeness should be given
vicinity of the source, a ‘‘regional site’’ may be modeled as nearby sources. To reiterate, to adequate characterization of transport and
be used to determine background. A in these examples and other appropriate dispersion between the source(s) of concern
‘‘regional site’’ is one that is located away cases, the burden is on the primary source and areas where maximum design
from the area of interest but is impacted by being modeled to make the appropriate concentrations are anticipated to occur. The
similar natural and distant man-made demonstration to the satisfaction of the representativeness of data that were collected
sources. appropriate reviewing authority. off-site should be judged, in part, by
8.2.3 Recommendations (Multi-Source e. The impact of the nearby sources should comparing the surface characteristics in the
Areas) be examined at locations where interactions vicinity of the meteorological monitoring site
between the plume of the point source under with the surface characteristics that generally
a. In multi-source areas, two components consideration and those of nearby sources describe the analysis domain. The surface
of background should be determined: (plus natural background) can occur. characteristics input to AERMET should be
contributions from nearby sources and Significant locations include: (1) the area of based on the topographic conditions in the
contributions from other sources. maximum impact of the point source; (2) the vicinity of the meteorological tower.
b. Nearby Sources: All sources expected to area of maximum impact of nearby sources; Furthermore, since the spatial scope of each
cause a significant concentration gradient in and (3) the area where all sources combine variable could be different,
the vicinity of the source or sources under representativeness should be judged for each
to cause maximum impact. These locations
consideration for emission limit(s) should be variable separately. For example, for a
may be identified through trial and error
explicitly modeled. The number of such variable such as wind direction, the data may
analyses.
sources is expected to be small except in need to be collected very near plume height
f. Other Sources: That portion of the
unusual situations. Owing to both the to be adequately representative, whereas, for
background attributable to all other sources
uniqueness of each modeling situation and a variable such as temperature, data from a
(e.g., natural sources, minor sources and
the large number of variables involved in station several kilometers away from the
distant major sources) should be determined
identifying nearby sources, no attempt is source may in some cases be considered to
by the procedures found in subsection 89.2.2
made here to comprehensively define this be adequately representative.
or by application of a model using Table 8–
term. Rather, identification of nearby sources d. For long range transport modeling
calls for the exercise of professional 1 or 8–2.
assessments (subsection 6.2.3) or for
judgement by the appropriate reviewing 8.3 Meteorological Input Data assessments where the transport winds are
authority (paragraph 3.0(b)). This guidance is complex and the application involves a non-
a. The meteorological data used as input to
not intended to alter the exercise of that steady-state dispersion model (subsection
a dispersion model should be selected on the
judgement or to comprehensively define 7.2.8), use of output from prognostic
basis of spatial and climatological (temporal)
which sources are nearby sources. mesoscale meteorological models is
representativeness as well as the ability of
c. For compliance with the short-term and encouraged.84 85 86 Some diagnostic
the individual parameters selected to
annual ambient standards, the nearby sources meteorological processors are designed to
characterize the transport and dispersion
as well as the primary source(s) should be appropriately blend available NWS
conditions in the area of concern. The
evaluated using an appropriate Appendix A comparable meteorological observations,
model with the emission input data shown representativeness of the data is dependent
on: (1) The proximity of the meteorological local site specific meteorological
in Table 8–1 or 8–2. When modeling a nearby observations, and prognostic mesoscale
source that does not have a permit and the monitoring site to the area under
consideration; (2) the complexity of the meteorological data, using empirical
emission limit contained in the SIP for a relationships, to diagnostically adjust the
particular source category is greater than the terrain; (3) the exposure of the meteorological
monitoring site; and (4) the period of time wind field for mesoscale and local-scale
emissions possible given the source’s effects. These diagnostic adjustments can
maximum physical capacity to emit, the during which data are collected. The spatial
representativeness of the data can be sometimes be improved through the use of
‘‘maximum allowable emission limit’’ for strategically placed site specific
such a nearby source may be calculated as adversely affected by large distances between
the source and receptors of interest and the meteorological observations. The placement
the emission rate representative of the nearby of these special meteorological observations
source’s maximum physical capacity to emit, complex topographic characteristics of the
area. Temporal representativeness is a (often more than one location is needed)
considering its design specifications and
function of the year-to-year variations in involves expert judgement, and is specific to
allowable fuels and process materials.
weather conditions. Where appropriate, data the terrain and land use of the modeling
However, the burden is on the permit
representativeness should be viewed in terms domain. Acceptance for use of output from
applicant to sufficiently document what the
of the appropriateness of the data for prognostic mesoscale meteorological models
maximum physical capacity to emit is for
constructing realistic boundary layer profiles is contingent on concurrence by the
such a nearby source.
and three dimensional meteorological fields, appropriate reviewing authorities (paragraph
d. It is appropriate to model nearby sources
as described in paragraphs (c) and (d) below. 3.0(b)) that the data are of acceptable quality,
only during those times when they, by their
b. Model input data are normally obtained which can be demonstrated through
nature, operate at the same time as the
either from the National Weather Service or statistical comparisons with observations of
primary source(s) being modeled. Where a
primary source believes that a nearby source as part of a site specific measurement winds aloft and at the surface at several
does not, by its nature, operate at the same program. Local universities, Federal Aviation appropriate locations.
time as the primary source being modeled, Administration (FAA), military stations, 8.3.1 Length of Record of Meteorological
the burden is on the primary source to industry and pollution control agencies may Data
demonstrate to the satisfaction of the also be sources of such data. Some
recommendations for the use of each type of 8.3.1.1 Discussion
appropriate reviewing authority (paragraph
3.0(b)) that this is, in fact, the case. Whether data are included in this subsection. a. The model user should acquire enough
or not the primary source has adequately c. Regulatory application of AERMOD meteorological data to ensure that worst-case
demonstrated that fact is a matter of requires careful consideration of minimum meteorological conditions are adequately
professional judgement left to the discretion data for input to AERMET. Data represented in the model results. The trend
of the appropriate reviewing authority. The representativeness, in the case of AERMOD, toward statistically based standards suggests
following examples illustrate two cases in means utilizing data of an appropriate type a need for all meteorological conditions to be
which a nearby source may be shown not to for constructing realistic boundary layer adequately represented in the data set
operate at the same time as the primary profiles. Of paramount importance is the selected for model input. The number of
source(s) being modeled. Some sources are requirement that all meteorological data used years of record needed to obtain a stable
only used during certain seasons of the year. as input to AERMOD must be both laterally distribution of conditions depends on the
Those sources would not be modeled as and vertically representative of the transport variable being measured and has been
nearby sources during times in which they and dispersion within the analysis domain. estimated by Landsberg and Jacobs 87 for
do not operate. Similarly, emergency backup Where surface conditions vary significantly various parameters. Although that study
generators that never operate simultaneously over the analysis domain, the emphasis in indicates in excess of 10 years may be
required to achieve stability in the frequency data are being relied upon as the basis for data in close proximity to the actual site of
distributions of some meteorological characterizing the meteorological conditions, the source(s). Site specific measured data are
variables, such long periods are not a data base of at least 1 full-year of therefore preferred as model input, provided
reasonable for model input data. This is due meteorological data is required. If more data that appropriate instrumentation and quality
in part to the fact that hourly data in model are available, they should be used. Site assurance procedures are followed and that
input format are frequently not available for specific meteorological data may have to be the data collected are adequately
such periods and that hourly calculations of collected at multiple locations. Such data representative (free from inappropriate local
concentration for long periods may be should have been subjected to quality or microscale influences) and compatible
prohibitively expensive. Another study 88 assurance procedures as described in with the input requirements of the model to
compared various periods from a 17-year paragraph 8.3.3.2(a), and should be reviewed be used. It should be noted that, while site
data set to determine the minimum number for spatial and temporal representativeness. specific measurements are frequently made
of years of data needed to approximate the 8.3.2 National Weather Service Data ‘‘on-property’’ (i.e., on the source’s premises),
concentrations modeled with a 17-year acquisition of adequately representative site
period of meteorological data from one 8.3.2.1 Discussion specific data does not preclude collection of
station. This study indicated that the a. The NWS meteorological data are data from a location off property. Conversely,
variability of model estimates due to the routinely available and familiar to most collection of meteorological data on a
meteorological data input was adequately model users. Although the NWS does not source’s property does not of itself guarantee
reduced if a 5-year period of record of provide direct measurements of all the adequate representativeness. For help in
meteorological input was used. needed dispersion model input variables, determining representativeness of site
8.3.1.2 Recommendations methods have been developed and specific measurements, technical guidance 92
a. Five years of representative successfully used to translate the basic NWS is available. Site specific data should always
meteorological data should be used when data to the needed model input. Site specific be reviewed for representativeness and
estimating concentrations with an air quality measurements of model input parameters consistency by a qualified meteorologist.
model. Consecutive years from the most have been made for many modeling studies,
8.3.3.2 Recommendations
recent, readily available 5-year period are and those methods and techniques are
becoming more widely applied, especially in a. EPA guidance 92 provides
preferred. The meteorological data should be recommendations on the collection and use
adequately representative, and may be site situations such as complex terrain
applications, where available NWS data are of site specific meteorological data.
specific or from a nearby NWS station. Where
not adequately representative. However, Recommendations on characteristics, siting,
professional judgment indicates NWS-
there are many model applications where and exposure of meteorological instruments
collected ASOS (automated surface observing
NWS data are adequately representative, and and on data recording, processing,
stations) data are inadequate {for cloud cover
the applications still rely heavily on the NWS completeness requirements, reporting, and
observations}, the most recent 5 years of
data. archiving are also included. This publication
NWS data that are observer-based may be
b. Many models use the standard hourly should be used as a supplement to other
considered for use.
weather observations available from the limited guidance on these subjects.83 93 94
b. The use of 5 years of NWS
National Climatic Data Center (NCDC). These Detailed information on quality assurance is
meteorological data or at least l year of site
specific data is required. If one year or more observations are then preprocessed before also available.95 As a minimum, site specific
(including partial years), up to five years, of they can be used in the models. measurements of ambient air temperature,
site specific data is available, these data are transport wind speed and direction, and the
8.3.2.2 Recommendations
preferred for use in air quality analyses. Such variables necessary to estimate atmospheric
a. The preferred models listed in Appendix dispersion should be available in
data should have been subjected to quality A all accept as input the NWS meteorological
assurance procedures as described in meteorological data sets to be used in
data preprocessed into model compatible modeling. Care should be taken to ensure
subsection 8.3.3.2. form. If NWS data are judged to be
c. For permitted sources whose emission that meteorological instruments are located
adequately representative for a particular to provide representative characterization of
limitations are based on a specific year of modeling application, they may be used.
meteorological data, that year should be pollutant transport between sources and
NCDC makes available surface 89 90 and upper receptors of interest. The appropriate
added to any longer period being used (e.g.,
air 91 meteorological data in CD–ROM format. reviewing authority (paragraph 3.0(b)) is
5 years of NWS data) when modeling the
b. Although most NWS measurements are available to help determine the
facility at a later time.
made at a standard height of 10 meters, the appropriateness of the measurement
d. For LRT situations (subsection 6.2.3)
actual anemometer height should be used as locations.
and for complex wind situations (paragraph
input to the preferred model. Note that b. All site specific data should be reduced
7.2.8(a)), if only NWS or comparable
AERMOD at a minimum requires wind to hourly averages. Table 8–3 lists the wind
standard meteorological observations are
observations at a height above ground related parameters and the averaging time
employed, five years of meteorological data
between seven times the local surface requirements.
(within and near the modeling domain)
roughness height and 100 meters. c. Missing Data Substitution. After valid
should be used. Consecutive years from the
most recent, readily available 5-year period c. Wind directions observed by the data retrieval requirements have been met 92,
are preferred. Less than five, but at least National Weather Service are reported to the hours in the record having missing data
three, years of meteorological data (need not nearest 10 degrees. A specific set of randomly should be treated according to an established
be consecutive) may be used if mesoscale generated numbers has been developed for data substitution protocol provided that data
meteorological fields are available, as use with the preferred EPA models and from an adequately representative alternative
discussed in paragraph 8.3(d). These should be used with NWS data to ensure a site are available. Such protocols are usually
mesoscale meteorological fields should be lack of bias in wind direction assignments part of the approved monitoring program
used in conjunction with available standard within the models. plan. Data substitution guidance is provided
NWS or comparable meteorological d. Data from universities, FAA, military in Section 5.3 of reference 92. If no
observations within and near the modeling stations, industry and pollution control representative alternative data are available
domain. agencies may be used if such data are for substitution, the absent data should be
e. For solely LRT applications (subsection equivalent in accuracy and detail to the NWS coded as missing using missing data codes
6.2.3), if site specific meteorological data are data, and they are judged to be adequately appropriate to the applicable meteorological
available, these data may be helpful when representative for the particular application. pre-processor. Appropriate model options for
used in conjunction with available standard 8.3.3 Site Specific Data treating missing data, if available in the
NWS or comparable observations and model, should be employed.
mesoscale meteorological fields as described 8.3.3.1 Discussion d. Solar Radiation Measurements. Total
in paragraph 8.3.1.2(d). a. Spatial or geographical solar radiation or net radiation should be
f. For complex wind situations (paragraph representativeness is best achieved by measured with a reliable pyranometer or net
7.2.8(a)) where site specific meteorological collection of all of the needed model input radiometer, sited and operated in accordance
with established site specific meteorological categories, as originally defined, couple near- TABLE 8–3.—AVERAGING TIMES FOR
guidance.92 95 surface measurements of wind speed with
e. Temperature Measurements.
SITE SPECIFIC WIND AND TURBU-
subjectively determined insolation
Temperature measurements should be made assessments based on hourly cloud cover and LENCE MEASUREMENTS—Continued
at standard shelter height (2m) in accordance ceiling height observations. The wind speed
with established site specific meteorological measurements are made at or near 10m. The Averaging
guidance.92 insolation rate is typically assessed using Parameter time
f. Temperature Difference Measurements. (hour)
observations of cloud cover and ceiling
Temperature difference (DT) measurements height based on criteria outlined by Turner.70 Turbulence measurements (sE
should be obtained using matched It is recommended that the P–G stability
thermometers or a reliable thermocouple and sA) for use in stability
category be estimated using the Turner determinations ....................... 11
system to achieve adequate accuracy. Siting, method with site specific wind speed
probe placement, and operation of DT Turbulence measurements for
measured at or near 10m and representative direct input to dispersion
systems should be based on guidance found cloud cover and ceiling height.
in Chapter 3 of reference 92, and such models ................................... 1
Implementation of the Turner method, as
guidance should be followed when obtaining well as considerations in determining 1 To minimize meander effects in sA when
vertical temperature gradient data. AERMET representativeness of cloud cover and ceiling wind conditions are light and/or variable, de-
employs the Bulk Richardson scheme which height in cases for which site specific cloud termine the hourly average s value from four
requires measurements of temperature sequential 15-minute s’s according to the fol-
observations are unavailable, may be found lowing formula:
difference. To ensure correct application and in Section 6 of reference 92. In the absence
acceptance, AERMOD users should consult of requisite data to implement the Turner
with the appropriate Reviewing Authority
method, the SRDT method or wind σ 15
2
+ σ 15
2
+ σ 15
2
+ σ 15
2
before using the Bulk Richardson scheme for
their analysis.
fluctuation statistics (i.e., the sE and sA σ1− hr =
g. Winds Aloft. For simulation of plume
methods) may be used. 4
j. The SRDT method, described in Section
rise and dispersion of a plume emitted from 8.3.4 Treatment of Near-Calms and Calms
6.4.4.2 of reference 92, is modified slightly
a stack, characterization of the wind profile
from that published from earlier work 96 and 8.3.4.1 Discussion
up through the layer in which the plume
has been evaluated with three site specific
disperses is required. This is especially a. Treatment of calm or light and variable
