450481501
REGIONAL WORKSHOPS ON
AIR QUALITY MODELING:
A SIWIARY REPORT
APRIL 1981
SOURCE RECEPTOR ANALYSIS BRANCH
MONITORING AND DATA .ANALYSIS DIVISION
OFFICE OF AIR QUALITY PLANNING AND STANDARDS
U, S, ENVIRONMENTAL PROTECTION AGENCY
RESEARCH TRIANGLE PARK, NORTH CAROLINA
.-/ / - 5 of
~- rv»"v>
mi
-------
TABLE OF CONTENTS
Page
1.0 INTRODUCTION 1
2.0 DATA BASES 3
2.1 Discussion 3
2.2 Recommendations * 3
2.2.1 Acquisition of Data Bases 3
2.2.2 Background Concentrations 4
2.2.3 Source Data 4
2.2.4 Meteorological Data 4
3.0 FLAT TERRAIN MODELS 6
3.1 Discussion 6
3.2 Recommendations 6
3.2.1 Screening Techniques 6
3.2.2 Refined Analytical Techniques . V. . . '." . 6
3.2.3 Model Options 8
4.0 COMPLEX TERRAIN MODELS ,. 9
4.1 Discussion 9
4.2 Recommendations 9
4.2.1 Screening Techniques 9
4.2.2 Refined Analytical Techniques 10
5.0 MOBILE SOURCE MODELS 11
6.0 GENERAL MODELING ISSUES 12
6.1 Discussion 12
6.2 Recommendations 12
6.2.1 Design Concentrations 12
6.2.2 Critical Receptor Sites 12
6.2.3 Long-Range Transport 13
6.2.4 Pollutant Half-Life 14
6.2.5 Urban/Rural Classification 14
6.2.6 General Model Evaluation 15
11
-------
7.0 USE OF NONGUIDELINE MODELS 16
7.1 Discussion . . . 16
7.2 Definition of Guideline vs Nonguideline
Models ' 16
7.3 Recommendations 17
8.0 USE OF MEASURED DATA IN LIEU OF MODEL ESTIMATES 18
8.1 Discussion 18
8.2 Recommendations ..... 18
9.0 REGIONAL MODELING PROCEDURES . . -. 21
9.1 Discussion 21
9.2 Recommendations 21
Appendix A. Acquisition of Site Specific Meteorological Data . . A-l
Appendix B. Air Quality Analysis Checklist B-l
(111)
-------
1.0 INTRODUCTION
The requirements placed on air quality control agencies by the
Clean Air Act have dramatically increased the need for improved air
quality modeling. The resulting increase'in the use of models has also
led to a substantial increase in the number and complexity of situations
in which models are employed. The modeling guideline (Guideline on Air
Quality Models, EPA-450/2-78-027, April 1978) addresses many of the
problems in this relatively new and growing field, but much is left to
the discretion of the reviewing agency since many complex problems are
best solved on a case-by-case basis. However, because of the variety of
technically correct solutions to any complex problem, different approaches
with differing results have led to inconsistency in model applications
from Region to Region. In an effort to improve consistency in the use
of modeling techniques, three in-house workshops have been held since
1978. These workshops provide a forum for the Regional Office and Head-
quarters groups to discuss common problem areas and arrive at generally
acceptable solutions.
Many recommendations were made in the course of the workshops.
These have been reviewed by OAQPS and some have necessarily been modi-
fied and supplemented to ensure consistency with other modeling policies.
This report clarifies preferred data bases and procedures for the appli-
cation of specific models and modeling techniques in situations where
the guideline permits a case-by-case analysis.
-------
Recommendations contained in this report should be followed by
the EPA Regional Offices until such time as the 1978 guideline is
formally revised. Issues concerning the use of models not specified
in this summary report or in the 1978 guideline, should be directed
to the OAQPS Model Clearinghouse for review. The current procedures
for submitting issues are provided in the Clearinghouse Operating Plan,
-------
2.0 DATA BASES
2.1 Discussion
Estimated concentrations can vary widely depending on the
source, meteorological, and air quality data used in preparing the
estimates. Thus the need for consistency in the use of data and in the
selection of data bases is apparent. Also, an accurate and reliable air
quality data base is needed to evaluate the performance of a model.
«•
Inconsistencies occur because adequate data frequently are not
available for model input. Requirements for pre-application monitoring
under PSD have alleviated some of the inconsistencies in data collection
and use. However, additional guidance is still needed in the collection
and interpretation of meteorological data.