data bases.97 The two methods of stability
important in complex terrain and/or complex wind poses a special problem in model
classification which use wind fluctuation
wind situations where wind measurements at applications since steady-state Gaussian
statistics, the sE and sA methods, are also
heights up to hundreds of meters above stack plume models assume that concentration is
described in detail in Section 6.4.4 of
base may be required in some circumstances. inversely proportional to wind speed.
reference 92 (note applicable tables in
For tall stacks when site specific data are Furthermore, concentrations may become
Section 6). For additional information on the
needed, these winds have been obtained unrealistically large when wind speeds less
wind fluctuation methods, several references
traditionally using meteorological sensors than 1 m/s are input to the model.
are available.98 99 100 101
mounted on tall towers. A feasible alternative Procedures have been developed to prevent
k. Meteorological Data Preprocessors. The
to tall towers is the use of meteorological the occurrence of overly conservative
following meteorological preprocessors are
remote sensing instruments (e.g., acoustic concentration estimates during periods of
recommended by EPA: AERMET,102
sounders or radar wind profilers) to provide calms. These procedures acknowledge that a
PCRAMMET,103 MPRM,104 METPRO,105 and
winds aloft, coupled with 10-meter towers to steady-state Gaussian plume model does not
CALMET 106 AERMET, which is patterned
provide the near-surface winds. (For specific apply during calm conditions, and that our
after MPRM, should be used to preprocess all
requirements for AERMOD and CTDMPLUS, knowledge of wind patterns and plume
data for use with AERMOD. Except for
see Appendix A.) Specifications for wind behavior during these conditions does not, at
applications that employ AERMOD,
measuring instruments and systems are present, permit the development of a better
PCRAMMET is the recommended
contained in reference 92. technique. Therefore, the procedures
meteorological preprocessor for use in
h. Turbulence. There are several dispersion disregard hours which are identified as calm.
applications employing hourly NWS data.
models that are capable of using direct The hour is treated as missing and a
MPRM is a general purpose meteorological
measurements of turbulence (wind convention for handling missing hours is
data preprocessor which supports regulatory
fluctuations) in the characterization of the recommended.
models requiring PCRAMMET formatted
vertical and lateral dispersion (e.g., b. AERMOD, while fundamentally a
(NWS) data. MPRM is available for use in
CTDMPLUS, AERMOD, and CALPUFF). For steady-state Gaussian plume model, contains
applications employing site specific
specific requirements for CTDMPLUS, algorithms for dealing with low wind speed
meteorological data. The latest version
AERMOD, and CALPUFF, see Appendix A. (near calm) conditions. As a result, AERMOD
(MPRM 1.3) has been configured to
For technical guidance on measurement and can produce model estimates for conditions
implement the SRDT method for estimating
processing of turbulence parameters, see when the wind speed may be less than 1 m/
P–G stability categories. METPRO is the
reference 92. When turbulence data are used s, but still greater than the instrument
required meteorological data preprocessor for
in this manner to directly characterize the threshold. Required input to AERMET, the
use with CTDMPLUS. CALMET is available
vertical and lateral dispersion, the averaging meteorological processor for AERMOD,
for use with applications of CALPUFF. All of
time for the turbulence measurements should includes a threshold wind speed and a
the above mentioned data preprocessors are
be one hour (Table 8–3). There are other reference wind speed. The threshold wind
available for downloading from EPA’s
dispersion models (e.g., BLP, and CALINE3) speed is typically the threshold of the
Internet SCRAM Web site (subsection 2.3).
that employ P–G stability categories for the instrument used to collect the wind speed
characterization of the vertical and lateral data. The reference wind speed is selected by
dispersion. Methods for using site specific TABLE 8–3.—AVERAGING TIMES FOR the model as the lowest level of non-missing
turbulence data for the characterization of P– SITE SPECIFIC WIND AND TURBU- wind speed and direction data where the
G stability categories are discussed in LENCE MEASUREMENTS speed is greater than the wind speed
reference 92. When turbulence data are used threshold, and the height of the measurement
in this manner to determine the P–G stability Averaging is between seven times the local surface
category, the averaging time for the Parameter time roughness and 100 meters. If the only valid
turbulence measurements should be 15 (hour) observation of the reference wind speed
minutes. between these heights is less than the
i. Stability Categories. For dispersion Surface wind speed (for use in threshold, the hour is considered calm, and
models that employ P–G stability categories stability determinations) ........ 1 no concentration is calculated. None of the
for the characterization of the vertical and Transport direction .................... 1 observed wind speeds in a measured wind
ER09NO05.002</MATH>
lateral dispersion, the P–G stability Dilution wind speed .................. 1 profile that are less than the threshold speed
are used in construction of the modeled wind uncertainty in the regulatory process. The comparison of model concentration estimates
speed profile in AERMOD. Second (EPA) Conference on Air Quality with measured air quality data.118 The
8.3.4.2 Recommendations Modeling, August 1982 110, was devoted to statement of accuracy is based on statistical
that subject. tests or performance measures such as bias,
a. Hourly concentrations calculated with b. To better deduce the statistical noise, correlation, etc.11 However,
steady-state Gaussian plume models using significance of differences seen in model information that allows a distinction between
calms should not be considered valid; the performance in the face of unaccounted for contributions of the various elements of
wind and concentration estimates for these uncertainties and variations, investigators inherent and reducible uncertainty is only
hours should be disregarded and considered have more recently explored the use of now beginning to emerge.16 As a result most
to be missing. Critical concentrations for bootstrap techniques.111 112 Work is discussions of the accuracy of models make
3-, 8-, and 24-hour averages should be underway to develop a new generation of no quantitative distinction between (1)
calculated by dividing the sum of the hourly evaluation metrics 16 that takes into account limitations of the model versus (2)
concentrations for the period by the number the statistical differences (in error limitations of the data base and of knowledge
of valid or non-missing hours. If the total distributions) between model predictions and concerning atmospheric variability. The
number of valid hours is less than 18 for 24- observations.113 Even though the procedures reader should be aware that statements on
hour averages, less than 6 for 8-hour averages and measures are still evolving to describe model accuracy and uncertainty may imply
or less than 3 for 3-hour averages, the total performance of models that characterize the need for improvements in model
concentration should be divided by 18 for the atmospheric fate, transport and performance that even the ‘‘perfect’’ model
24-hour average, 6 for the 8-hour average and diffusion 114 115 116, there has been general could not satisfy.
3 for the 3-hour average. For annual averages, acceptance of a need to address the
the sum of all valid hourly concentrations is 9.1.2 Studies of Model Accuracy
uncertainties inherent in atmospheric
divided by the number of non-calm hours processes. a. A number of studies 119 120 have been
during the year. AERMOD has been coded to conducted to examine model accuracy,
implement these instructions. For models 9.1.1 Overview of Model Uncertainty particularly with respect to the reliability of
listed in Appendix A, a post-processor a. Dispersion models generally attempt to short-term concentrations required for
computer program, CALMPRO 107 has been estimate concentrations at specific sites that ambient standard and increment evaluations.
prepared, is available on the SCRAM Internet really represent an ensemble average of The results of these studies are not
Web site (subsection 2.3), and should be numerous repetitions of the same event.16 surprising. Basically, they confirm what
used. The event is characterized by measured or expert atmospheric scientists have said for
b. Stagnant conditions that include ‘‘known’’ conditions that are input to the some time: (1) Models are more reliable for
extended periods of calms often produce models, e.g., wind speed, mixed layer height, estimating longer time-averaged
high concentrations over wide areas for surface heat flux, emission characteristics, concentrations than for estimating short-term
relatively long averaging periods. The etc. However, in addition to the known concentrations at specific locations; and (2)
standard steady-state Gaussian plume models conditions, there are unmeasured or the models are reasonably reliable in
are often not applicable to such situations. unknown variations in the conditions of this estimating the magnitude of highest
When stagnation conditions are of concern, event, e.g., unresolved details of the concentrations occurring sometime,
other modeling techniques should be atmospheric flow such as the turbulent somewhere within an area. For example,
considered on a case-by-case basis (see also velocity field. These unknown conditions, errors in highest estimated concentrations of
subsection 7.2.8). may vary among repetitions of the event. As ± 10 to 40 percent are found to be
c. When used in steady-state Gaussian a result, deviations in observed typical 121 122, i.e., certainly well within the
plume models, measured site specific wind concentrations from their ensemble average, often quoted factor-of-two accuracy that has
speeds of less than 1 m/s but higher than the and from the concentrations estimated by the long been recognized for these models.
response threshold of the instrument should model, are likely to occur even though the However, estimates of concentrations that
be input as 1 m/s; the corresponding wind known conditions are fixed. Even with a occur at a specific time and site, are poorly
direction should also be input. Wind perfect model that predicts the correct correlated with actually observed
observations below the response threshold of ensemble average, there are likely to be concentrations and are much less reliable.
the instrument should be set to zero, with the deviations from the observed concentrations b. As noted above, poor correlations
input file in ASCII format. For input to in individual repetitions of the event, due to between paired concentrations at fixed
AERMOD, no adjustment should be made to variations in the unknown conditions. The stations may be due to ‘‘reducible’’
the site specific wind data. In all cases statistics of these concentration residuals are uncertainties in knowledge of the precise
involving steady-state Gaussian plume termed ‘‘inherent’’ uncertainty. Available plume location and to unquantified inherent
models, calm hours should be treated as evidence suggests that this source of uncertainties. For example, Pasquill 123
missing, and concentrations should be uncertainty alone may be responsible for a estimates that, apart from data input errors,
calculated as in paragraph (a) of this typical range of variation in concentrations of maximum ground-level concentrations at a
subsection. as much as ±50 percent.117 given hour for a point source in flat terrain
b. Moreover, there is ‘‘reducible’’ could be in error by 50 percent due to these
9.0 Accuracy and Uncertainty of Models uncertainty 108 associated with the model and uncertainties. Uncertainty of five to 10
its input conditions; neither models nor data degrees in the measured wind direction,
9.1 Discussion
bases are perfect. Reducible uncertainties are which transports the plume, can result in
a. Increasing reliance has been placed on caused by: (1) Uncertainties in the input concentration errors of 20 to 70 percent for
concentration estimates from models as the values of the known conditions (i.e., a particular time and location, depending on
primary basis for regulatory decisions emission characteristics and meteorological stability and station location. Such
concerning source permits and emission data); (2) errors in the measured uncertainties do not indicate that an
control requirements. In many situations, concentrations which are used to compute estimated concentration does not occur, only
such as review of a proposed source, no the concentration residuals; and (3) that the precise time and locations are in
practical alternative exists. Therefore, there is inadequate model physics and formulation. doubt.
an obvious need to know how accurate The ‘‘reducible’’ uncertainties can be
models really are and how any uncertainty in minimized through better (more accurate and 9.1.3 Use of Uncertainty in Decision-
the estimates affects regulatory decisions. more representative) measurements and Making
During the 1980’s, attempts were made to better model physics. a. The accuracy of model estimates varies
encourage development of standardized c. To use the terminology correctly, with the model used, the type of application,
evaluation methods.11 108 EPA recognized the reference to model accuracy should be and site specific characteristics. Thus, it is
need for incorporating such information and limited to that portion of reducible desirable to quantify the accuracy or
has sponsored workshops 109 on model uncertainty which deals with the physics and uncertainty associated with concentration
accuracy, the possible ways to quantify the formulation of the model. The accuracy estimates used in decision-making.
accuracy, and on considerations in the of the model is normally determined by an Communications between modelers and
incorporation of model accuracy and evaluation procedure which involves the decision-makers must be fostered and further
developed. Communications concerning performance measures, including measures they can provide additional information on
concentration estimates currently exist in of difference (or residuals) such as bias, the effect of inaccuracies in the data bases
most cases, but the communications dealing variance of difference and gross variability of and on the uncertainty in model estimates.
with the accuracy of models and its meaning the difference, and measures of correlation Sensitivity analyses can aid in determining
to the decision-maker are limited by the lack such as time, space, and time and space the effect of inaccuracies of variations or
of a technical basis for quantifying and combined as recommended by the AMS uncertainties in the data bases on the range
directly including uncertainty in decisions. Woods Hole Workshop 11, were generally of likely concentrations. Uncertainty analyses
Procedures for quantifying and interpreting followed. Third, more specific information can aid in determining the range of likely
uncertainty in the practical application of has been provided for justifying the site concentration values, resulting from
such concepts are only beginning to evolve; specific use of alternative models in uncertainties in the model inputs, the model
much study is still required.108 109 110 124 125 previously cited EPA guidance 15, and new formulations, and parameterizations. Such
b. In all applications of models an effort is models are under consideration and information may be used to determine source
encouraged to identify the reliability of the review.16 Together these documents provide impact and to evaluate control strategies.
model estimates for that particular area and methods that allow a judgement to be made Where possible, information from such
to determine the magnitude and sources of as to what models are most appropriate for sensitivity analyses should be made available
error associated with the use of the model. a specific application. For the present, to the decision-maker with an appropriate
The analyst is responsible for recognizing performance and the theoretical evaluation of interpretation of the effect on the critical
and quantifying limitations in the accuracy, models are being used as an indirect means concentrations.
precision and sensitivity of the procedure. to quantify one element of uncertainty in air
Information that might be useful to the pollution regulatory decisions. 9.2 Recommendations
decision-maker in recognizing the b. EPA has participated in a series of a. No specific guidance on the
seriousness of potential air quality violations conferences entitled, ‘‘Harmonisation within quantification of model uncertainty for use in
includes such model accuracy estimates as Atmospheric Dispersion Modelling for decision-making is being given at this time.
accuracy of peak predictions, bias, noise, Regulatory Purposes.’’ 128 for the purpose of As procedures for considering uncertainty
correlation, frequency distribution, spatial promoting the development of improved develop and become implementable, this
extent of high concentration, etc. Both space/ methods for the characterization of model guidance will be changed and expanded. For
time pairing of estimates and measurements performance. There is a consensus the present, continued use of the ‘‘best
and unpaired comparisons are developing on what should be considered in estimate’’ is acceptable; however, in specific
recommended. Emphasis should be on the the evaluation of air quality models 129, circumstances for O3, PM–2.5 and regional
highest concentrations and the averaging namely quality assurance planning, haze, additional information and/or
times of the standards or increments of documentation and scrutiny should be procedures may be appropriate.32 33
concern. Where possible, confidence consistent with the intended use, and should
10.0 Regulatory Application of Models
intervals about the statistical values should include:
be provided. However, while such • Scientific peer review; 10.1 Discussion
information can be provided by the modeler • Supportive analyses (diagnostic a. Procedures with respect to the review
to the decision-maker, it is unclear how this evaluations, code verification, sensitivity and and analysis of air quality modeling and data
information should be used to make an air uncertainty analyses); analyses in support of SIP revisions, PSD
pollution control decision. Given a range of • Diagnostic and performance evaluations permitting or other regulatory requirements
possible outcomes, it is easiest and tends to with data obtained in trial locations, and need a certain amount of standardization to
ensure consistency if the decision-maker • Statistical performance evaluations in ensure consistency in the depth and
confines his judgement to use of the ‘‘best the circumstances of the intended comprehensiveness of both the review and
estimate’’ provided by the modeler (i.e., the applications. the analysis itself. This section recommends
design concentration estimated by a model Performance evaluations and diagnostic procedures that permit some degree of
recommended in the Guideline or an evaluations assess different qualities of how standardization while at the same time
alternate model of known accuracy). This is well a model is performing, and both are allowing the flexibility needed to assure the
an indication of the practical limitations needed to establish credibility within the technically best analysis for each regulatory
imposed by current abilities of the technical client and scientific community. Performance application.
community. evaluations allow us to decide how well the b. Dispersion model estimates, especially
c. To improve the basis for decision- model simulates the average temporal and with the support of measured air quality
making, EPA has developed and is spatial patterns seen in the observations, and data, are the preferred basis for air quality
continuing to study procedures for employ large spatial/temporal scale data sets demonstrations. Nevertheless, there are
determining the accuracy of models, (e.g., national data sets). Performance instances where the performance of
quantifying the uncertainty, and expressing evaluations also allow determination of recommended dispersion modeling
confidence levels in decisions that are made relative performance of a model in techniques, by comparison with observed air