Also, appropriate source data to reflect-short-term variations
in emissions are often unavailable. The relationship of source emission
data to worst-case conditions can be another area of inconsistency.
This section identifies a few of the more frequent problem
areas and provides recommendations to ensure consistency in the select-
ion and use of data.
2.2 Recommendations
2.2.1 Acquisition of Data Bases
Guidance provided in the "Ambient Monitoring Guidelines
for Prevention of Significant Deterioration (PSD)," EPA-450/4-80-012,
November 1980 should be used for the establishment of a special monitor-
ing network for air quality analyses, including both air quality and
meteorological monitoring techniques. Additional information is avail-
able in 40CFR Part 58 and in the quality assurance and site selection
EPA guidance documents published on a pollutant-by-pollutant basis. The
EPA Regional Office should review the network design prior to operation.
-------
2.2.2 Background Concentrations
Techniques discussed in the Guideline on Air Quality
Models should be used in establishing background concentrations.
2.2.3 Source Data
The load or operating condition of a plant that causes
the highest ground-level concentrations should be determined through a
screening analysis and this load should be used to establish emission
limitations. As a minimum all sources should be modeled using 100
percent design capacity; however, when modeling large sources, e.g., 500
MW power plants or equivalent, 50 and 75 percent capacity should also be
modeled.
Hourly sequential emissions determined for existing
sources from continuous in-stack monitoring should be used in model
evaluation where possible. Hourly emissions are critical where short-
term concentrations are of concern in such evaluations.
2.2.4 Meteorological Data
Five years~of representative meteorological data should
be used when estimating concentrations with an air quality model.
Consecutive years from the most recently available five-year period are
preferred. The meteorological data may be data collected either on-site
or at the nearest National Weather Service (NWS) station. .If the source
is large, e.g., emissions equivalent to a 500 MW power plant, the use of
five years of NWS meteorological data or at least one year of on-site
data is required.
Five years of on-site data are often not available.
When considering shorter periods of meteorological data, care must be
taken to ensure that the data used contain the appropriate worst-case
conditions. On-site data should also be subjected to quality assurance
procedures that will ensure that the data is at least as accurate and in
as much detail as NWS data.
Hourly average wind directions reported to the nearest
degree should be used where on-site data are used. The CRSTER randomi-
zation sequence (i.e., the established sequential random number set
designated for use with the meteorological preprocessor to CRSTER)
should be used with'NWS wind data.
The surface wind reference height used in the model
should be defined to agree with the actual height of the surface wind
sensor. When wind is monitored at heights closer to plume height, the
-------
wind direction should be used to define the plume transport and the
speed should be utilized to develop the appropriate vertical wind speed
profile.
Guidance provided in Appendix A should be followed in
the design of site-specific, on-site meteorological data collection
programs.
-------
3.0 FLAT TERRAIN MODELS
3.1 Discussion
Flat terrain, as used here, is considered to be an area where
terrain features are all lower in elevation than the top of the stack of
the source in question. Most Gaussian models perform adequately in such
situations.
A number of models have been made available by EPA and others
for those applications where receptors are located at elevations less
than the top of the stack. However, inconsistencies have resulted from
the use of these models. Such inconsistencies occur in part because
models may be developed at different times for specific applications and
the various algorithms are improved, changed or added to accommodate a
specific problem or to reflect recent research. This section provides
recommendations to resolve these inconsistencies without limiting the
range of applicability of flat terrain models.
3.2 Recommendations
3.2.1 Screening Techniques
Screening techniques and options as provided in "Guide-
lines for Air Quality Maintenance Planning and Analysis Volume 10 (R):
Procedures for Evaluating Air Quality Impact of New Stationary Sources"
should be used.
Where possible, screening procedures should be site and
problem specific. Consideration should be given to: (1) terrain; (2)
urban or rural dispersion coefficients; and (3) worst-case conditions
when representative meteorological data or applicable detailed modeling
techniques are not available. If screening is the sole basis for the
analysis, adequate justification and documentation should be required
for the use of averaging time factors.
3.2.2 Refined Analytical Techniques
The following table lists the preferred models for the
indicated applications. These models should be used in Regional Office
-------
applications of models. For use of models and in applications that do
not appear in this table or in the 1978 guideline, Regional Offices
should follow Section 7 of this report.
Table 1. Preferred Model Use
SHORT TERM
Single Source
Multiple Source
Industrial Complexes
LONG TERM
Single Source
Multiple Source
Industrial Complexes
Rural
Urban
Rural
Urban
Rural/Urban
Rural
-Urban
Rural
Urban
Rural/Urban
MODEL
CRSTER
RAM
MPTER
RAM
ISC
CRSTER
CDMQC or RAM*
MPTER
CDMQC or RAM*
ISC
For all model applications in a rural area, the CRSTER
techniques for wind speed profile, plume rise and terrain adjustment
should be used unless other techniques can be shown on a case-by-case
"basis to provide more appropriate and accurate estimates.
Dispersion coefficients appropriate to either urban or
rural settings should be used in accordance with Section 6.2.5. Sector
averaging should be accepted only for seasonal or annual estimates where
estimates are based on statistically summarized meteorological data.
*The choice of RAM or, CDMQC in urban applications is a function of the
number of sources and the size of the area to be modeled, e.g., if only
three or four sources in an urban area are to be modeled, RAM should be
used.
-------
3.2.3 Model Options
The options that are found in the ISC, CRSTER, MPTER,
and PTPLU have greatly increased the technical options available. To
ensure consistency in the use of these options, Regional Office users -
should follow the guidance below:
a. Stack Tip Downwash (CRSTER, MPTER, ISC, PTPLU*)
This option should not be used unless demonstrated
to be applicable on a case-by-case basis. Although there is evidence
that this phenomenon can occur, there are no data to support wide use of
the option.
b. Plume Rise (CRSTER, MPTER, ISC, PTPLU)
In all cases, except stable conditions in complex
terrain, the final plume rise option should be used. The restriction on
the use of gradual plume rise is based upon the lack of specific data
needed to quantify the dispersion during plume rise. In complex terrain
where plume impaction is the identified problem, the use of transitional
plume rise, during stable conditions, may be required to ensure that
impaction on close-in terrain is considered.
c. Rural/Urban Options (ISC)
, - *.
The selection of the rural or urban option should
be based upon the determinations as outlined in Section 6.2.5 for deter-
mining whether an area is urban or rural.
d. Momentum Plume Rise (ISC, CRSTER, MPTER, PTPLU)
This is optional in the CRSTER and MPTER models and
an integral part of the ISC and PTPLU models. It should be used in the
CRSTER and MPTER models.
*PTPLU is found in UNAMAP (Version 4) and is a replacement for PTMAX.
-------
4.0 COMPLEX TERRAIN MODELING
4.1 Discussion
Although the need for a refined complex terrain dispersion
model has been acknowledged for several years, such a model has not yet
been developed. The lack of extensive data bases and basic knowledge
concerning the behavior of atmospheric variables in the vicinity of
complex terrain presents a considerable obstacle to the solution of the
»
problems and the development of a refined model.
This is complicated by the fact that each complex terrain
location can be considered unique. A comprehensive study was begun
recently by ORD in the vicinity of a 3-dimensional hill in Idaho, the
purpose of which is to gain the basic knowledge of dispersion in complex
terrain. Practical information from this study is still forthcoming.
Until changes in the physical behavior of plumes in complex terrain are
documented and new mathematical constructs developed, the existing
dispersion algorithms adapted to complex terrain must be used.
For the purpose of this report, complex terrain is defined as
any terrain exceeding the height of the stack being modeled.
4.2 Recommendations
4.2.1 Screening Techniques
The screening technique recommended for use in complex
terrain situations to determine annual and 24-hour average concentrations
is the Valley Model. The following worst-case assumptions should be
used to determine 24-hour averages: (1) P-G stability of "F"; (2) wind
speed of 2.5 m/s; (3) six hours of occurrence. For screening, the use
of a sector greater than 22-1/2° in the Valley Model should not be
allowed. Full ground reflection should always be used in screening
analyses.
-------
Multiple sources should be treated individually in the
Valley Model and the concentrations for specific wind directions summed.
Only one wind direction should be used for 24-hour averages (see User's
Manual, pages 2-15) even if individual runs are made for each source.
The receptor grid found in the Valley Model User's
Guide may not be sufficient for all analyses if only one geographical
scale factor is used. The Valley Model is very sensitive to ground-
level elevation at the receptor, and the use of the standard polar grid
could miss the worst-case receptor. If this situation occurs, the user
should choose an additional set of receptors at appropriate downwind
distances whose elevations are equal to plume height minus 10 meters.