concerning emissions controls.126 127 comparison with alternative modeling quality data, may be shown to be less than
However, work in this area involves
systems. Diagnostic evaluations allow acceptable. Also, there may be no
‘‘breaking new ground’’ with slow and
determination of a model capability to recommended modeling procedure suitable
sporadic progress likely. As a result, it may
simulate individual processes that affect the for the situation. In these instances, emission
be necessary to continue using the ‘‘best
results, and usually employ smaller spatial/ limitations may be established solely on the
estimate’’ until sufficient technical progress
temporal scale date sets (e.g., field studies). basis of observed air quality data as would
has been made to meaningfully implement
Diagnostic evaluations allow us to decide if be applied to a modeling analysis. The same
such concepts dealing with uncertainty.
we get the right answer for the right reason. care should be given to the analyses of the
9.1.4 Evaluation of Models The objective comparison of modeled air quality data as would be applied to a
a. A number of actions have been taken to concentrations with observed field data modeling analysis.
ensure that the best model is used correctly provides only a partial means for assessing c. The current NAAQS for SO2 and CO are
for each regulatory application and that a model performance. Due to the limited both stated in terms of a concentration not to
model is not arbitrarily imposed. First, the supply of evaluation data sets, there are be exceeded more than once a year. There is
Guideline clearly recommends the most severe practical limits in assessing model only an annual standard for NO2 and a
appropriate model be used in each case. performance. For this reason, the conclusions quarterly standard for Pb. Standards for fine
Preferred models, based on a number of reached in the science peer reviews and the particulate matter (PM–2.5) are expressed in
factors, are identified for many uses. General supportive analyses have particular relevance terms of both long-term (annual) and short-
guidance on using alternatives to the in deciding whether a model will be useful term (daily) averages. The long-term standard
preferred models is also provided. Second, for its intended purposes. is calculated using the three year average of
the models have been subjected to a c. To extend information from diagnostic the annual averages while the short-term
systematic performance evaluation and a and performance evaluations, sensitivity and standard is calculated using the three year
peer scientific review. Statistical uncertainty analyses are encouraged since average of the 98th percentile of the daily
average concentration. For PM–10, the d. Regional Offices should require permit highly desirable. The design of the network,
convention is to compare the arithmetic applicants to incorporate the pollutant the number, type and location of the
mean, averaged over 3 consecutive years, contributions of all sources into their monitors, the sampling period, averaging
with the concentration specified in the analysis. Where necessary this may include time as well as the need for meteorological
NAAQS (50 µg/m3). The 24-hour NAAQS emissions associated with growth in the area monitoring or the use of mobile sampling or
(150 µg/m3) is met if, over a 3-year period, of impact of the new or modified source. PSD plume tracking techniques, should all be
there is (on average) no more than one air quality assessments should consider the specified in the protocol and agreed upon
exceedance per year. As noted in subsection amount of the allowable air quality prior to start-up of the network.
7.2.1.1, the modeled compliance for this increment that has already been consumed 10.2.3 Emission Limits
NAAQS is based on the highest 6th highest by other sources. Therefore, the most recent
concentration over 5 years. For ozone the source applicant should model the existing 10.2.3.1 Design Concentrations
short term 1-hour standard is expressed in or permitted sources in addition to the one a. Emission limits should be based on
terms of an expected exceedance limit while currently under consideration. This would concentration estimates for the averaging
the short term 8-hour standard is expressed permit the use of newly acquired data or time that results in the most stringent control
in terms of a three year average of the annual improved modeling techniques if such have requirements. The concentration used in
fourth highest daily maximum 8-hour value. become available since the last source was specifying emission limits is called the
The NAAQS are subjected to extensive permitted. When remodeling, the worst case design value or design concentration and is
review and possible revision every 5 years. used in the previous modeling analysis a sum of the concentration contributed by the
d. This section discusses general should be one set of conditions modeled in primary source, other applicable sources,
requirements for concentration estimates and the new analysis. All sources should be
identifies the relationship to emission limits. and—for NAAQS assessments—the
modeled for each set of meteorological background concentration.
The following recommendations apply to: (1) conditions selected.
Revisions of State Implementation Plans and b. To determine the averaging time for the
10.2.2 Use of Measured Data in Lieu of design value, the most restrictive NAAQS or
(2) the review of new sources and the
Model Estimates PSD increment, as applicable, should be
prevention of significant deterioration (PSD).
a. Modeling is the preferred method for identified. For a NAAQS assessment, the
10.2 Recommendations determining emission limitations for both averaging time for the design value is
10.2.1 Analysis Requirements new and existing sources. When a preferred determined by calculating, for each averaging
model is available, model results alone time, the ratio of the difference between the
a. Every effort should be made by the
(including background) are sufficient. applicable NAAQS (S) and the background
Regional Office to meet with all parties
Monitoring will normally not be accepted as concentration (B) to the (model) predicted
involved in either a SIP revision or a PSD
permit application prior to the start of any the sole basis for emission limitation. In concentration (P) (i.e., (S–B)/P). For a PSD
work on such a project. During this meeting, some instances when the modeling technique increment assessment, the averaging time for
a protocol should be established between the available is only a screening technique, the the design value is determined by
preparing and reviewing parties to define the addition of air quality data to the analysis calculating, for each averaging time, the ratio
procedures to be followed, the data to be may lend credence to model results. of the applicable PSD increment (I) and the
collected, the model to be used, and the b. There are circumstances where there is model-predicted concentration (P) (i.e., I/P).
analysis of the source and concentration data. no applicable model, and measured data may The averaging time with the lowest ratio
An example of requirements for such an need to be used. However, only in the case identifies the most restrictive standard or
effort is contained in the Air Quality of a NAAQS assessment for an existing increment. If the annual average is the most
Analysis Checklist posted on EPA’s Internet source should monitoring data alone be a restrictive, the highest estimated annual
SCRAM Web site (subsection 2.3). This basis for emission limits. In addition, the average concentration from one or a number
checklist suggests the level of detail required following items (i–vi) should be considered of years of data is the design value. When
to assess the air quality resulting from the prior to the acceptance of the measured data: short term standards are most restrictive, it
proposed action. Special cases may require i. Does a monitoring network exist for the may be necessary to consider a broader range
additional data collection or analysis and this pollutants and averaging times of concern? of concentrations than the highest value. For
should be determined and agreed upon at ii. Has the monitoring network been example, for pollutants such as SO2, the
this preapplication meeting. The protocol designed to locate points of maximum highest, second-highest concentration is the
should be written and agreed upon by the concentration? design value. For pollutants with statistically
parties concerned, although a formal legal iii. Do the monitoring network and the data based NAAQS, the design value is found by
document is not intended. Changes in such reduction and storage procedures meet EPA determining the more restrictive of: (1) The
a protocol are often required as the data monitoring and quality assurance short-term concentration over the period
collection and analysis progresses. However, requirements? specified in the standard, or (2) the long-term
the protocol establishes a common iv. Do the data set and the analysis allow concentration that is not expected to exceed
understanding of the requirements. impact of the most important individual the long-term NAAQS. Determination of
b. An air quality analysis should begin sources to be identified if more than one design values for PM–10 is presented in more
with a screening model to determine the source or emission point is involved? detail in EPA guidance.34
potential of the proposed source or control v. Is at least one full year of valid ambient
10.2.3.2 NAAQS Analyses for New or
strategy to violate the PSD increment or data available?
Modified Sources
NAAQS. For traditional stationary sources, vi. Can it be demonstrated through the
EPA guidance 24 should be followed. comparison of monitored data with model a. For new or modified sources predicted
Guidance is also available for mobile results that available models are not to have a significant ambient impact 83 and to
sources.48 applicable? be located in areas designated attainment or
c. If the concentration estimates from c. The number of monitors required is a unclassifiable for the SO2, Pb, NO2, or CO
screening techniques indicate a significant function of the problem being considered. NAAQS, the demonstration as to whether the
impact or that the PSD increment or NAAQS The source configuration, terrain source will cause or contribute to an air
may be approached or exceeded, then a more configuration, and meteorological variations quality violation should be based on: (1) The
refined modeling analysis is appropriate and all have an impact on number and placement highest estimated annual average
the model user should select a model of monitors. Decisions can only be made on concentration determined from annual
according to recommendations in Sections 4– a case-by-case basis. Guidance is available for averages of individual years; or (2) the
8. In some instances, no refined technique establishing criteria for demonstrating that a highest, second-highest estimated
may be specified in this guide for the model is not applicable? concentration for averaging times of 24-hours
situation. The model user is then encouraged d. Sources should obtain approval from the or less; and (3) the significance of the spatial
to submit a model developed specifically for appropriate reviewing authority (paragraph and temporal contribution to any modeled
the case at hand. If that is not possible, a 3.0(b)) for the monitoring network prior to violation. For Pb, the highest estimated
screening technique may supply the needed the start of monitoring. A monitoring concentration based on an individual
results. protocol agreed to by all concerned parties is calendar quarter averaging period should be
used. Background concentrations should be 11.0 Bibliography a 1980, Raleigh, NC. Bulletin of the American
added to the estimated impact of the source. American Meteorological Society. Meteorological Society, 62(2): 255–261.
The most restrictive standard should be used Symposia on Turbulence, Diffusion, and Air Hunt, J.C.R., R.G. Holroyd, D.J. Carruthers,
in all cases to assess the threat of an air Pollution (1st–10th); 1971–1992. Symposia A.G. Robins, D.D. Apsley, F.B. Smith and D.J.
quality violation. For new or modified on Boundary Layers & Turb. 11th–12th; Thompson, 1990. Developments in Modeling
sources predicted to have a significant 1995–1997. Boston, MA. Air Pollution for Regulatory Uses. In
ambient impact 83 in areas designated American Meteorological Society, 1977– Proceedings of the 18th NATO/CCMS
attainment or unclassifiable for the PM–10 1998. Joint Conferences on Applications of International Technical Meeting on Air
NAAQS, the demonstration of whether or not Air Pollution Meteorology (1st–10th). Pollution Modeling and its Application,
the source will cause or contribute to an air Sponsored by the American Meteorological Vancouver, Canada. Also In Air Pollution
quality violation should be based on Society and the Air & Waste Management Modeling and its Application VIII (1991). H.
Association. Boston, MA. van Dop and D.G. Steyn, eds. Plenum Press,
sufficient data to show whether: (1) The
American Meteorological Society, 1978. New York, NY. pp. 17–59
projected 24-hour average concentrations
Accuracy of Dispersion Models. Bulletin of Pasquill, F. and F.B. Smith, 1983.
will exceed the 24-hour NAAQS more than
the American Meteorological Society, 59(8): Atmospheric Diffusion, 3rd Edition. Ellis
once per year, on average; (2) the expected Horwood Limited, Chichester, West Sussex,
(i.e., average) annual mean concentration will 1025–1026.
American Meteorological Society, 1981. England, 438pp.
exceed the annual NAAQS; and (3) the Randerson, D., Ed., 1984. Atmospheric
source contributes significantly, in a Air Quality Modeling and the Clean Air Act:
Recommendations to EPA on Dispersion Science and Power Production. DOE/TIC
temporal and spatial sense, to any modeled 2760l. Office of Scientific and Technical
Modeling for Regulatory Applications.
violation. Information, U.S. Department of Energy, Oak
Boston, MA.
10.2.3.3 PSD Air Quality Increments and Briggs, G.A., 1969. Plume Rise. U.S. Ridge, TN.
Impacts Atomic Energy Commission Critical Review Scire, J.S. and L.L. Schulman, 1980:
Series, Oak Ridge National Laboratory, Oak Modeling plume rise from low-level buoyant
a. The allowable PSD increments for
Ridge, TN. line and point sources. AMS/APCA Second
criteria pollutants are established by
Drake, R.L. and S.M. Barrager, 1979. Joint Conference on Applications of Air
regulation and cited in 40 CFR 51.166. These
Mathematical Models for Atmospheric Pollution Meteorology, March 24–27, New
maximum allowable increases in pollutant Orleans, LA.
Pollutants. EPRI EA–1131. Electric Power
concentrations may be exceeded once per Smith, M.E., Ed., 1973. Recommended
Research Institute, Palo Alto, CA.
year at each site, except for the annual Guide for the Prediction of the Dispersion of
Environmental Protection Agency, 1978.
increment that may not be exceeded. The Workbook for Comparison of Air Quality Airborne Effluents. The American Society of
highest, second-highest increase in estimated Models. Publication No. EPA–450/2–78–028a Mechanical Engineers, New York, NY.
concentrations for the short term averages as and b. Office of Air Quality Planning & Stern, A.C., Ed., 1976. Air Pollution, Third
determined by a model should be less than Standards, Research Triangle Park, NC. Edition, Volume I: Air Pollutants, Their
or equal to the permitted increment. The Erisman J.W., Van Pul A. and Wyers P. Transformation and Transport. Academic
modeled annual averages should not exceed (1994) Parameterization of surface resistance Press, New York, NY.
the increment. for the quantification of atmospheric Turner, D.B., 1979. Atmospheric
b. Screening techniques defined in deposition of acidifying pollutants and Dispersion Modeling: A Critical Review.
subsection 4.2.1 can sometimes be used to ozone. Atmos. Environ., 28: 2595–2607. Journal of the Air Pollution Control
estimate short term incremental Fox, D.G., and J.E. Fairobent, 1981. NCAQ Association, 29(5): 502–519.
concentrations for the first new source that Panel Examines Uses and Limitations of Air Venkatram, A. and J.C. Wyngaard, Editors,
triggers the baseline in a given area. Quality Models. Bulletin of the American 1988. Lectures on Air Pollution Modeling.
However, when multiple increment- Meteorological Society, 62(2): 218–221. American Meteorological Society, Boston,
consuming sources are involved in the Gifford, F.A., 1976. Turbulent Diffusion MA. 390pp.
calculation, the use of a refined model with Typing Schemes: A Review. Nuclear Safety,
17(1): 68–86. 12.0 References
at least 1 year of site specific or 5 years of
(off-site) NWS data is normally required Gudiksen, P.H., and M.H. Dickerson, Eds., 1. Code of Federal Regulations; Title 40
(subsection 8.3.1.2). In such cases, sequential Executive Summary: Atmospheric Studies in (Protection of Environment). Sections 51.112,
modeling must demonstrate that the Complex Terrain Technical Progress Report 51.117, 51.150, 51.160.
allowable increments are not exceeded FY–1979 Through FY–1983. Lawrence 2. Environmental Protection Agency, 1990.
temporally and spatially, i.e., for all receptors Livermore National Laboratory, Livermore, New Source Review Workshop Manual:
for each time period throughout the year(s) CA. (Docket Reference No. II–I–103). Prevention of Significant Deterioration and
(time period means the appropriate PSD Hanna, S.R., G.A. Briggs, J. Deardorff, B.A. Nonattainment Area Permitting (Draft). Office
averaging time, e.g., 3-hour, 24-hour, etc.). Egan, G.A. Gifford and F. Pasquill, 1977. of Air Quality Planning & Standards,
c. The PSD regulations require an AMS Workshop on Stability Classification Research Triangle Park, NC. (Available at:
estimation of the SO2, particulate matter Schemes And Sigma Curves—Summary of http://www.epa.gov/ttn/nsr/)
(PM–10), and NO2 impact on any Class I area. Recommendations. Bulletin of the American 3. Code of Federal Regulations; Title 40
Normally, steady-state Gaussian plume Meteorological Society, 58(12): 1305–1309. (Protection of Environment). Sections 51.166
models should not be applied at distances Hanna, S.R., G.A. Briggs and R.P. Hosker, and 52.21.
greater than can be accommodated by the Jr., 1982. Handbook on Atmospheric 4. Code of Federal Regulations (Title 40,
Diffusion. Technical Information Center, U.S. Part 50): Protection of the Environment;
steady state assumptions inherent in such
Department of Energy, Washington, D.C. National Primary and Secondary Ambient
models. The maximum distance for refined
Haugen, D.A., Workshop Coordinator, Air Quality Standards.
steady-state Gaussian plume model
1975. Lectures on Air Pollution and 5. Environmental Protection Agency, 1988.
application for regulatory purposes is
Environmental Impact Analyses. Sponsored Model Clearinghouse: Operational Plan
generally considered to be 50km. Beyond the
by the American Meteorological Society, (Revised). Staff Report. Office of Air Quality
50km range, screening techniques may be Planning & Standards, Research Triangle
Boston, MA.
used to determine if more refined modeling Park, NC. (Docket No. A–88–04, II–J–1)
Hoffnagle, G.F., M.E. Smith, T.V. Crawford
is needed. If refined models are needed, long and T.J. Lockhart, 1981. On-site 6. Environmental Protection Agency, 1980.
range transport models should be considered Meteorological Instrumentation Guidelines on Air Quality Models. Federal
in accordance with subsection 6.2.3. As Requirements to Characterize Diffusion from Register, 45(61): 20157–20158.
previously noted in Sections 3 and 7, the Point Sources—A Workshop, 15–17 January 7. Scire, J.S. and L.L. Schulman, 1981.
need to involve the Federal Land Manager in Evaluation of the BLP and ISC Models with
decisions on potential air quality impacts, a The documents listed here are major sources of SF6 Tracer Data and SO2 Measurements at
particularly in relation to PSD Class I areas, supplemental information on the theory and Aluminum Reduction Plants. APCA
cannot be overemphasized. application of mathematical air quality models. Specialty Conference on Dispersion
Modeling for Complex Sources, St. Louis, Florida. J. Climate and Appl. Met., 24(11): Other Analyses in Attainment
MO. 1111—1207. Demonstrations for the PM–2.5 NAAQS and
8. Environmental Protection Agency, 1986. 22. Environmental Protection Agency, Regional Haze Goals. Office of Air Quality
Evaluation of Mobile Source Air Quality 2002. AERMOD: Description of Model Planning & Standards, Research Triangle
Simulation Models. Publication No. EPA– Formulation. Research Triangle Park, NC. Park, NC. (As of May 2005, this document
450/4–86–002. Office of Air Quality Planning EPA Report No. EPA–454/R–02–002d; April has not been finalized; latest version