4.2.2 Refined Analytical Techniques
If the results of the screening analysis demonstrate a
possible violation of NAAQS or the controlling PSD increments, a more
refined analysis should be conducted. In the absence of other models
demonstrated to be more appropriate, the Valley Model is acceptable
for selecting emission limitations in complex terrain situations.
10
-------
5.0 MOBILE SOURCE MODELS
Regional meteorologists have not been involved in significant
consistency problems with carbon monoxide or ozone models. Some guid-
ance is found in the 1978 Guideline on Air Quality Models, and additional
guidance with respect to the use of models and data bases for SIP
revisions is contained in the Federal Register Volume 46, No. 14, p.
7182, entitled, "State Implementation Plans: Approval of 1982 Ozone and
Carbon Monoxide Plan Revisions for Areas Needing an Attainment Date
Extension."
11
-------
6.0 GENERAL MODELING ISSUES
6.1 Discussion
This section contains recommendations concerning a number of-
different issues. The problem areas addressed are not specific to any
one program or modeling area but need resolution in nearly all modeling
situations.
«•
6.2 Recommendations
6.2.1 Design Concentrations
If five years of NWS data are used in an analysis or if
one year of one-site meteorological data are used, then the highest,
second-highest short term concentration estimate should be used to
determine the impact of the source. If less than five years of NWS data
or less than one year of on^site data are used, then the highest con-
centration estimate should be used as an approximation to the second-
highest short term concentration.
Block averaging times should continue to be used for
modeling purposes.
6.2.2 Critical Receptor Sites
Receptor sites should be utilized in sufficient detail
to allow estimates of the highest concentrations and the probability of
a violation of a NAAQS or a PSD increment. The procedures listed below
should be followed to locate receptor sites when a large source, such as
a 500 MW power plant, is being modeled.
a. Apply PTPLU to identify the distance to the highest
estimated concentration for each combination of atmospheric stability
class and wind speed. PTPLU should be run using 0.10, 0.15, 0.20. 0.25,
0.30, and 0.30 as wind profile exponents for the six stability classes A
through F respectively. The receptor elevation in PTPLU should be set
to the highest terrain elevation above stack base found within a one-
kilometer radius of the stack. Select the smallest of the distances
obtained from PTPLU as the first receptor distance.
b. Select eight more distances by multiplying the
first receptor distance by each of the following constants: 1.3, 1.7,
2.3, 3.0, 3.9, 5.2, 6.8, and 9.0. This geometric progression allows the
12
-------
user to most closely approximate the location of maximum concentration.
There is no need for a receptor spacing closer than 0.1 kilometer. A
tenth receptor distance may be used to locate receptors in a potentially
high concentration area beyond the ninth receptor distance.
c. Check the PTPLU output to be sure that high con-
centrations with P-G D stability are not expected beyond the last
receptor distance. If high concentrations are expected beyond the last
receptor distance, then add additional rings to include those cases.
d. If the elevation of individual receptors is signi-
ficant, those elevations should be specified as the greatest terrain
elevation along the appropriate 10 degree arc for the receptor distance
of concern; the height should not be limited to the center of the 10
degree sector. In some instances it may be desirable to locate re-
ceptors at the plant boundary. Additional rings may be needed for this.
For models capable of using a rectangular grid, includ-
ing multi-source models, a one-kilometer square receptor grid extended
outward in all directions from the source to a distance of 10 kilometers,
or about 400 receptor sites, should be used. The grid should be ex-
tended farther if maximum 1-haur concentrations are estimated to occur
beyond 10 kilometers. For urban models, this grid should cover.the
entire area being modeled. In addition, to identify concentrations that
might be missed by the spacing o.f the rectangular grid, individual
isolated receptor sites should be located downwind from the major source(s)
for prevailing wind directions during conditions of maximum concentration.
For each direction, four downwind distances associated with maximum one-
hour concentrations for Pasquill-Gifford stability Classes A, B, C, and
D as determined by PTPLU should be selected. Receptor sites should also
be located at sites where monitored air quality data are available and
sites where plume interactions from multiple sources are likely to be
greatest. If the height of individual receptors is significant, those
should be specified as the actual terrain height at the receptor loca-
tion.