& Standards, Research Triangle Park, NC. 2002; AND Cimorelli, A. et al., 2005. available on SCRAM Web site as draft-
(NTIS No. PB 86–167293) AERMOD: A Dispersion Model for Industrial pm.pdf; see subsection 2.3)
9. Strimaitis, D.G., J.S. Scire and J.C. Source Applications. Part I: General Model 34. Environmental Protection Agency,
Chang. 1998. Evaluation of the CALPUFF Formulation and Boundary Layer 1987. PM–10 SIP Development Guideline.
Dispersion Model with Two Power Plant Characterization. Journal of Applied Publication No. EPA–450/2–86–001. Office of
Data Sets. Tenth Joint Conference on the Meteorology, 44(5): 682–693. Air Quality Planning & Standards, Research
Application of Air Pollution Meteorology, 23. L.L. Schulman, D.G. Strimaitis and J.S. Triangle Park, NC. (NTIS No. PB 87–206488)
Phoenix, Arizona. American Meteorological Scire, 2002. Development and evaluation of 35. U.S. Forest Service, 1996. User
Society, Boston, MA. January 11–16, 1998. the PRIME plume rise and building Assessment of Smoke-Dispersion Models for
10. Environmental Protection Agency, downwash model. Journal of the Air & Waste Wildland Biomass Burning. USDA, Pacific
2003. AERMOD: Latest Features and Management Association, 50: 378–390. Northwest Research Station, Portland, OR.
Evaluation Results. Publication No. EPA– 24. Environmental Protection Agency, General Technical Report PNW–GTR–379.
454/R–03–003. U.S. Environmental 1992. Screening Procedures for Estimating 30pp. (NTIS No. PB 97–163380)
Protection Agency, Research Triangle Park, the Air Quality Impact of Stationary Sources, 36. Hanrahan, P.L., 1999. The Polar
NC. (Available at http://www.epa.gov/ Revised. Publication No. EPA–454/R–92– Volume Polar Ratio Method for Determining
scram001/) 019. U.S. Environmental Protection Agency, NO2 / NOX Ratios in Modeling—Part I:
11. Fox, D.G., 1981. Judging Air Quality Research Triangle Park, NC. (NTIS No. PB Methodology. J. Air & Waste Manage. Assoc.,
Model Performance. Bulletin of the American 93–219095) 49: 1324–1331.
Meteorological Society, 62(5): 599–609. 25. Environmental Protection Agency, 37. Environmental Protection Agency,
12. American Meteorological Society, 1983. 1995. SCREEN3 User’s Guide. Publication 1997. Guidance for Siting Ambient Air
Synthesis of the Rural Model Reviews. No. EPA–454/B–95–004. U.S. Environmental Monitors around Stationary Lead Sources.
Publication No. EPA–600/3–83–108. Office of Protection Agency, Research Triangle Park, Publication No. EPA–454/R–92–009R. Office
Research & Development, Research Triangle NC. (NTIS No. PB 95–222766) of Air Quality Planning & Standards,
Park, NC. (NTIS No. PB 84–121037) 26. Perry, S.G., D.J. Burns and A.J. Research Triangle Park, NC. (NTIS No. PB
13. Allwine, K.J., W.F. Dabberdt and L.L. Cimorelli, 1990. User’s Guide to CTDMPLUS: 97–208094)
Simmons. 1998. Peer Review of the Volume 2. The Screening Mode (CTSCREEN). 38. Environmental Protection Agency,
CALMET/CALPUFF Modeling System. Publication No. EPA–600/8–90–087. U.S. 1993. Lead Guideline Document. Publication
Prepared by the KEVRIC Company, Inc. Environmental Protection Agency, Research No. EPA–452/R–93–009. Office of Air
under EPA Contract No. 68–D–98–092 for Triangle Park, NC. (NTIS No. PB 91–136564) Quality Planning & Standards, Research
Environmental Protection Agency, Research 27. Mills, M.T., R.J. Paine, E.A. Insley and Triangle Park, NC. (NTIS No. PB 94–111846)
Triangle Park, NC. (Docket No. A–99–05, II– B.A. Egan, 1987. The Complex Terrain 39. Environmental Protection Agency,
A–8) Dispersion Model Terrain Preprocessor 1998. EPA Third-Generation Air Quality
14. Hanna, S., M. Garrison and B. Turner, System—User’s Guide and Program Modeling System. Models-3, Volume 9b:
1998. AERMOD Peer Review report. Prepared Description. Publication No. EPA–600/8–88– User Manual. Publication No. EPA–600/R–
by SAI, Inc. under EPA Contract No. 68–D6– 003. U.S. Environmental Protection Agency, 98/069(b). Office of Research and
0064/1–14 for Environmental Protection Research Triangle Park, NC. (NTIS No. PB Development, Washington, D.C.
Agency, Research Triangle Park, NC. 12pp. & 88–162094) 40. Gery, M.W. and R.R. Crouse, 1991.
appendices (Docket No. A–99–05, II–A–6) 28. Burns, D.J., S.G. Perry and A.J. User’s Guide for Executing OZIPR.
15. Environmental Protection Agency, Cimorelli, 1991. An Advanced Screening Publication No. EPA–600/8–90–069. Office of
1992. Protocol for Determining the Best Model for Complex Terrain Applications. Research & Development, Research Triangle
Performing Model. Publication No. EPA–454/ Paper presented at the 7th Joint Conference Park, NC. (NTIS No. PB 91–175877)
R–92–025. Office of Air Quality Planning & on Applications of Air Pollution Meteorology 41. Environmental Protection Agency,
Standards, Research Triangle Park, NC. (NTIS (cosponsored by the American 2002. User’s Guide to the Regulatory
No. PB 93–226082) Meteorological Society and the Air & Waste Modeling System for Aerosols and
16. ASTM D6589: Standard Guide for Management Association), January 13–18, Deposition (REMSAD) Version 7. Prepared
Statistical Evaluation of Atmospheric 1991, New Orleans, LA. for Environmental Protection Agency under
Dispersion Model Performance. (2000) 29. Environmental Research and Contract No. GS–10F–0124J by ICF
17. Environmental Protection Agency, Technology, 1987. User’s Guide to the Rough Consulting, July 2002. (Available at http://
1995. User’s Guide for the Industrial Source Terrain Diffusion Model (RTDM), Rev. 3.20. www.epa.gov/scram001/)
Complex (ISC3) Dispersion Models, Volumes ERT Document No. P–D535–585. 42. Environmental Protection Agency,
1 and 2. Publication Nos. EPA–454/B–95– Environmental Research and Technology, 2004. EPA–CMB8.2 Users Manual.
003a & b. U.S. Environmental Protection Inc., Concord, MA. (NTIS No. PB 88–171467) Publication No. EPA–452/R–04–011. Office
Agency, Research Triangle Park, NC. (NTIS 30. Meng, Z.D. Dabdub and J.H. Seinfeld, of Air Quality Planning & Standards,
Nos. PB 95–222741 and PB 95–222758, 1997. Chemical Coupling between Research Triangle Park, NC; December 2004.
respectively) Atmospheric Ozone and Particulate Matter. (Available at http://www.ega.gov/scram001/)
18. Hanna, S.R. and R.J. Paine, 1989. Science, 277: 116–119. 43. Environmental Protection Agency,
Hybrid Plume Dispersion Model (HPDM) 31. Hidy, G.M, P.M. Roth, J.M. Hales and 2004. Protocol for Applying and Validating
Development and Evaluation. J. Appl. R.D. Scheffe, 1998. Fine Particles and the CMB Model for PM2.5 and VOC.
Meteorol., 28: 206–224. Oxidant Pollution: Developing an Agenda for Publication No. EPA–451/R–04–001. Office
19. Hanna, S.R. and J.C. Chang, 1992. Cooperative Research. JAWMA, 50: 613–632. of Air Quality Planning & Standards,
Boundary layer parameterizations for applied 32. Environmental Protection Agency, Research Triangle Park, NC; December 2004.
dispersion modeling over urban areas. 2005. Guidance on the Use of Models and (Available at http://www.ega.gov/scram001/)
Bound. Lay. Meteorol., 58, 229–259. Other Analyses in Attainment 44. Environmental Protection Agency,
20. Hanna, S.R. and J.C. Chang, 1993. Demonstrations for the 8-hr Ozone NAAQS 1988. Chemical Mass Balance Model
Hybrid Plume Dispersion Model (HPDM) (Draft Final). Office of Air Quality Planning Diagnostics. Publication No. EPA–450/4–88–
Improvements and Testing at Three Field & Standards, Research Triangle Park, NC. 005. Office of Air Quality Planning &
Sites. Atmos. Environ., 27A: 1491–1508. (Latest version available on SCRAM Web site Standards, Research Triangle Park, NC. (NTIS
21. American Meteorological Society, 1984. as draft-final-O3.pdf; see subsection 2.3) No. PB 88–208319)
Workshop on Updating Applied Diffusion 33. Environmental Protection Agency, 45. Paatero, P. and U. Tapper, 1994.
Models. 24–27 January 1984. Clearwater, 2005. Guidance on the Use of Models and Positive Matrix Factorization: A Non-
negative Factor Model with Optimal Sources on Annual NO2 Concentrations. 3–85–022. Office of Research & Development,
Utilization of Error Estimates of Data Values. Proceedings, 84th Annual Meeting & Research Triangle Park, NC. (NTIS No. PB
Environmetrics, 5: 111–126. (Other Exhibition of the Air & Waste Management 85–203107)
documents related to PMF may be accessed Association, Vancouver, B.C.; 16–21 June 68. Bennett, M.J, M.E. Yansura, I.G.
via FTP at ftp://rock.helsinki.fi/pub/misc/ 1991. (16pp.) (Docket No. A–92–65, II–A–9) Hornyik, J.M. Nall, D.G. Caniparoli and C.G.
pmf.) 57. Cole, H.S. and J.E. Summerhays, 1979. Ashmore, 2002. Evaluation of the CALPUFF
46. Lewis, C.W., G.A. Norris, R.C. Henry A Review of Techniques Available for Long-range Transport Screening Technique
and T.L. Conner, 2003. Source Estimation of Short-Term NO2 by Comparison to Refined CALPUFF Results
Apportionment of Phoenix PM–2.5 Aerosol Concentrations. Journal of the Air Pollution for Several Power Plants in Both the Eastern
with the Unmix Receptor Model. Journal of Control Association, 29(8): 812–817. and Western United States. Proceedings of
the Air & Waste Management Association, 58. U.S. Department of Housing and Urban the Air & Waste Management Association’s
53(3): 325–338. Development, 1980. Air Quality 95th Annual Conference, June 23–27, 2002;
47. Environmental Protection Agency, Considerations in Residential Planning. U.S. Baltimore, MD. Paper #43454.
1994. Guidelines for PM10 Sampling and Superintendent of Documents, Washington, 69. Environmental Protection Agency,
Analysis Applicable to Receptor Modeling. DC. (GPO Order Nos. 023–000–00577–8, 1999. Guideline of Data Handling
Publication No. EPA–452/R–94–009. Office 023–000–00576–0, 023–000–00575–1) Conventions for the PM NAAQS. Publication
of Air Quality Planning & Standards, 59. Environmental Protection Agency, No. EPA–454/R–99–008. Office of Air
Research Triangle Park, NC. (NTIS No. PB 1986. Evaluation of Short-Term Long-Range Quality Planning & Standards, Research
94–177441) Transport Models, Volumes I and II. Triangle Park. (NTIS PB 99–149023)
48. Environmental Protection Agency, Publication Nos. EPA–450/4–86–016a and b. 70. Turner, D.B., 1969. Workbook of
1992. Guideline for Modeling Carbon Office of Air Quality Planning & Standards, Atmospheric Dispersion Estimates. PHS
Monoxide from Roadway Intersections. Research Triangle Park, NC. (NTIS Nos. PB Publication No. 999–AP–26. U.S. Department
Publication No. EPA–454/R–92–005. Office 87–142337 and PB 87–142345) of Health, Education and Welfare, Public
of Air Quality Planning & Standards, 60. Environmental Protection Agency, Health Service, Cincinnati, OH. (NTIS No.
Research Triangle Park, NC. (NTIS No. PB 1998. Interagency Workgroup on Air Quality PB–191482)
93–210391) Modeling (IWAQM) Phase 2 Summary Report 71. McElroy, J.L. and F. Pooler, Jr., 1968.
49. Environmental Protection Agency, and Recommendations for Modeling Long- St. Louis Dispersion Study, Volume II—
1992. User’s Guide for CAL3QHC Version 2: Range Transport Impacts. Publication No. Analysis. National Air Pollution Control
A Modeling Methodology for Predicting EPA–454/R–98–019. Office of Air Quality Administration Publication No. AP–53, U.S.
Pollutant Concentrations near Roadway Planning & Standards, Research Triangle Department of Health, Education and
Intersections. Publication No. EPA–454/R– Park, NC.(NTIS No. PB 99–121089) Welfare, Public Health Service, Arlington,
92–006. Office of Air Quality Planning & 61. National Acid Precipitation Assessment VA. (NTIS No. PB–190255)
Standards, Research Triangle Park, NC. (NTIS Program (NAPAP), 1991. Acid Deposition: 72. Irwin, J.S., 1978. Proposed Criteria for
No. PB 93–210250) State of Science and Technology. Volume III Selection of Urban Versus Rural Dispersion
50. Environmental Protection Agency, Terrestrial, Materials, Health and Visibility Coefficients. (Draft Staff Report). Meteorology
1992. Evaluation of CO Intersection Modeling Effects. Report 24, Visibility: Existing and and Assessment Division, U.S.
techniques Using a New York City Database. Historical Conditions—Causes and Effects Environmental Protection Agency, Research
Publication No. EPA–454/R–92–004. Office Edited by Patricia M. Irving. Washington, DC Triangle Park, NC. (Docket No. A–80–46, II-
of Air Quality Planning & Standards, RTP, 129pp. B–8)
NC 27711. (NTIS No. PB 93–105559) 62. National Research Council, 1993. 73. Auer, Jr., A.H., 1978. Correlation of
51. Environmental Protection Agency, Protecting Visibility in National Parks and Land Use and Cover with Meteorological
1995. Addendum to the User’s Guide to Wilderness Areas. National Academy Press, Anomalies. Journal of Applied Meteorology,
CAL3QHC Version 2.0. Staff Report. Office of Washington, DC 446pp. 17(5): 636–643.
Air Quality Planning & Standards, Research 63. Environmental Protection Agency, 74. Pasquill, F., 1976. Atmospheric
Triangle Park, NC. (Available at http:// 1992. Workbook for Plume Visual Impact Dispersion Parameters in Gaussian Plume
www.epa.gov/scram001/) Screening and Analysis (Revised). Modeling, Part II. Possible Requirements for
52. Shannon, J.D., 1987. Mobile Source Publication No. EPA–454/R–92–023. Office Change in the Turner Workbook Values.
Modeling Review. A report prepared under a of Air Quality Planning & Standards, Publication No. EPA–600/4–76–030b. Office
cooperative agreement with the Research Triangle Park, NC. (NTIS No. PB of Research & Development, Research
Environmental Protection Agency. 5pp. 93–223592) Triangle Park, NC. (NTIS No. PB–258036/
(Docket No. A–88–04, II–J–2) 64. Environmental Protection Agency, 3BA)
53. Environmental Protection Agency, 1981. Guideline for Use of Fluid Modeling to 75. Turner, D.B., 1964. A Diffusion Model
1991. Emission Inventory Requirements for Determine Good Engineering Practice (GEP) for an Urban Area. Journal of Applied
Carbon Monoxide State Implementation Stack Height. Publication No. EPA–450/4– Meteorology, 3(1): 83–91.
Plans. Publication No. EPA–450/4–91–011. 81–003. Office of Air Quality Planning & 76. Briggs, G.A., 1975. Plume Rise
Office of Air Quality Planning & Standards, Standards, Research Triangle Park, NC. (NTIS Predictions. Chapter 3 in Lectures on Air
Research Triangle Park, NC. (NTIS No. PB No. PB 82–145327) Pollution and Environmental Impact
92–112150) 65. Lawson, Jr., R.E. and W.H. Snyder, Analyses. American Meteorological Society,
54. Environmental Protection Agency, 1983. Determination of Good Engineering Boston, MA; pp. 59–111.
1992. Guideline for Regulatory Application Practice Stack Height: A Demonstration 77. Hanna, S.R., G.A. Briggs and R.P.
of the Urban Airshed Model for Areawide Study for a Power Plant. Publication No. Hosker, Jr., 1982. Plume Rise. Chapter 2 in
Carbon Monoxide. Publication No. EPA–450/ EPA–600/3–83–024. Office of Research & Handbook on Atmospheric Diffusion.
4–92–011a and b. Office of Air Quality Development, Research Triangle Park, NC. Technical Information Center, U.S.
Planning & Standards, Research Triangle (NTIS No. PB 83–207407) Department of Energy, Washington, DC; pp.
Park, NC. (NTIS Nos. PB 92–213222 and PB 66. Environmental Protection Agency, 11–24. DOE/TIC–11223 (DE 82002045)
92–213230) 1985. Guideline for Determination of Good 78. Weil, J.C., L.A. Corio and R.P. Brower,
55. Environmental Protection Agency, Engineering Practice Stack Height (Technical 1997. A PDF dispersion model for buoyant
1992. Technical Support Document to Aid Support Document for the Stack Height plumes in the convective boundary layer.
States with the Development of Carbon Regulations), Revised. Publication No. EPA– Journal of Applied Meteorology, 36: 982–
Monoxide State Implementation Plans. 450/4–80–023R. Office of Air Quality 1003.
Publication No. EPA–452/R–92–003. Office Planning & Standards, Research Triangle 79. Stull, R.B., 1988. An Introduction to
of Air Quality Planning & Standards, Park, NC. (NTIS No. PB 85–225241) Boundary Layer Meteorology. Kluwer
Research Triangle Park, NC. (NTIS No. PB 67. Snyder, W.H. and R.E. Lawson, Jr., Academic Publishers, Boston, MA. 666pp.
92–233055) 1985. Fluid Modeling Demonstration of Good 80. Environmental Protection Agency,
56. Chu, S.H. and E.L. Meyer, 1991. Use of Engineering-Practice Stack Height in 1988. User’s Guide to SDM—A Shoreline
Ambient Ratios to Estimate Impact of NOX Complex Terrain. Publication No. EPA–600/ Dispersion Model. Publication No. EPA–450/
4–88–017. Office of Air Quality Planning & Can be ordered from NOAA National Data 104. Environmental Protection Agency,
Standards, Research Triangle Park, NC. (NTIS Center’s Internet Web site at http:// 1996. Meteorological Processor for
No. PB 89–164305) lwf.ncdc.noaa.gov/oa/ncdc.html Regulatory Models (MPRM) User’s Guide.
81. Environmental Protection Agency, 92. Environmental Protection Agency, Publication No. EPA–454/B–96–002. Office
1987. Analysis and Evaluation of Statistical 2000. Meteorological Monitoring Guidance of Air Quality Planning & Standards,
Coastal Fumigation Models. Publication No. for Regulatory Modeling Applications. Research Triangle Park, NC. (NTIS No. PB
EPA–450/4–87–002. Office of Air Quality Publication No. EPA–454/R–99–005. Office 96–180518)
Planning & Standards, Research Triangle of Air Quality Planning & Standards, 105. Paine, R.J., 1987. User’s Guide to the
Park, NC. (NTIS No. PB 87–175519) Research Triangle Park, NC. (PB 2001– CTDM Meteorological Preprocessor Program.
82. Environmental Protection Agency, 103606) (Available at http://www.epa.gov/ Publication No. EPA–600/8–88–004. Office of
1995. Compilation of Air Pollutant Emission scram001/) Research & Development, Research Triangle
Factors, Volume I: Stationary Point and Area 93. ASTM D5527: Standard Practice for Park, NC. (NTIS No. PB 88–162102)
Sources (Fifth Edition, AP–42: GPO Stock Measuring Surface Winds and Temperature 106. Scire, J.S., F.R. Francoise, M.E. Fernau
No. 055–000–00500–1), and Supplements A– by Acoustic Means. (1994) and R.J. Yamartino, 1998. A User’s Guide for
D; Volume II: Mobile Sources (Fifth Edition). 94. ASTM D5741: Standard Practice for the CALMET Meteorological Model (Version
Office of Air Quality Planning & Standards, Characterizing Surface Wind Using Wind 5.0). Earth Tech, Inc., Concord, MA. (http://
Research Triangle Park, NC. Volume I can be Vane and Rotating Anemometer. (1996) www.src.com/calpuff/calpuff1.htm)
downloaded from EPA’s Internet Web site at 95. Environmental Protection Agency, 107. Environmental Protection Agency,
http://www.epa.gov/ttn/chief/ap42.html; 1995. Quality Assurance for Air Pollution 1984. Calms Processor (CALMPRO) User’s
Volume II can be downloaded from http:// Measurement Systems, Volume IV— Guide. Publication No. EPA–901/9–84–001.
www.epa.gov/omswww/ap42.htm Meteorological Measurements. Publication Office of Air Quality Planning & Standards,
83. Environmental Protection Agency, No. EPA600/R–94/038d. Office of Air Quality Region I, Boston, MA. (NTIS No. PB 84–
1987. Ambient Air Monitoring Guidelines for Planning & Standards, Research Triangle 229467)
Prevention of Significant Deterioration (PSD). Park, NC. Note: for copies of this handbook, 108. Fox, D.G., 1984. Uncertainty in air
Publication No. EPA–450/4–87–007. Office of you may make inquiry to ORD Publications, quality modeling. Bulletin of the American
Air Quality Planning & Standards, Research 26 West Martin Luther King Dr., Cincinatti, Meteorological Society, 65(1): 27–36.
Triangle Park, NC. (NTIS No. PB 90–168030) OH 45268. Phone (513) 569–7562 or (800) 109. Burton, C.S., 1981. The Role of
84. Stauffer, D.R. and Seaman, N.L., 1990. 490–9198 (automated request line) Atmospheric Models in Regulatory Decision-
Use of four-dimensional data assimilation in 96. Bowen, B.M., J.M. Dewart and A.I. Making: Summary Report. Systems
a limited-area mesoscale model. Part I: Chen, 1983. Stability Class Determination: A Applications, Inc., San Rafael, CA. Prepared
Experiments with synoptic-scale data. Comparison for One Site. Proceedings, Sixth under contract No. 68–01–5845 for U.S.
Environmental Protection Agency, Research
Monthly Weather Review, 118: 1250–1277. Symposium on Turbulence and Diffusion.
Triangle Park, NC. (Docket No. A–80–46,
85. Stauffer, D.R., N.L. Seaman and F.S. American Meteorological Society, Boston,
II–M–6)
Binkowski, 1991. Use of four-dimensional MA; pp. 211–214. (Docket No. A–92–65,
110. Environmental Protection Agency,
data assimilation in a limited-area mesoscale II–A–7)
1981. Proceedings of the Second Conference
model. Part II: Effect of data assimilation 97. Environmental Protection Agency,
on Air Quality Modeling, Washington, DC.
within the planetary boundary layer. Monthly 1993. An Evaluation of a Solar Radiation/ Office of Air Quality Planning & Standards,
Weather Review, 119: 734–754. Delta-T (SRDT) Method for Estimating Research Triangle Park, NC. (Docket No. A–
86. Grell, G.A., J. Dudhia, and D.R. Pasquill-Gifford (P–G) Stability Categories. 80–46,
Stauffer, 1994. A Description of the Fifth- Publication No. EPA–454/R–93–055. Office II–M–16)