For sources smaller than those equivalent to a 500 MW
power plant, receptors should be located following the above procedures,
but in the actual model runs it may not be necessary to include all
receptors for all directions and all distances. The selection of recep-
tor sites is left to the discretion of the Regional Office, but should
be based on wind roses for the area and the results of calculations
using PTPLU or other comparable screening procedures.
6.2.3 Long-Range Transport
Long-range transport should be considered where impact
on Federal Class I areas is possible. The application of simple Gaussian
models for downwind transport distances greater than 50 km should be
evaluated on a case-by-case basis. Models that are more appropriate
13
-------
at these transport distances should be evaluated a-s alternatives to the
simple Gaussian models. Models for long-range transport included in the
references to the 1978 Guideline on Air Quality Models can be used as
screening techniques. More complex and thoroughly documented models
such as MESOPUFF, MESOPLUME or MESOGRID may be considered on a case-by- '
case basis for use as refined models within their established limitations.
analyses.
6.2.4 Pollutant Half-life
Pollutant half-life should not be used in screening
For a refined analysis, if the need for half-life for
S0? can be demonstrated, site-specific data should be used to define a
rate of conversion of SO,. Otherwise only those refined models with
built-in conversion provisions should be used where conversion appears
to be an obvious problem.
For nitrogen oxides, complete conversion from NO to
nitrogen dioxide (I^) should be used in screening analyses. In refined
analyses, case-by-case half-life conversion rates should be determined
on the basis of scientific technical studies appropriate to the site in
question. The methods suggested by Cole and Summerhays* should be
considered.
An infinite half-life should be used for estimates of
total suspended particulate concentrations when simple Gaussian models
with exponential decay terms are employed. Deposition and removal
should be directly considered in the model if it is a significant
factor.
6.2.5 Urban/Rural Classification
The selection of either rural or urban dispersion
coefficients in a specific application should follow the procedure below
using land use or population density.
Land Use Procedure: (1) Classify the land use within
the total area, A0, circumscribed by a 3 km radius circle about the
source using the meteorological land use typing scheme proposed by
Auer**; (2) If land use types II, 12, Cl, R2, and R3 account for 50
percent or more of A0, use urban dispersion coefficients; otherwise, use
appropriate rural .dispersion coefficients.
* Cole, H. S., and J. E. Summerhays, A Review of Techniques for
Estimating Short Term N0? Concentrations, JAPCA, Vol. 29, No. 8,
pp. 812-817, 1979. L
**Auer, A. H., Correlation of Land Use and Cover with Meteorological
Anomalies, JAM, Vol. 17, pp 636-643, 1978.
14
-------
Population Density Procedures: (1) Compute the
average population density, p, per square kilometer with A0 as defined
above; (2) If p, is greater than 750 people/km , use urban dispersion •
coefficients; otherwise, use appropriate rural dispersion coefficients.
The land use procedure is considered the most defin-
itive. Population density should be used with caution, especially in a
highly industrialized area where the population density may be low but
the area is sufficiently built up so that the land use criteria would be
satisfied. Impacts from sources beyond the three (3) kilometers should
be included in the background.
For analyses of urban complexes, the entire area should
be modeled as an urban region if most of the sources are located in
areas classified as urban.
6.2.6 General Model Evaluation
A model evaluation study should assess how closely the
mathematical assumptions inherent to the model describe the physics
and/or chemistry of the atmosphere. The process of model evaluation
should consider all of the following: (1) assumptions inherent to
design/algorithms of model; (2) purpose/objective of model; (3) purpose/
objectives of monitors(s); (4) applicability of monitored data for
comparison; (5) comparison of model estimates with monitor observations
for upper end of frequency distribution, statistical analyses, and
analyses of weakness in individual algorithms; and "(6) analyses of
critical meteorological and source conditions and their effect on indi-
vidual algorithms. An analysis should also be made of the sensitivity
of the algorithms to the meteorological input data.
For short-term model evaluation, all input data should
be based on measured hourly averages. This includes mass emission
rates, stack dynamic operating parameters and meteorological input. The
spatial applicability of measured air quality data and tracer studies
should be consistent with the scale of the model comparisons.
Calibration for short-term air quality concentrations
is not recommended. Determination of the need for calibration and
calibration procedures are contained in the 1978 guideline.