Generation Penn State/NCAR Mesoscale of Air Quality Planning & Standards, 111. Hanna, S.R., 1989. Confidence limits
Model (MM5). NCAR Technical Note, NCAR/ Research Triangle Park, NC. (NTIS No. PB for air quality model evaluations, as
TN–398+STR, National Center for 94–113958) estimated by bootstrap and jackknife
Atmospheric Research, Boulder, CO; 138pp. 98. Irwin, J.S., 1980. Dispersion Estimate resampling methods. Atmospheric
http://www.mmm.ucar.edu/mm5/mm5- Suggestion #8: Estimation of Pasquill Environment, 23(6): 1385–1398.
home.html Stability Categories. Office of Air Quality 112. Cox, W.M. and J.A. Tikvart, 1990. A
87. Landsberg, H.E. and W.C. Jacobs, 1951. Planning & Standards, Research Triangle statistical procedure for determining the best
Compendium of Meteorology. American Park, NC (Docket No. A–80–46, II–B–10) performing air quality simulation model.
Meteorological Society, Boston, MA; pp. 99. Mitchell, Jr., A.E. and K.O. Timbre, Atmos. Environ., 24A(9): 2387–2395.
976–992. 1979. Atmospheric Stability Class from 113. Oreskes, N.K., K. Shrader-Frechette
88. Burton, C.S., T.E. Stoeckenius and J.P. Horizontal Wind Fluctuation. Presented at and K. Beliz, 1994. Verification, validation
Nordin, 1983. The Temporal 72nd Annual Meeting of Air Pollution and confirmation of numerical models in the
Representativeness of Short-Term Control Association, Cincinnati, OH; June earth sciences. Science, 263: 641–646.
Meteorological Data Sets: Implications for Air 24–29, 1979. (Docket No. A–80–46, II–P–9) 114. Dekker, C.M., A. Groenendijk, C.J.
Quality Impact Assessments. Systems 100. Smedman—Hogstrom, A. and V. Sliggers and G.K. Verboom, 1990. Quality
Applications, Inc., San Rafael, CA. (Docket Hogstrom, 1978. A Practical Method for Criteria for Models to Calculate Air Pollution.
No. A–80–46, II-G–11) Determining Wind Frequency Distributions Lucht (Air) 90, Ministry of Housing, Physical
89. Solar and Meteorological Surface for the Lowest 200m from Routine Planning and Environment, Postbus 450,
Observation Network, 1961–1990; 3-volume Meteorological Data. J. of Applied 2260 MB Leidschendam, The Netherlands;
CD–ROM. Version 1.0, September 1993. Meteorology, 17(7): 942–954. 52pp.
Produced jointly by National Climatic Data 101. Smith, T.B. and S.M. Howard, 1972. 115. Weil, J.C., R.I. Sykes and A.
Center and National Renewable Energy Methodology for Treating Diffusivity. MRI 72 Venkatram, 1992. Evaluating air-quality
Laboratory. Can be ordered from NOAA FR–1030. Meteorology Research, Inc., models: review and outlook. Journal of
National Data Center’s Internet Web site at Altadena, CA. (Docket No. A–80–46, II–P–8) Applied Meteorology, 31: 1121–1145.
http://www.NNDC.NOAA.GOV/. 102. Environmental Protection Agency, 116. Cole, S.T. and P.J. Wicks, Editors
90. Hourly United States Weather 2004. User’s Guide for the AERMOD (1995): Model Evaluation Group: Report of
Observations, 1990–1995 (CD–ROM). October Meteorological Preprocessor (AERMET). the Second Open Meeting. EUR 15990 EN,
1997. Produced jointly by National Climatic Publication No. EPA–454/B–03–002. U.S. European Commission, Directorate-General
Data Center and Environmental Protection Environmental Protection Agency, Research XII, Environmental Research Programme, L–
Agency. Can be ordered from NOAA National Triangle Park, NC. (Available at http:// 2920 Luxembourg; 77pp.
Data Center’s Internet Web site at http:// www.epa.gov/scram001/) 117. Hanna, S.R., 1982. Natural Variability
lwf.ncdc.noaa.gov/oa/ncdc.html 103. Environmental Protection Agency, of Observed Hourly SO2 and CO
91. Radiosonde Data of North America, 1993. PCRAMMET User’s Guide. Publication Concentrations in St. Louis. Atmospheric
1946–1996; 4-volume CD–ROM. August No. EPA–454/R–96–001. Office of Air Environment, 16(6): 1435–1440.
1996. Produced jointly by Forecast Systems Quality Planning & Standards, Research 118. Bowne, N.E., 1981. Validation and
laboratory and National Climatic Data Center. Triangle Park, NC. (NTIS No. PB 97–147912) Performance Criteria for Air Quality Models.
Appendix F in Air Quality Modeling and the A.2 Buoyant Line and Point Source 454/B–03–001. U.S. Environmental
Clean Air Act: Recommendations to EPA on Dispersion Model (BLP) Protection Agency, Research Triangle Park,
Dispersion Modeling for Regulatory A.3 CALINE3 NC 27711; September 2004. (Available at
Applications. American Meteorological A.4 CALPUFF http://www.epa.gov/scram001/)
Society, Boston, MA; pp. 159–171. (Docket A.5 Complex Terrain Dispersion Model Environmental Protection Agency, 2004.
No. A–80–46, II–A–106) Plus Algorithms for Unstable Situations User’s Guide for the AERMOD
119. Bowne, N.E. and R.J. Londergan, 1983. (CTDMPLUS) Meteorological Preprocessor (AERMET).
Overview, Results, and Conclusions for the A.6 Offshore and Coastal Dispersion Model Publication No. EPA–454/B–03–002. U.S.
EPRI Plume Model Validation and (OCD) Environmental Protection Agency, Research
Development Project: Plains Site. EPRI EA– A.REF References Triangle Park, NC 27711; November 2004.
3074. Electric Power Research Institute, Palo (Available at http://www.epa.gov/scram001/)
A.0 Introduction and Availability Environmental Protection Agency, 2004.
Alto, CA.
120. Moore, G.E., T.E. Stoeckenius and (1) This appendix summarizes key features User’s Guide for the AERMOD Terrain
D.A. Stewart, 1982. A Survey of Statistical of refined air quality models preferred for Preprocessor (AERMAP). Publication No.
Measures of Model Performance and specific regulatory applications. For each EPA–454/B–03–003. U.S. Environmental
Accuracy for Several Air Quality Models. model, information is provided on Protection Agency, Research Triangle Park,
Publication No. EPA–450/4–83–001. Office of availability, approximate cost (where NC 27711; October 2004. (Available at http://
Air Quality Planning & Standards, Research applicable), regulatory use, data input, www.epa.gov/scram001/)
Triangle Park, NC. (NTIS No. PB 83–260810) output format and options, simulation of Schulman, L.L., D.G. Strimaitis and J.S.
121. Rhoads, R.G., 1981. Accuracy of Air atmospheric physics, and accuracy. These Scire, 2000. Development and evaluation of
Quality Models. Staff Report. Office of Air models may be used without a formal the PRIME plume rise and building
Quality Planning & Standards, Research demonstration of applicability provided they downwash model. Journal of the Air and
Triangle Park, NC. (Docket No. A–80–46, satisfy the recommendations for regulatory Waste Management Association, 50: 378–
II–G–6) use; not all options in the models are 390.
122. Hanna, S.R., 1993. Uncertainties in air necessarily recommended for regulatory use.
(2) Many of these models have been Availability
quality model predictions. Boundary-Layer
Meteorology, 62: 3–20. subjected to a performance evaluation using The model codes and associated
123. Pasquill, F., 1974. Atmospheric comparisons with observed air quality data. documentation are available on EPA’s
Diffusion, 2nd Edition. John Wiley and Sons, Where possible, several of the models Internet SCRAM Web site (Section A.0).
New York, NY; 479pp. contained herein have been subjected to
evaluation exercises, including (1) statistical Abstract
124. Morgan, M.G. and M. Henrion, 1990.
performance tests recommended by the AERMOD is a steady-state plume
Uncertainty, A Guide to Dealing With
American Meteorological Society and (2) dispersion model for assessment of pollutant
Uncertainty in Quantitative Risk and Policy concentrations from a variety of sources.
peer scientific reviews. The models in this
Analysis. Cambridge University Press. New AERMOD simulates transport and dispersion
appendix have been selected on the basis of
York, NY; 332pp. from multiple point, area, or volume sources
the results of the model evaluations,
125. Irwin, J.S., K. Steinberg, C. based on an up-to-date characterization of the
experience with previous use, familiarity of
Hakkarinen and H. Feldman, 2001. atmospheric boundary layer. Sources may be
the model to various air quality programs,
Uncertainty in Air Quality Modeling for Risk located in rural or urban areas, and receptors
and the costs and resource requirements for
Calculations. (CD–ROM) Proceedings of may be located in simple or complex terrain.
use.
Guideline on Air Quality Models: A New AERMOD accounts for building wake effects
(3) Codes and documentation for all
Beginning. April 4–6, 2001, Newport, RI, Air models listed in this appendix are available (i.e., plume downwash) based on the PRIME
& Waste Management Association. from EPA’s Support Center for Regulatory Air building downwash algorithms. The model
Pittsburgh, PA; 17pp. Models (SCRAM) Web site at http:// employs hourly sequential preprocessed
126. Austin, B.S., T.E. Stoeckenius, M.C. www.epa.gov/scram001. Documentation is meteorological data to estimate
Dudik and T.S. Stocking, 1988. User’s Guide also available from the National Technical concentrations for averaging times from one
to the Expected Exceedances System. Information Service (NTIS), http:// hour to one year (also multiple years).
Systems Applications, Inc., San Rafael, CA. www.ntis.gov or U.S. Department of AERMOD is designed to operate in concert
Prepared under Contract No. 68–02–4352 Commerce, Springfield, VA 22161; phone: with two pre-processor codes: AERMET
Option I for the U.S. Environmental (800) 553–6847. Where possible, accession processes meteorological data for input to
Protection Agency, Research Triangle Park, numbers are provided. AERMOD, and AERMAP processes terrain
NC. (Docket No. A–88–04, II–I–3) elevation data and generates receptor
127. Thrall, A.D., T.E. Stoeckenius and C.S. A.1 AMS/EPA Regulatory Model— information for input to AERMOD.
Burton, 1985. A Method for Calculating AERMOD
a. Recommendations for Regulatory Use
Dispersion Modeling Uncertainty Applied to References
the Regulation of an Emission Source. (1) AERMOD is appropriate for the
Systems Applications, Inc., San Rafael, CA. Environmental Protection Agency, 2004. following applications:
Prepared for the U.S. Environmental AERMOD: Description of Model • Point, volume, and area sources;
Protection Agency, Research Triangle Park, Formulation. Publication No. EPA–454/R– • Surface, near-surface, and elevated
NC. (Docket No. A–80–46, IV–G–1) 03–004. U.S. Environmental Protection releases;
128. ‘‘Ten years of Harmonisation Agency, Research Triangle Park, NC 27711; • Rural or urban areas;
activities: Past, present and future’’ at http:// September 2004. (Available at http:// • Simple and complex terrain;
www.dmu.dk/AtmosphericEnvironment/ www.epa.gov/scram001/) • Transport distances over which steady-
Harmoni/Conferences/Belgirate/ Cimorelli, A. et al., 2005. AERMOD: A state assumptions are appropriate, up to
BelgiratePapers.asp. Dispersion Model for Industrial Source 50km;
129. ‘‘A platform for model evaluation’’ at Applications. Part I: General Model • 1-hour to annual averaging times; and
http://www.dmu.dk/ Formulation and Boundary Layer • Continuous toxic air emissions.
AtmosphericEnvironment/Harmoni/ Characterization. Journal of Applied (2) For regulatory applications of
Conferences/Belgirate/BelgiratePapers.asp. Meteorology, 44(5): 682–693. AERMOD, the regulatory default option
Perry, S. et al., 2005. AERMOD: A should be set, i.e., the parameter DFAULT
APPENDIX A TO APPENDIX W OF Dispersion Model for Industrial Source should be employed in the MODELOPT
PART 51—SUMMARIES OF Applications. Part II: Model Performance record in the COntrol Pathway. The DFAULT
PREFERRED AIR QUALITY MODELS against 17 Field Study Databases. Journal of option requires the use of terrain elevation
Applied Meteorology, 44(5): 694–708. data, stack-tip downwash, sequential date
Table of Contents Environmental Protection Agency, 2004. checking, and does not permit the use of the
A.0 Introduction and Availability User’s Guide for the AMS/EPA Regulatory model in the SCREEN mode. In the
A.1 Aermod Model—AERMOD. Publication No. EPA– regulatory default mode, pollutant half life or
decay options are not employed, except in (ii) For recommendations regarding the the mixed layer, but can disperse downward
the case of an urban source of sulfur dioxide length of meteorological record needed to and re-enter the mixed layer. In the CBL,
where a four-hour half life is applied. Terrain perform a regulatory analysis with AERMOD, plume rise is superposed on the
elevation data from the U.S. Geological see Section 8.3.1. displacements by random convective
Survey 7.5-Minute Digital Elevation Model (3) Receptor data: Receptor coordinates, velocities (Weil et al., 1997).
(edcwww.cr.usgs.gov/doc/edchome/ndcdb/ elevations, height above ground, and hill (2) In the stable boundary layer, plume rise
ndcdb.html) or equivalent (approx. 30-meter height scales are produced by the AERMAP is estimated using an iterative approach,
resolution) should be used in all terrain preprocessor for input to AERMOD. similar to that in the CTDMPLUS model (see
applications. In some cases, exceptions of the Discrete receptors and/or multiple receptor A.5 in this appendix).
terrain data requirement may be made in grids, Cartesian and/or polar, may be (3) Stack-tip downwash and buoyancy
consultation with the permit/SIP reviewing employed in AERMOD. AERMAP requires induced dispersion effects are modeled.
authority. input of Digital Elevation Model (DEM) Building wake effects are simulated for stacks
b. Input Requirements terrain data produced by the U.S. Geological less than good engineering practice height
Survey (USGS), or other equivalent data. using the methods contained in the PRIME
(1) Source data: Required input includes AERMAP can be used optionally to estimate downwash algorithms (Schulman, et al.,
source type, location, emission rate, stack source elevations. 2000). For plume rise affected by the
height, stack inside diameter, stack gas exit presence of a building, the PRIME downwash
velocity, stack gas temperature, area and c. Output
algorithm uses a numerical solution of the
volume source dimensions, and source Printed output options include input mass, energy and momentum conservation
elevation. Building dimensions and variable information, high concentration summary laws (Zhang and Ghoniem, 1993). Streamline
emission rates are optional. tables by receptor for user-specified deflection and the position of the stack
(2) Meteorological data: The AERMET averaging periods, maximum concentration relative to the building affect plume
meteorological preprocessor requires input of summary tables, and concurrent values trajectory and dispersion. Enhanced
surface characteristics, including surface summarized by receptor for each day dispersion is based on the approach of Weil
roughness (zo), Bowen ratio, and albedo, as processed. Optional output files can be (1996). Plume mass captured by the cavity is
well as, hourly observations of wind speed generated for: a listing of occurrences of well-mixed within the cavity. The captured
between 7zo and 100m (reference wind speed exceedances of user-specified threshold plume mass is re-emitted to the far wake as
measurement from which a vertical profile value; a listing of concurrent (raw) results at a volume source.
can be developed), wind direction, cloud each receptor for each hour modeled, suitable (4) For elevated terrain, AERMOD
cover, and temperature between zo and 100m for post-processing; a listing of design values incorporates the concept of the critical
(reference temperature measurement from that can be imported into graphics software dividing streamline height, in which flow
which a vertical profile can be developed). for plotting contours; an unformatted listing below this height remains horizontal, and
Surface characteristics may be varied by of raw results above a threshold value with flow above this height tends to rise up and
wind sector and by season or month. A a special structure for use with the TOXX over terrain (Snyder et al., 1985). Plume
morning sounding (in National Weather model component of TOXST; a listing of concentration estimates are the weighted sum
Service format) from a representative upper concentrations by rank (e.g., for use in of these two limiting plume states. However,
air station, latitude, longitude, time zone, and quantile-quantile plots); and, a listing of consistent with the steady-state assumption
wind speed threshold are also required in concentrations, including arc-maximum of uniform horizontal wind direction over the
AERMET (instrument threshold is only normalized concentrations, suitable for modeling domain, straight-line plume
required for site specific data). Additionally, model evaluation studies. trajectories are assumed, with adjustment in
measured profiles of wind, temperature, d. Type of Model the plume/receptor geometry used to account
vertical and lateral turbulence may be for the terrain effects.
required in certain applications (e.g., in AERMOD is a steady-state plume model,
using Gaussian distributions in the vertical h. Horizontal Winds
complex terrain) to adequately represent the
meteorology affecting plume transport and and horizontal for stable conditions, and in Vertical profiles of wind are calculated for
dispersion. Optionally, measurements of the horizontal for convective conditions. The each hour based on measurements and
solar, or net radiation may be input to vertical concentration distribution for surface-layer similarity (scaling)
AERMET. Two files are produced by the convective conditions results from an relationships. At a given height above
AERMET meteorological preprocessor for assumed bi-Gaussian probability density ground, for a given hour, winds are assumed
input to the AERMOD dispersion model. The function of the vertical velocity. constant over the modeling domain. The
surface file contains observed and calculated e. Pollutant Types effect of the vertical variation in horizontal
surface variables, one record per hour. The wind speed on dispersion is accounted for
AERMOD is applicable to primary through simple averaging over the plume
profile file contains the observations made at pollutants and continuous releases of toxic
each level of a meteorological tower (or depth.
and hazardous waste pollutants. Chemical
remote sensor), or the one-level observations transformation is treated by simple i. Vertical Wind Speed
taken from other representative data (e.g., exponential decay. In convective conditions, the effects of
National Weather Service surface random vertical updraft and downdraft
observations), one record per level per hour. f. Source-Receptor Relationships
velocities are simulated with a bi-Gaussian
(i) Data used as input to AERMET should AERMOD applies user-specified locations probability density function. In both
possess an adequate degree of for sources and receptors. Actual separation convective and stable conditions, the mean
representativeness to insure that the wind, between each source-receptor pair is used. vertical wind speed is assumed equal to zero.
temperature and turbulence profiles derived Source and receptor elevations are user input
by AERMOD are both laterally and vertically j. Horizontal Dispersion
or are determined by AERMAP using USGS
representative of the source area. The DEM terrain data. Receptors may be located Gaussian horizontal dispersion coefficients
adequacy of input data should be judged at user-specified heights above ground level. are estimated as continuous functions of the
independently for each variable. The values parameterized (or measured) ambient lateral
for surface roughness, Bowen ratio, and g. Plume Behavior
turbulence and also account for buoyancy-
albedo should reflect the surface (1) In the convective boundary layer (CBL), induced and building wake-induced
characteristics in the vicinity of the the transport and dispersion of a plume is turbulence. Vertical profiles of lateral
meteorological tower, and should be characterized as the superposition of three turbulence are developed from measurements
adequately representative of the modeling modeled plumes: The direct plume (from the and similarity (scaling) relationships.
domain. Finally, the primary atmospheric stack), the indirect plume, and the penetrated Effective turbulence values are determined
input variables including wind speed and plume, where the indirect plume accounts from the portion of the vertical profile of
direction, ambient temperature, cloud cover, for the lofting of a buoyant plume near the lateral turbulence between the plume height
and a morning upper air sounding should top of the boundary layer, and the penetrated and the receptor height. The effective lateral
also be adequately representative of the plume accounts for the portion of a plume turbulence is then used to estimate
source area. that, due to its buoyancy, penetrates above horizontal dispersion.
k. Vertical Dispersion Availability c. Output

In the stable boundary layer, Gaussian The computer code is available on EPA’s (1) Printed output (from a separate post-
vertical dispersion coefficients are estimated Internet SCRAM Web site and also on processor program) includes:
as continuous functions of parameterized diskette (as PB 2002–500051) from the (2) Total concentration or, optionally,
vertical turbulence. In the convective National Technical Information Service (see source contribution analysis; monthly and
boundary layer, vertical dispersion is Section A.0). annual frequency distributions for 1-, 3-, and
characterized by a bi-Gaussian probability 24-hour average concentrations; tables of
density function, and is also estimated as a Abstract
1-, 3-, and 24-hour average concentrations at
continuous function of parameterized BLP is a Gaussian plume dispersion model each receptor; table of the annual (or length
vertical turbulence. Vertical turbulence designed to handle unique modeling of run) average concentrations at each
profiles are developed from measurements problems associated with aluminum receptor;
and similarity (scaling) relationships. These reduction plants, and other industrial sources (3) Five highest 1-, 3-, and 24-hour average
turbulence profiles account for both where plume rise and downwash effects from concentrations at each receptor; and
convective and mechanical turbulence. stationary line sources are important. (4) Fifty highest 1-, 3-, and 24-hour
Effective turbulence values are determined concentrations over the receptor field.
from the portion of the vertical profile of a. Recommendations for Regulatory Use
vertical turbulence between the plume height (1) The BLP model is appropriate for the d. Type of Model
and the receptor height. The effective vertical following applications: BLP is a gaussian plume model.
turbulence is then used to estimate vertical • Aluminum reduction plants which e. Pollutant Types
dispersion. contain buoyant, elevated line sources;
• Rural areas; BLP may be used to model primary
l. Chemical Transformation
• Transport distances less than 50 pollutants. This model does not treat settling
Chemical transformations are generally not and deposition.
treated by AERMOD. However, AERMOD kilometers;
does contain an option to treat chemical • Simple terrain; and f. Source-Receptor Relationship
transformation using simple exponential • One hour to one year averaging times. (1) BLP treats up to 50 point sources, 10
decay, although this option is typically not (2) The following options should be parallel line sources, and 100 receptors
used in regulatory applications, except for selected for regulatory applications: arbitrarily located.
sources of sulfur dioxide in urban areas. (i) Rural (IRU=1) mixing height option; (2) User-input topographic elevation is
Either a decay coefficient or a half life is (ii) Default (no selection) for plume rise applied for each stack and each receptor.
input by the user. Note also that the Plume wind shear (LSHEAR), transitional point
source plume rise (LTRANS), vertical g. Plume Behavior
Volume Molar Ratio Method (subsection 5.1)
and the Ozone Limiting Method (subsection potential temperature gradient (DTHTA), (1) BLP uses plume rise formulas of
5.2.4) and for point-source NO2 analyses are vertical wind speed power law profile Schulman and Scire (1980).
available as non-regulatory options. exponents (PEXP), maximum variation in (2) Vertical potential temperature gradients
number of stability classes per hour (IDELS), of 0.02 Kelvin per meter for E stability and
m. Physical Removal pollutant decay (DECFAC), the constant in 0.035 Kelvin per meter are used for stable
AERMOD can be used to treat dry and wet Briggs’ stable plume rise equation (CONST2), plume rise calculations. An option for user
deposition for both gases and particles. constant in Briggs’ neutral plume rise input values is included.
n. Evaluation Studies equation (CONST3), convergence criterion (3) Transitional rise is used for line
for the line source calculations (CRIT), and sources.
American Petroleum Institute, 1998. maximum iterations allowed for line source (4) Option to suppress the use of
Evaluation of State of the Science of Air calculations (MAXIT); and transitional plume rise for point sources is
Quality Dispersion Model, Scientific (iii) Terrain option (TERAN) set equal to included.
Evaluation, prepared by Woodward-Clyde
0.0, 0.0, 0.0, 0.0, 0.0, 0.0 (5) The building downwash algorithm of
Consultants, Lexington, Massachusetts, for
(3) For other applications, BLP can be used Schulman and Scire (1980) is used.
American Petroleum Institute, Washington,
if it can be demonstrated to give the same h. Horizontal Winds
D.C., 20005–4070.
estimates as a recommended model for the
Brode, R.W., 2002. Implementation and (1) Constant, uniform (steady-state) wind is
same application, and will subsequently be
Evaluation of PRIME in AERMOD. Preprints assumed for an hour.
executed in that mode.
of the 12th Joint Conference on Applications Straight line plume transport is assumed to
(4) BLP can be used on a case-by-case basis
of Air Pollution Meteorology, May 20–24, all downwind distances.
with specific options not available in a
2002; American Meteorological Society, (2) Wind speeds profile exponents of 0.10,
recommended model if it can be
Boston, MA. 0.15, 0.20, 0.25, 0.30, and 0.30 are used for
Brode, R.W., 2004. Implementation and demonstrated, using the criteria in Section
3.2, that the model is more appropriate for a stability classes A through F, respectively.
Evaluation of Bulk Richardson Number An option for user-defined values and an
Scheme in AERMOD. 13th Joint Conference specific application.
option to suppress the use of the wind speed
on Applications of Air Pollution b. Input Requirements profile feature are included.
Meteorology, August 23–26, 2004; American (1) Source data: point sources require stack
Meteorological Society, Boston, MA. i. Vertical Wind Speed
location, elevation of stack base, physical
Environmental Protection Agency, 2003. stack height, stack inside diameter, stack gas Vertical wind speed is assumed equal to
AERMOD: Latest Features and Evaluation exit velocity, stack gas exit temperature, and zero.
Results. Publication No. EPA–454/R–03–003. pollutant emission rate. Line sources require j. Horizontal Dispersion
U.S. Environmental Protection Agency, coordinates of the end points of the line,
Research Triangle Park, NC. Available at (1) Rural dispersion coefficients are from
release height, emission rate, average line Turner (1969), with no adjustment made for
http://www.epa.gov/scram001/. source width, average building width, variations in surface roughness or averaging
A.2 Buoyant Line and Point Source average spacing between buildings, and time.
Dispersion Model (BLP) average line source buoyancy parameter. (2) Six stability classes are used.
(2) Meteorological data: surface weather
Reference data from a preprocessor such as k. Vertical Dispersion
Schulman, Lloyd L., and Joseph S. Scire, PCRAMMET which provides hourly stability (1) Rural dispersion coefficients are from
1980. Buoyant Line and Point Source (BLP) class, wind direction, wind speed, Turner (1969), with no adjustment made for
Dispersion Model User’s Guide. Document temperature, and mixing height. variations in surface roughness.
P–7304B. Environmental Research and (3) Receptor data: locations and elevations (2) Six stability classes are used.
Technology, Inc., Concord, MA. (NTIS No. of receptors, or location and size of receptor (3) Mixing height is accounted for with
PB 81–164642; also available at http:// grid or request automatically generated multiple reflections until the vertical plume
www.epa.gov/scram001/) receptor grid. standard deviation equals 1.6 times the
mixing height; uniform mixing is assumed (2) Meteorological data: wind speed, wind EPA–450/4–86–002. Office of Air Quality
beyond that point. angle (measured in degrees clockwise from Planning & Standards, Research Triangle
(4) Perfect reflection at the ground is the Y axis), stability class, mixing height, Park, NC. (NTIS No. PB 86–167293)
assumed. ambient (background to the highway)
concentration of pollutant. A.4 CALPUFF
(3) Receptor data: coordinates and height References
Chemical transformations are treated using above ground for each receptor.
linear decay. Decay rate is input by the user. Scire, J.S., D.G. Strimaitis and R.J.
c. Output Yamartino, 2000. A User’s Guide for the
m. Physical Removal
Printed output includes concentration at CALPUFF Dispersion Model (Version 5.0).
Physical removal is not explicitly treated. each receptor for the specified meteorological Earth Tech, Inc., Concord, MA.
n. Evaluation Studies condition. Scire J.S., F.R. Robe, M.E. Fernau and R.J.
d. Type of Model Yamartino, 2000. A User’s Guide for the
Schulman, L.L. and J.S. Scire, 1980.
CALMET Meteorological Model (Version
Buoyant Line and Point Source (BLP) CALINE–3 is a Gaussian plume model.
5.0). Earth Tech, Inc., Concord, MA.
Dispersion Model User’s Guide, P–7304B. e. Pollutant Types
Environmental Research and Technology, Availability
Inc., Concord, MA. CALINE–3 may be used to model primary
pollutants. The model code and its documentation are
Scire, J.S. and L.L. Schulman, 1981. available at no cost for download from the
Evaluation of the BLP and ISC Models with f. Source-Receptor Relationship
model developers’ Internet Web site: http://
SF6 Tracer Data and SO2 Measurements at (1) Up to 20 highway links are treated. www.src.com/calpuff/calpuff1.htm. You may
Aluminum Reduction Plants. APCA (2) CALINE–3 applies user input location also contact Joseph Scire, Earth Tech, Inc.,
Specialty Conference on Dispersion and emission rate for each link. User-input 196 Baker Avenue, Concord, MA 01742;
Modeling for Complex Sources, St. Louis, receptor locations are applied. Telephone: (978) 371–4270; Fax: (978) 371–
MO.
g. Plume Behavior 2468; e-mail: JScire@alum.mit.edu.
A.3 CALINE3 Plume rise is not treated. Abstract
Reference h. Horizontal Winds CALPUFF is a multi-layer, multi-species
Benson, Paul E., 1979. CALINE3—A (1) User-input hourly wind speed and non-steady-state puff dispersion modeling
Versatile Dispersion Model for Predicting Air direction are applied. system that simulates the effects of time- and
Pollutant Levels Near Highways and Arterial (2) Constant, uniform (steady-state) wind is space-varying meteorological conditions on
Streets. Interim Report, Report Number assumed for an hour. pollutant transport, transformation, and
FHWA/CA/TL–79/23. Federal Highway removal. CALPUFF is intended for use on
i. Vertical Wind Speed
Administration, Washington, DC (NTIS No. scales from tens of meters from a source to
PB 80–220841). Vertical wind speed is assumed equal to hundreds of kilometers. It includes
zero. algorithms for near-field effects such as stack
Availability
j. Horizontal Dispersion tip downwash, building downwash,
The CALINE3 model is available on transitional buoyant and momentum plume
diskette (as PB 95–502712) from NTIS. The (1) Six stability classes are used.
(2) Rural dispersion coefficients from rise, rain cap effects, partial plume
source code and user’s guide are also penetration, subgrid scale terrain and coastal
available on EPA’s Internet SCRAM Web site Turner (1969) are used, with adjustment for
roughness length and averaging time. interactions effects, and terrain impingement
( Section A.0). as well as longer range effects such as
(3) Initial traffic-induced dispersion is
Abstract handled implicitly by plume size parameters. pollutant removal due to wet scavenging and
dry deposition, chemical transformation,
CALINE3 can be used to estimate the k. Vertical Dispersion vertical wind shear effects, overwater
concentrations of nonreactive pollutants from (1) Six stability classes are used. transport, plume fumigation, and visibility
highway traffic. This steady-state Gaussian (2) Empirical dispersion coefficients from effects of particulate matter concentrations.
model can be applied to determine air Benson (1979) are used including an
pollution concentrations at receptor locations a. Recommendations for Regulatory Use
adjustment for roughness length.
downwind of ‘‘at-grade,’’ ‘‘fill,’’ ‘‘bridge,’’ (3) Initial traffic-induced dispersion is (1) CALPUFF is appropriate for long range
and ‘‘cut section’’ highways located in handled implicitly by plume size parameters. transport (source-receptor distances of 50 to
relatively uncomplicated terrain. The model (4) Adjustment for averaging time is several hundred kilometers) of emissions
is applicable for any wind direction, highway included. from point, volume, area, and line sources.
orientation, and receptor location. The model The meteorological input data should be
has adjustments for averaging time and l. Chemical Transformation fully characterized with time-and-space-
surface roughness, and can handle up to 20 Not treated. varying three dimensional wind and
links and 20 receptors. It also contains an m. Physical Removal meteorological conditions using CALMET, as
algorithm for deposition and settling velocity discussed in paragraphs 8.3(d) and 8.3.1.2(d)
so that particulate concentrations can be Optional deposition calculations are
of Appendix W.
predicted. included.
(2) CALPUFF may also be used on a case-
a. Recommendations for Regulatory Use n. Evaluation Studies by-case basis if it can be demonstrated using
Bemis, G.R. et al., 1977. Air Pollution and the criteria in Section 3.2 that the model is
CALINE–3 is appropriate for the following
Roadway Location, Design, and Operation— more appropriate for the specific application.
applications:
Project Overview. FHWA–CA–TL–7080–77– The purpose of choosing a modeling system
• Highway (line) sources;
25, Federal Highway Administration, like CALPUFF is to fully treat stagnation,
• Urban or rural areas;
• Simple terrain; Washington, DC. wind reversals, and time and space variations
• Transport distances less than 50 Cadle, S.H. et al., 1976. Results of the of meteorological conditions on transport and
kilometers; and General Motors Sulfate Dispersion dispersion, as discussed in paragraph
• One-hour to 24-hour averaging times. Experiment, GMR–2107. General Motors 7.2.8(a).
Research Laboratories, Warren, MI. (3) For regulatory applications of CALMET
b. Input Requirements Dabberdt, W.F., 1975. Studies of Air and CALPUFF, the regulatory default option
(1) Source data: up to 20 highway links Quality on and Near Highways, Project 2761. should be used. Inevitably, some of the
classed as ‘‘at-grade,’’ ‘‘fill,’’ ‘‘bridge,’’ or Stanford Research Institute, Menlo Park, CA. model control options will have to be set
‘‘depressed’’; coordinates of link end points; Environmental Protection Agency, 1986. specific for the application using expert
traffic volume; emission factor; source height; Evaluation of Mobile Source Air Quality judgment and in consultation with the
and mixing zone width. Simulation Models. EPA Publication No. appropriate reviewing authorities.
b. Input Requirements MM5, RUC, Eta and RAMS) can be used by and POSTUTIL allows the re-partitioning of
Source Data: CALMET as well (paragraph 8.3.1.2(d)). nitric acid and nitrate to account for the
1. Point sources: Source location, stack CALMET contains an option to be run in effects of ammonia limitation (Scire et al.,
height, diameter, exit velocity, exit ‘‘No-observations’’ mode (Robe et al., 2002), 2001; Escoffier-Czaja and Scire, 2002).
temperature, base elevation, wind direction which allows the 3–D CALMET CALPUFF contains an options to output
specific building dimensions (for building meteorological fields to be based on liquid water concentrations for use in
downwash calculations), and emission rates prognostic model output alone, without computing visible plume lengths and
for each pollutant. Particle size distributions observations. This allows CALMET and frequency of icing and fogging from cooling
may be entered for particulate matter. CALPUFF to be run in prognostic mode for towers and other water vapor sources. The
Temporal emission factors (diurnal cycle, forecast applications. CALPRO Graphical User Interface (GUI)
monthly cycle, hour/season, wind speed/ 2. Single station surface and upper air contains options for creating graphics such as
stability class, or temperature-dependent meteorological data in CTDMPLUS data file contour plots, vector plots and other displays
emission factors) may also be entered. formats (SURFACE.DAT and PROFILE.DAT when linked to graphics software.
Arbitrarily-varying point source parameters files) or AERMOD data file formats. These d. Type of Model
may be entered from an external file. options allow a vertical variation in the
meteorological parameters but no horizontal (1) CALPUFF is a non-steady-state time-
2. Area sources: Source location and shape, and space-dependent Gaussian puff model.
release height, base elevation, initial vertical spatial variability.
3. Single station meteorological data in CALPUFF treats primary pollutants and
distribution (sz) and emission rates for each simulates secondary pollutant formation
pollutant. Particle size distributions may be ISCST3 data file format. This option does not
account for variability of the meteorological using a parameterized, quasi-linear chemical
entered for particulate matter. Temporal conversion mechanism. Pollutants treated
emission factors (diurnal cycle, monthly parameters in the horizontal or vertical,
except as provided for by the use of stability- include SO2, SO4=, NOX (i.e., NO + NO2),
cycle, hour/season, wind speed/stability HNO3, NO3-, NH3, PM–10, PM–2.5, toxic
class, or temperature-dependent emission dependent wind shear exponents and average
temperature lapse rates. pollutants and others pollutant species that
factors) may also be entered. Arbitrarily- are either inert or subject to quasi-linear
varying area source parameters may be Gridded terrain and land use data are
required as input into CALMET when Option chemical reactions. The model includes a
entered from an external file. Area sources resistance-based dry deposition model for
specified in the external file are allowed to 1 is used. Geophysical processor programs
are provided that interface the modeling both gaseous pollutants and particulate
be buoyant and their location, size, shape, matter. Wet deposition is treated using a
and other source characteristics are allowed system to standard terrain and land use data
bases available from various sources such as scavenging coefficient approach. The model
to change in time. has detailed parameterizations of complex
3. Volume sources: Source location, release the U.S. Geological Survey (USGS) and the
National Aeronautics and Space terrain effects, including terrain
height, base elevation, initial horizontal and
Administration (NASA). impingement, side-wall scrapping, and steep-
vertical distributions (sy, sz) and emission
Receptor Data: walled terrain influences on lateral plume
rates for each pollutant. Particle size
CALPUFF includes options for gridded and growth. A subgrid-scale complex terrain
distributions may be entered for particulate
non-gridded (discrete) receptors. Special module based on a dividing streamline
matter. Temporal emission factors (diurnal
subgrid-scale receptors are used with the concept divides the flow into a lift
cycle, monthly cycle, hour/season, wind
subgrid-scale complex terrain option. An component traveling over the obstacle and a
speed/stability class, or temperature-
option is provided for discrete receptors to be wrap component deflected around the
dependent emission factors) may also be
placed at ground-level or above the local obstacle.
entered. Arbitrarily-varying volume source
parameters may be entered from an external ground level (i.e., flagpole receptors). (2) The meteorological fields used by
file. Volume sources with buoyancy can be Gridded and subgrid-scale receptors are CALPUFF are produced by the CALMET
simulated by treating the source as a point placed at the local ground level only. meteorological model. CALMET includes a
source and entering initial plume size Other Input: diagnostic wind field model containing
parameters—initial (sy, sz)—to define the CALPUFF accepts hourly observations of parameterized treatments of slope flows,
initial size of the volume source. ozone concentrations for use in its chemical valley flows, terrain blocking effects, and
4. Line sources: Source location, release transformation algorithm. Monthly kinematic terrain effects, lake and sea breeze
height, base elevation, average buoyancy concentrations of ammonia concentrations circulations, a divergence minimization
parameter, and emission rates for each can be specified in the CALPUFF input file, procedure, and objective analysis of
pollutant. Building data may be entered for although higher time-resolution ammonia observational data. An energy-balance
line source emissions experiencing building variability can be computed using the scheme is used to compute sensible and
downwash effects. Particle size distributions POSTUTIL program. Subgrid-scale coastlines latent heat fluxes and turbulence parameters
may be entered for particulate matter. can be specified in its coastal boundary file. over land surfaces. A profile method is used
Temporal emission factors (diurnal cycle, Optional, user-specified deposition velocities over water. CALMET contains interfaces to
monthly cycle, hour/season, wind speed/ and chemical transformation rates can also be prognostic meteorological models such as the
stability class, or temperature-dependent entered. CALPUFF accepts the CTDMPLUS Penn State/NCAR Mesoscale Model (e.g.,
terrain and receptor files for use in its MM5; Section 12.0, ref. 86), as well as the
emission factors) may also be entered.
subgrid-scale terrain algorithm. Inflow RAMS, Ruc and Eta models.
Arbitrarily-varying line source parameters
boundary conditions of modeled pollutants
may be entered from an external file. e. Pollutant Types
can be specified in a boundary condition file.
Meteorological Data (different forms of CALPUFF may be used to model gaseous
Liquid water content variables including
meteorological input can be used by pollutants or particulate matter that are inert
cloud water/ice and precipitation water/ice
CALPUFF): or which undergo quasi-linear chemical
can be used as input for visibility analyses
1. Time-dependent three-dimensional (3– reactions, such as SO2, SO4 =, NOX (i.e., NO
and other CALPUFF modules.
D) meteorological fields generated by + NO2), HNO3, NO3-, NH3, PM–10, PM–2.5
CALMET. This is the preferred mode for c. Output
and toxic pollutants. For regional haze
running CALPUFF. Data inputs used by CALPUFF produces files of hourly analyses, sulfate and nitrate particulate
CALMET include surface observations of concentrations of ambient concentrations for components are explicitly treated.
wind speed, wind direction, temperature, each modeled species, wet deposition fluxes,
cloud cover, ceiling height, relative dry deposition fluxes, and for visibility f. Source-Receptor Relationships
humidity, surface pressure, and precipitation applications, extinction coefficients. CALPUFF contains no fundamental
(type and amount), and upper air sounding Postprocessing programs (PRTMET, limitations on the number of sources or
data (wind speed, wind direction, CALPOST, CALSUM, APPEND, and receptors. Parameter files are provided that
temperature, and height) and air-sea POSTUTIL) provide options for summing, allow the user to specify the maximum
temperature differences (over water). scaling, analyzing and displaying the number of sources, receptors, puffs, species,
Optional 3–D meteorological prognostic modeling results. CALPOST contains options grid cells, vertical layers, and other model
model output (e.g., from models such as for computing of light extinction (visibility) parameters. Its algorithms are designed to be
suitable for source-receptor distances from m. Physical Removal Dispersion Model Plus Algorithms for
tens of meters to hundreds of kilometers. Dry deposition of gaseous pollutants and Unstable Situations (CTDMPLUS). Volume 1:
g. Plume Behavior particulate matter is parameterized in terms Model Descriptions and User Instructions.
of a resistance-based deposition model. EPA Publication No. EPA–600/8–89–041.
Momentum and buoyant plume rise is Environmental Protection Agency, Research
Gravitational settling, inertial impaction, and
treated according to the plume rise equations Triangle Park, NC. (NTIS No. PB 89–181424)
Brownian motion effects on deposition of
of Briggs (1975) for non-downwashing point particulate matter is included. CALPUFF Perry, S.G., 1992. CTDMPLUS: A
sources, Schulman and Scire (1980) for line contains an option to evaluate the effects of Dispersion Model for Sources near Complex
sources and point sources subject to building plume tilt resulting from gravitational Topography. Part I: Technical Formulations.
downwash effects using the Schulman-Scire settling. Wet deposition of gases and Journal of Applied Meteorology, 31(7): 633–
downwash algorithm, and Zhang (1993) for particulate matter is parameterized in terms 645.
buoyant area sources and point sources of a scavenging coefficient approach. Availability
affected by building downwash when using
n. Evaluation Studies This model code is available on EPA’s
the PRIME building downwash method.
Stack tip downwash effects and partial Berman, S., J.Y. Ku, J. Zhang and S.T. Rao, Internet SCRAM Web site and also on
plume penetration into elevated temperature 1977. Uncertainties in estimating the mixing diskette (as PB 90–504119) from the National
depth—Comparing three mixing depth Technical Information Service (Section A.0).
inversions are included. An algorithm to treat
horizontally-oriented vents and stacks with models with profiler measurements, Abstract
rain caps is included. Atmospheric Environment, 31: 3023–3039.
Chang, J.C., P. Franzese, K. Chayantrakom CTDMPLUS is a refined point source
h. Horizontal Winds and S.R. Hanna, 2001. Evaluations of Gaussian air quality model for use in all