15
-------
7.0 NONGUIDELINE MODELS
7.1 Discussion
Only a limited number of models or modeling techniques have
undergone a sufficient evaluation to be considered "guideline" models,
or to be recommended procedures for modeling certain aspects of plume
behavior. There remain a large number of circumstances when no recom-
mended technique is available and no guideline model is totally appli-
cable. There are also circumstances when a model other than a recommended
model may appear suitable. In those cases the Regional Office must
decide on the acceptable procedures and approve or disapprove specific
nonguideline modeling approaches for use in each specific situation.
7.2 Definition of Guideline vs. Nonguideline Models
Guideline models are those models specifically recommended
for (general) use in the 1978 Guideline on Air Quality Models. All
other models require review and evaluation on a case-by-case basis.
Changes made to a guideline model that do not affect the
concentration estimates do not change the guideline status of that
model. Examples of such changes are those required to run the program
on a different computer or those that affect only the format of the
model results. When such changes are made, the Regional Offices may
require a test case example to demonstrate that the concentrations
are not affected. ]
Use of a guideline model with other than recommended options
changes the status of the model to nonguideline. Similarly, if a guide-
line model has been revised or changed such that it produces concentra-
tions different from the original model for the same input data, the
status of the model is changed to nonguideline.
16
-------
7.3 Recommendations
The determination of the acceptability of a nonguideline model '
is a Regional Office responsibility. Proposed models should be eval-
uated from both a theoretical and a performance perspective. Proper
support and documentation for the use of a nonguideline model will
normally include air quality and meteorological data that have been
collected using appropriate techniques and procedures as outlined in
the "Ambient Monitoring Guideline for Prevention of Significant Dete-
rioration (PSD)" EPA 450/4-80-012, November 1980. Data bases for other
than the specific site in question may be acceptable if it can be shown
that the data available represent similar topography, climatology, and
source configurations. Any data base^used must include appropriate
periods of worst-case conditions.
An OAQPS document entitled "Interim Procedures for Evaluating
Air Quality Models" is in preparation and will describe procedures and
techniques for determining the acceptability of a nonguideline model for
use in a specific application. When completed (Summer 1981) this document
may assist in the Regional Office determinations.
17
-------
8.0 USE OF MEASURED AIR QUALITY DATA IN LIEU OF MODEL ESTIMATES
8.1 Discussion
Dispersion model estimates, especially with air monitoring
support, are the preferred basis for air quality demonstration deci-
sions. Nevertheless, there may arise instances where the performance
of recommended dispersion modeling techniques may be demonstrated by
observed air quality data to be less than acceptable. Occasionally
»-
there may be no recommended modeling procedure. In these instances,
air pollutant emission limitations may be established on the basis of
observed air quality data.
8.2 Recommendations:
Modeling is the preferred method for determining emission
limitations for both new and existing sources. Where a well-accepted,
well-verified model is available, model results alone are sufficient.
Monitoring will normally not be accepted as the sole basis..for emis-
sion limitation determination in flat terrain areas. In some instan-
ces where the modeling technique available is only a screening tech-
nique, the addition of air quality data to the analysis may lend
credence to model results.
In some instances a model that is applicable to the situa-
tion may not be available. Measured data may have to be used. Examples
of such situations are: (1) complex terrain locations, (2) aerodynamic
downwash situations, (3) land/water interface areas, and (4) urban
locations with a large fraction of particulate emissions from non-
traditional sources. However, only in the case of an existing source
would monitoring data alone be an acceptable basis for emission limits.
In addition, there are other requirements for the acceptance of an
analysis based only on monitoring:
a. A monitoring network exists for the pollutants and
averaging times of concern;
b. It can be demonstrated that the monitors in the network
are located as close as possible to all points of maximum concentra-
tion;
c. The monitoring network and the data reduction and stor-
age procedures meet all EPA monitoring and quality assurance require-
ments;
18
-------
d. The data set and the analysis identify conclusively each
individual source impact if more than one source or emission point is
involved;
e. At least one full year of valid ambient data is avail-
able and a demonstration that the year was not sufficiently atypical
to influence the resulting emission limits;
f. A demonstration that EPA recommended models are not
applicable through the comparison of the'monitored data with model
results.
Sources should obtain approval from the Regional Office for
the monitoring network prior to the start of monitoring to ensure that
the situation requires the use of monitoring and to obtain approval of
the monitoring network design and procedures.