A three-dimensional wind field is CALPUFF, HPAC and VLSTRACK with Two stability conditions for complex terrain
Mesoscale Field Datasets. Journal of Applied applications. The model contains, in its
computed by the CALMET meteorological
Meteorology, 42(4): 453–466. entirety, the technology of CTDM for stable
model. CALMET combines an objective
Environmental Protection Agency, 1998. and neutral conditions. However,
analysis procedure using wind observations
Interagency Workgroup on Air Quality CTDMPLUS can also simulate daytime,
with parameterized treatments of slope flows, unstable conditions, and has a number of
valley flows, terrain kinematic effects, terrain Modeling (IWAQM) Phase 2 Summary Report
and Recommendations for Modeling Long- additional capabilities for improved user
blocking effects, and sea/lake breeze friendliness. Its use of meteorological data
Range Transport Impacts. EPA Publication
circulations. CALPUFF may optionally use and terrain information is different from
No. EPA–454/R–98–019. Office of Air
single station (horizontally-constant) wind other EPA models; considerable detail for
Quality Planning & Standards, Research
fields in the CTDMPLUS, AERMOD or both types of input data is required and is
Triangle Park, NC.
ISCST3 data formats. Irwin, J.S., 1997. A Comparison of supplied by preprocessors specifically
i. Vertical Wind Speed CALPUFF Modeling Results with 1997 INEL designed for CTDMPLUS. CTDMPLUS
Field Data Results. In Air Pollution Modeling requires the parameterization of individual
Vertical wind speeds are not used
and its Application, XII. Edited by S.E. hill shapes using the terrain preprocessor and
explicitly by CALPUFF. Vertical winds are
Gyrning and N. Chaumerliac. Plenum Press, the association of each model receptor with
used in the development of the horizontal
New York, NY. a particular hill.
wind components by CALMET.
Irwin, J.S., J.S. Scire and D.G. Strimaitis, a. Recommendation for Regulatory Use
j. Horizontal Dispersion 1996. A Comparison of CALPUFF Modeling
CTDMPLUS is appropriate for the
Turbulence-based dispersion coefficients Results with CAPTEX Field Data Results. In
following applications:
provide estimates of horizontal plume Air Pollution Modeling and its Application,
XI. Edited by S.E. Gyrning and F.A. • Elevated point sources;
dispersion based on measured or computed • Terrain elevations above stack top;
values of sv. The effects of building Schiermeier. Plenum Press, New York, NY.
Morrison, K, Z–X Wu, J.S. Scire, J. Chenier • Rural or urban areas;
downwash and buoyancy-induced dispersion • Transport distances less than 50
are included. The effects of vertical wind and T. Jeffs-Schonewille, 2003. CALPUFF-
Based Predictive and Reactive Emission kilometers; and
shear are included through the puff splitting • One hour to annual averaging times
algorithm. Options are provided to use Control System. 96th A&WMA Annual
Conference & Exhibition, 22–26 June 2003; when used with a post-processor program
Pasquill-Gifford (rural) and McElroy-Pooler such as CHAVG.
San Diego, CA.
(urban) dispersion coefficients. Initial plume
Schulman, L.L., D.G. Strimaitis and J.S. b. Input Requirements
size from area or volume sources is allowed.
Scire, 2000. Development and evaluation of (1) Source data: For each source, user
k. Vertical Dispersion the PRIME Plume Rise and Building supplies source location, height, stack
Turbulence-based dispersion coefficients Downwash Model. JAWMA, 50: 378–390. diameter, stack exit velocity, stack exit
provide estimates of vertical plume Scire, J.S., Z–X Wu, D.G. Strimaitis and temperature, and emission rate; if variable
dispersion based on measured or computed G.E. Moore, 2001. The Southwest Wyoming emissions are appropriate, the user supplies
values of sw. The effects of building Regional CALPUFF Air Quality Modeling hourly values for emission rate, stack exit
downwash and buoyancy-induced dispersion Study—Volume I. Prepared for the Wyoming velocity, and stack exit temperature.
are included. Vertical dispersion during Dept. of Environmental Quality. Available (2) Meteorological data: For applications of
convective conditions is simulated with a from Earth Tech at http://www.src.com. CTDMPLUS, multiple level (typically three
Strimaitis, D.G., J.S. Scire and J.C. Chang, or more) measurements of wind speed and
probability density function (pdf) model
1998. Evaluation of the CALPUFF Dispersion direction, temperature and turbulence (wind
based on Weil et al. (1997). Options are
Model with Two Power Plant Data Sets. fluctuation statistics) are required to create
provided to use Pasquill-Gifford (rural) and
Tenth Joint Conference on the Application of the basic meteorological data file
McElroy-Pooler (urban) dispersion
Air Pollution Meteorology, Phoenix, Arizona. (‘‘PROFILE’’). Such measurements should be
coefficients. Initial plume size from area or
American Meteorological Society, Boston, obtained up to the representative plume
volume sources is allowed. MA. January 11–16, 1998. height(s) of interest (i.e., the plume height(s)
A.5 Complex Terrain Dispersion Model under those conditions important to the
Gas phase chemical transformations are Plus Algorithms for Unstable Situations determination of the design concentration).
treated using parameterized models of SO2 (CTDMPLUS) The representative plume height(s) of interest
conversion to SO4= and NO conversion to should be determined using an appropriate
NO3-, HNO3, and NO2. Organic aerosol Reference complex terrain screening procedure (e.g.,
formation is treated. The POSTUTIL program Perry, S.G., D.J. Burns, L.H. Adams, R.J. CTSCREEN) and should be documented in
contains an option to re-partition HNO3 and Paine, M.G. Dennis, M.T. Mills, D.G. the monitoring/modeling protocol. The
NO3- in order to treat the effects of ammonia Strimaitis, R.J. Yamartino and E.M. Insley, necessary meteorological measurements
limitation. 1989. User’s Guide to the Complex Terrain should be obtained from an appropriately
sited meteorological tower augmented by e. Pollutant Types observed vertical turbulence intensity, e.g.,
SODAR and/or RASS if the representative CTDMPLUS may be used to model non- sw (standard deviation of the vertical velocity
plume height(s) of interest is above the levels reactive, primary pollutants. fluctuation). In simulating unstable
represented by the tower measurements. (convective) conditions, CTDMPLUS relies
Meteorological preprocessors then create a f. Source-Receptor Relationship on a skewed, bi-Gaussian probability density
SURFACE data file (hourly values of mixed Up to 40 point sources, 400 receptors and function (pdf) description of the vertical
layer heights, surface friction velocity, 25 hills may be used. Receptors and sources velocities to estimate the vertical distribution
Monin-Obukhov length and surface are allowed at any location. Hill slopes are of pollutant concentration.
roughness length) and a RAWINsonde data assumed not to exceed 15°, so that the l. Chemical Transformation
file (upper air measurements of pressure, linearized equation of motion for Boussinesq
flow are applicable. Receptors upwind of the Chemical transformation is not treated by
temperature, wind direction, and wind
impingement point, or those associated with CTDMPLUS.
speed).
(3) Receptor data: receptor names (up to any of the hills in the modeling domain, m. Physical Removal
400) and coordinates, and hill number (each require separate treatment. Physical removal is not treated by
receptor must have a hill number assigned). g. Plume Behavior CTDMPLUS (complete reflection at the
(4) Terrain data: user inputs digitized (1) As in CTDM, the basic plume rise ground/hill surface is assumed).
contour information to the terrain algorithms are based on Briggs’ (1975) n. Evaluation Studies
preprocessor which creates the TERRAIN recommendations.
data file (for up to 25 hills). Burns, D.J., L.H. Adams and S.G. Perry,
(2) A central feature of CTDMPLUS for 1990. Testing and Evaluation of the
c. Output neutral/stable conditions is its use of a CTDMPLUS Dispersion Model: Daytime
(1) When CTDMPLUS is run, it produces critical dividing-streamline height (Hc) to Convective Conditions. Environmental
separate the flow in the vicinity of a hill into Protection Agency, Research Triangle Park,
a concentration file, in either binary or text
two separate layers. The plume component in NC.
format (user’s choice), and a list file
the upper layer has sufficient kinetic energy Paumier, J.O., S.G. Perry and D.J. Burns,
containing a verification of model inputs, i.e.,
to pass over the top of the hill while
• Input meteorological data from 1990. An Analysis of CTDMPLUS Model
streamlines in the lower portion are Predictions with the Lovett Power Plant Data
‘‘SURFACE’’ and ‘‘PROFILE’’.
constrained to flow in a horizontal plane
• Stack data for each source. Base. Environmental Protection Agency,
around the hill. Two separate components of
• Terrain information. Research Triangle Park, NC.
CTDMPLUS compute ground-level Paumier, J.O., S.G. Perry and D.J. Burns,
• Receptor information.
concentrations resulting from plume material 1992. CTDMPLUS: A Dispersion Model for
• Source-receptor location (line printer
in each of these flows. Sources near Complex Topography. Part II:
map). (3) The model calculates on an hourly (or
(2) In addition, if the case-study option is Performance Characteristics. Journal of
appropriate steady averaging period) basis Applied Meteorology, 31(7): 646–660.
selected, the listing includes: how the plume trajectory (and, in stable/
• Meteorological variables at plume height. neutral conditions, the shape) is deformed by A.6 Offshore and Coastal Dispersion Model
• Geometrical relationships between the each hill. Hourly profiles of wind and (OCD)
source and the hill. temperature measurements are used by
• Plume characteristics at each receptor, CTDMPLUS to compute plume rise, plume
Reference
i.e., penetration (a formulation is included to DiCristofaro, D.C. and S.R. Hanna, 1989.
—Distance in along-flow and cross flow handle penetration into elevated stable OCD: The Offshore and Coastal Dispersion
direction layers, based on Briggs (1984)), convective Model, Version 4. Volume I: User’s Guide,
—Effective plume-receptor height difference scaling parameters, the value of Hc, and the and Volume II: Appendices. Sigma Research
—Effective sy & sz values, both flat terrain Froude number above Hc. Corporation, Westford, MA. (NTIS Nos. PB
and hill induced (the difference shows the 93–144384 and PB 93–144392; also available
h. Horizontal Winds
effect of the hill) at http://www.epa.gov/scram001/)
—Concentration components due to WRAP, CTDMPLUS does not simulate calm
meteorological conditions. Both scalar and Availability
LIFT and FLAT.
(3) If the user selects the TOPN option, a vector wind speed observations can be read This model code is available on EPA’s
summary table of the top 4 concentrations at by the model. If vector wind speed is Internet SCRAM Web site and also on
each receptor is given. If the ISOR option is unavailable, it is calculated from the scalar diskette (as PB 91–505230) from the National
selected, a source contribution table for every wind speed. The assignment of wind speed Technical Information Service (see Section
hour will be printed. (either vector or scalar) at plume height is A.0). Official contact at Minerals
(4) A separate disk file of predicted (1-hour done by either: Management Service: Mr. Dirk Herkhof,
only) concentrations (‘‘CONC’’) is written if • Interpolating between observations Parkway Atrium Building, 381 Elden Street,
the user chooses this option. Three forms of above and below the plume height, or Herndon, VA 20170, Phone: (703) 787–1735.
output are possible: • Extrapolating (within the surface layer)
from the nearest measurement height to the Abstract
(i) A binary file of concentrations, one
value for each receptor in the hourly plume height. (1) OCD is a straight-line Gaussian model
sequence as run; i. Vertical Wind Speed developed to determine the impact of
(ii) A text file of concentrations, one value offshore emissions from point, area or line
Vertical flow is treated for the plume sources on the air quality of coastal regions.
for each receptor in the hourly sequence as component above the critical dividing
run; or OCD incorporates overwater plume transport
streamline height (Hc); see ‘‘Plume and dispersion as well as changes that occur
(iii) A text file as described above, but with Behavior’’.
a listing of receptor information (names, as the plume crosses the shoreline. Hourly
positions, hill number) at the beginning of j. Horizontal Dispersion meteorological data are needed from both
the file. Horizontal dispersion for stable/neutral offshore and onshore locations. These
(3) Hourly information provided to these conditions is related to the turbulence include water surface temperature, overwater
files besides the concentrations themselves velocity scale for lateral fluctuations, sv, for air temperature, mixing height, and relative
includes the year, month, day, and hour which a minimum value of 0.2 m/s is used. humidity.
information as well as the receptor number Convective scaling formulations are used to (2) Some of the key features include
with the highest concentration. estimate horizontal dispersion for unstable platform building downwash, partial plume
conditions. penetration into elevated inversions, direct
d. Type of Model use of turbulence intensities for plume
CTDMPLUS is a refined steady-state, point k. Vertical Dispersion dispersion, interaction with the overland
source plume model for use in all stability Direct estimates of vertical dispersion for internal boundary layer, and continuous
conditions for complex terrain applications. stable/neutral conditions are based on shoreline fumigation.
a. Recommendations for Regulatory Use f. Source-Receptor Relationship theory as default in the model. For very
OCD has been recommended for use by the (1) Up to 250 point sources, 5 area sources, stable conditions, vertical dispersion is also
Minerals Management Service for emissions or 1 line source and 180 receptors may be a function of lapse rate.
located on the Outer Continental Shelf (50 FR used. (2) Vertical dispersion may be enhanced
12248; 28 March 1985). OCD is applicable for (2) Receptors and sources are allowed at because of obstructions near the source. A
overwater sources where onshore receptors any location. virtual source technique is used to simulate
are below the lowest source height. Where (3) The coastal configuration is determined the initial plume dilution due to downwash.
onshore receptors are above the lowest by a grid of up to 3600 rectangles. Each (3) Formulas recommended by Pasquill
source height, offshore plume transport and element of the grid is designated as either (1976) are used to calculate buoyant plume
dispersion may be modeled on a case-by-case land or water to identify the coastline. enhancement.
basis in consultation with the appropriate (4) At the water/land interface, the change
g. Plume Behavior to overland dispersion rates is modeled using
reviewing authority (paragraph 3.0(b)).
(1) As in ISC, the basic plume rise a virtual source. The overland dispersion
b. Input Requirements algorithms are based on Briggs’ rates can be calculated from either vertical
(1) Source data: Point, area or line source recommendations. turbulence intensity or the Pasquill-Gifford
location, pollutant emission rate, building (2) Momentum rise includes consideration coefficients. The change is implemented
height, stack height, stack gas temperature, of the stack angle from the vertical. where the plume intercepts the rising
stack inside diameter, stack gas exit velocity, (3) The effect of drilling platforms, ships, internal boundary layer.
stack angle from vertical, elevation of stack or any overwater obstructions near the source 1. Chemical Transformation
base above water surface and gridded are used to decrease plume rise using a
specification of the land/water surfaces. As Chemical transformations are treated using
revised platform downwash algorithm based
an option, emission rate, stack gas exit exponential decay. Different rates can be
on laboratory experiments.
velocity and temperature can be varied specified by month and by day or night.
(4) Partial plume penetration of elevated
hourly. inversions is included using the suggestions m. Physical Removal
(2) Meteorological data (over water): Wind of Briggs (1975) and Weil and Brower (1984). Physical removal is also treated using
direction, wind speed, mixing height, relative (5) Continuous shoreline fumigation is exponential decay.
humidity, air temperature, water surface parameterized using the Turner method
temperature, vertical wind direction shear n. Evaluation Studies
where complete vertical mixing through the
(optional), vertical temperature gradient thermal internal boundary layer (TIBL) DiCristofaro, D.C. and S.R. Hanna, 1989.
(optional), turbulence intensities (optional). occurs as soon as the plume intercepts the OCD: The Offshore and Coastal Dispersion
(2) Meteorological data: TIBL. Model. Volume I: User’s Guide. Sigma
Over land: Surface weather data from a Research Corporation, Westford, MA.
preprocessor such as PCRAMMET which h. Horizontal Winds
Hanna, S.R., L.L. Schulman, R.J. Paine and
provides hourly stability class, wind (1) Constant, uniform wind is assumed for J.E. Pleim, 1984. The Offshore and Coastal
direction, wind speed, ambient temperature, each hour. Dispersion (OCD) Model User’s Guide,
and mixing height are required. (2) Overwater wind speed can be estimated Revised. OCS Study, MMS 84–0069.
Over water: Hourly values for mixing from overland wind speed using relationship Environmental Research & Technology, Inc.,
height, relative humidity, air temperature, of Hsu (1981). Concord, MA. (NTIS No. PB 86–159803).
and water surface temperature are required; (3) Wind speed profiles are estimated using Hanna, S.R., L.L. Schulman, R.J. Paine, J.E.
if wind speed/direction are missing, values similarity theory (Businger, 1973). Surface Pleim and M. Baer, 1985. Development and
over land will be used (if available); vertical layer fluxes for these formulas are calculated Evaluation of the Offshore and Coastal
wind direction shear, vertical temperature from bulk aerodynamic methods. Dispersion (OCD) Model. Journal of the Air
gradient, and turbulence intensities are i. Vertical Wind Speed Pollution Control Association, 35: 1039–
optional. 1047.
(3) Receptor data: Location, height above Vertical wind speed is assumed equal to
Hanna, S.R. and D.C. DiCristofaro, 1988.
local ground-level, ground-level elevation zero.
Development and Evaluation of the OCD/API
above the water surface. j. Horizontal Dispersion Model. Final Report, API Pub. 4461,
c. Output (1) Lateral turbulence intensity is American Petroleum Institute, Washington,
recommended as a direct estimate of DC.
(1) All input options, specification of
sources, receptors and land/water map horizontal dispersion. If lateral turbulence A. REFERENCES
including locations of sources and receptors. intensity is not available, it is estimated from
boundary layer theory. For wind speeds less Benson, P.E., 1979. CALINE3—A Versatile
(2) Summary tables of five highest Dispersion Model for Predicting Air
concentrations at each receptor for each than 8 m/s, lateral turbulence intensity is
assumed inversely proportional to wind Pollution Levels Near Highways and Arterial
averaging period, and average concentration Streets. Interim Report, Report Number
for entire run period at each receptor. speed.
(2) Horizontal dispersion may be enhanced FHWA/CA/TL–79/23. Federal Highway
(3) Optional case study printout with Administration, Washington, DC.
hourly plume and receptor characteristics. because of obstructions near the source. A
virtual source technique is used to simulate Briggs, G.A., 1975. Plume Rise Predictions.
Optional table of annual impact assessment Lectures on Air Pollution and Environmental
from non-permanent activities. the initial plume dilution due to downwash.
(3) Formulas recommended by Pasquill Impact Analyses. American Meteorological
(4) Concentration files written to disk or Society, Boston, MA, pp. 59–111.
tape can be used by ANALYSIS (1976) are used to calculate buoyant plume
enhancement and wind direction shear Briggs, G.A., 1984. Analytical
postprocessor to produce the highest Parameterizations of Diffusion: The
concentrations for each receptor, the enhancement.
(4) At the water/land interface, the change Convective Boundary Layer. Journal of
cumulative frequency distributions for each Climate and Applied Meteorology, 24(11):
receptor, the tabulation of all concentrations to overland dispersion rates is modeled using
a virtual source. The overland dispersion 1167–1186.
exceeding a given threshold, and the Environmental Protection Agency, 1980.
manipulation of hourly concentration files. rates can be calculated from either lateral
turbulence intensity or Pasquill-Gifford Recommendations on Modeling (October
d. Type of Model curves. The change is implemented where 1980 Meetings). Appendix G to: Summary of
OCD is a Gaussian plume model the plume intercepts the rising internal Comments and Responses on the October
constructed on the framework of the MPTER boundary layer. 1980 Proposed Revisions to the Guideline on
model. Air Quality Models. Meteorology and
k. Vertical Dispersion Assessment Division, Office of Research and
e. Pollutant Types (1) Observed vertical turbulence intensity Development, Research Triangle Park, NC
OCD may be used to model primary is not recommended as a direct estimate of 27711.
pollutants. Settling and deposition are not vertical dispersion. Turbulence intensity Environmental Protection Agency, 1998.
treated. should be estimated from boundary layer Interagency Workgroup on Air Quality
Modeling (IWAQM) Phase 2 Summary Report Impact Model. Systems Applications, Inc., Modeling. Journal of the Air Pollution
and Recommendations for Modeling Long- San Rafael, CA. Control Association, 23: 598–600.
Range Transport Impacts. Publication No. Pasquill, F., 1976. Atmospheric Dispersion Snyder, W.H., R.S. Thompson, R.E.
EPA–454/R–98–019. (NTIS No. PB 99– Parameters in Gaussian Plume Modeling Part Eskridge, R.E. Lawson, I.P. Castro, J.T. Lee,
121089). II. Possible Requirements for Change in the J.C.R. Hunt, and Y. Ogawa, 1985. The
Escoffier-Czaja, C. and J.S. Scire, 2002. The Turner Workbook Values. Publication No. structure of the strongly stratified flow over
Effects of Ammonia Limitation on Nitrate EPA–600/4–76–030b. Office of Air Quality hills: Dividing streamline concept. Journal of
Aerosol Formation and Visibility Impacts in Planning & Standards, Research Triangle Fluid Mechanics, 152: 249–288.
Class I Areas. Twelfth AMS/AWMA Park, NC 27711. Turner, D.B., 1969. Workbook of
Conference on the Application of Air Petersen, W.B., 1980. User’s Guide for Atmospheric Dispersion Estimates. PHS
Pollution Meteorology, 20–24 May 2002; HIWAY–2 A Highway Air Pollution Model. Publication No. 999–26. U.S. Environmental
Norfolk, VA. Publication No. EPA–600/8–80–018. Office of Protection Agency, Research Triangle, Park,
Gifford, F.A., Jr. 1976. Turbulent Diffusion Research & Development, Research Triangle NC 27711.
Typing Schemes—A Review. Nuclear Safety, Park, NC 27711. (NTIS PB 80–227556) Weil, J.C. and R.P. Brower, 1984. An
17: 68–86. Rao, T.R. and M.T. Keenan, 1980. Updated Gaussian Plume Model for Tall
Horst, T.W., 1983. A Correction to the Suggestions for Improvement of the EPA– Stacks. Journal of the Air Pollution Control
Gaussian Source-depletion Model. In HIWAY Model. Journal of the Air Pollution Association, 34: 818–827.
Precipitation Scavenging, Dry Deposition and Control Association, 30: 247–256 (and Weil, J.C., 1996. A new dispersion
Resuspension. H. R. Pruppacher, R.G. reprinted as Appendix C in Petersen, 1980). algorithm for stack sources in building
Semonin and W.G.N. Slinn, eds., Elsevier, Robe, F.R., Z–X. Wu and J.S. Scire, 2002: wakes, Paper 6.6. Ninth Joint Conference on
NY. Real-time SO2 Forecasting System with Applications of Air Pollution Meteorology
Hsu, S.A., 1981. Models for Estimating Combined ETA Analysis and CALPUFF with A&WMA, January 28–February 2, 1996.
Offshore Winds from Onshore Meteorological Modeling. Proceedings of the 8th Atlanta, GA.
Measurements. Boundary Layer Meteorology, International Conference on Harmonisation Weil, J.C., L.A. Corio, and R.P. Brower,
20: 341–352. within Atmospheric Dispersion Modelling 1997. A PDF dispersion model for buoyant
Huber, A.H. and W.H. Snyder, 1976. for Regulatory Purposes, 14–17 October 2002; plumes in the convective boundary layer.
Building Wake Effects on Short Stack Sofia, Bulgaria. Journal of Applied Meteorology, 36: 982–
Effluents. Third Symposium on Atmospheric Schulman, L.L. and J.S. Scire, 1980: 1003.
Turbulence, Diffusion and Air Quality, Buoyant Line and Point Source (BLP) Zhang, X., 1993. A computational analysis
American Meteorological Society, Boston, dispersion model user’s guide. The of the rise, dispersion, and deposition of
MA. Aluminum Association; Washington, DC. buoyant plumes. Ph.D. Thesis, Massachusetts
Irwin, J.S., 1979. A Theoretical Variation of (See A.2 in this appendix.) Institute of Technology, Cambridge, MA.
the Wind Profile Power-Law Exponent as a Schulman, L.L. and S.R. Hanna, 1986. Zhang, X. and A.F. Ghoniem, 1993. A
Function of Surface Roughness and Stability. Evaluation of Downwash Modification to the computational model for the rise and
Atmospheric Environment, 13: 191–194. Industrial Source Complex Model. Journal of dispersion of wind-blown, buoyancy-driven
Liu, M.K. et al., 1976. The Chemistry, the Air Pollution Control Association, 36: plumes—I. Neutrally stratified atmosphere.
Dispersion, and Transport of Air Pollutants 258–264. Atmospheric Environment, 15: 2295–2311.
Emitted from Fossil Fuel Power Plants in Segal, H.M., 1983. Microcomputer [FR Doc. 05–21627 Filed 11–8–05; 8:45 am]
California: Data Analysis and Emission Graphics in Atmospheric Dispersion BILLING CODE 6560–50–P

Rule: Air Quality Implementation Plans: Preparation, Adoption, and Submittal&#8212 Air Quality Models Guidelines

Загружено:

Сведения о документе

Оригинальное название

Авторское право

Доступные форматы

Поделиться этим документом

Поделиться или встроить документ

Параметры публикации

Этот документ был вам полезен?

Это неприемлемый материал?

Авторское право:

Доступные форматы

Rule: Air Quality Implementation Plans: Preparation, Adoption, and Submittal&#8212 Air Quality Models Guidelines

Загружено:

Авторское право:

Доступные форматы

Wednesday,

ENVIRONMENTAL PROTECTION telephone (919) 541–5562. SCRAM) at: www.epa.gov/scram001.

Appendix W to Part 51—Guideline on 4.0 Stationary-Source Models 10.2.3 Emission Limits

TABLE 4–1A.—NEUTRAL/STABLE METEOROLOGICAL MATRIX FOR CTSCREEN

Variable Specific values

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

Exceptions: (2) If sw = 0.75 m/s and U ≥ 3.0 m/s, then (4) sw ≤ sv

TABLE 4–1B.—UNSTABLE/CONVECTIVE METEOROLOGICAL MATRIX FOR CTSCREEN

Variable Specific values

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

TABLE 8–1.—MODEL EMISSION INPUT DATA FOR POINT SOURCES 1

Proposed Major New or Modified Source

k. Vertical Dispersion Availability c. Output

Вам также может понравиться