The following are examples for some common situations where
monitored data might be considered. It should be noted, however, that
since the adequacy of a "network is a function of the source configura-
tion as well as the topography and the meteorology of the site, a large
number of designs may need-to be considered and no set pattern is appli-
cable to any one of the problem areas.
a. For aerodynamic downwash, consider one or two background
monitors plus two to four downwind monitors. The number of downwind
monitors should be determined by a consideration of the frequency of
the downwash events, the expected magnitude of the impact, and the ,
area! extent of the impact.
b. For shoreline conditions consider one to two background
monitors and three to eight downwind monitors. The number of downwind
monitors should be determined by considering site characteristics, the
magnitude and the area! extent of the predicted impact. It may be
necessary to complement the stationary monitoring network with mobile
sampling and plume tracking techniques.
c. For complex terrain, the air quality monitors should
assess the maximum impacts for each averaging period for which an air
quality violation is expected to occur. Approximately three to eight
monitors should be considered necessary to monitor for each such
averaging time. T,he exact number depends on the magnitude and extent
of expected violations. At least two monitors for each contiguous
area where violations are expected to occur is necessary except where
these areas are large. In this case, more than two monitors could be
required. As a guide, a 22-1/2° sector should define the maximum size
19
-------
of a large contiguous area. Based upon meteorological judgment,
additional monitors may be required to evaluate the source impact
depending on the complexity of the terrain.
d. For urban situations where the concern is particulates
and the sources of violations appear to be fugitive and/or reentrained
dust, extensive monitoring and receptor models may be needed to accurately
assess the problem.
20
-------
9.0 REVIEW OF PSD PERMIT APPLICATIONS
9.1 Discussion
Certain procedures with respect to the review and analysis of
PSD permits also should be standardized to ensure consistency. A few
of these are discussed below.
9.2 Recommendations
In those Regions where the Regional Office has the responsibility
for permitting new sources, the Regional Office should provide permit
applicants with a uniform PSD/NSR guidance package, including screening
and modeling requirements. The attached Air Quality Analyses Checklist
(Appendix B) is recommended as a standardized set of data and a standard
basic degree of analysis to be required of PSD and SIP revision applicants.
This checklist suggests a level of detail, including the necessary grid
resolution, required to assess both PSD increments and the NAAQS.
Special cases may require additional guidance.
A pre-application meeting between source owner and Regional
Office staff should be the norm and the Regional Meteorologist should be
represented. ......
PSD air quality analyses should be based on information
considered valid for the start-up date for the new or modified source.
The Regional Office should allow permit applicants to use
"Procedures for Evaluating Air Quality Impact of New Stationary Sources"
(EPA-450/4-77-001) for screening purposes. Air quality concentration
estimates obtained using procedures in that Guideline on screening
techniques or using the refined analytical techniques incorporating
Pasquill-Gifford or McElroy-Pooler sigmas, are equivalent to one-hour
values. Time-scaling of such estimates from any period shorter than one
hour is generally not acceptable. Time-seal ing'of one hour estimates to
longer period averages is not acceptable when the purpose is to obtain
the highest or highest, second-highest concentration estimates and a
refined analytical.technique is appropriate for making one-hour estimates.
i
Regional Offices should require permit applicants to incorporate
the pollutant contributions of all sources into their analysis. This
should include emissions associated with area growth within the area of
the new or modified source's impact. PSD air quality assessments should
consider the amount of the allowable air quality increment that has
already been granted to any other new sources. The most recent source
applicant should be allowed the prerogative to re-model the existing or
permitted sources in addition to the one currently under consideration.
21
-------
This would permit the use of newly acquired data or improved modeling
techniques if such have become available since the last source was
permitted. When remodeling, the worst case conditions used in the
previous modeling analysis must be one set of conditions modeled in the
new analysis. All sources must be modeled for each set of meteorological
conditions selected and for all receptor sites used in the previous
applications as well as new sites specific to the new source.
22
-------
APPENDIX A
ACQUISITION OF SITE SPECIFIC METEOROLOGICAL DATA
Models recommended in the 1978 Guideline on Air Quality Models .
require as input the following parameters:
0 transport wind speed and direction;
0 ambient air temperature;
0 Pasquill-Gifford stability category.
•.
Wind Measurements
In addition to 10 m surface wind measurements, the transport wind
speed and direction should be measured at an elevation as close as
possible to the effective stack height. To approximate this, if a
source has a stack (or stacks) below 100 m, select the stack top height
as a wind measurement height. For sources with stacks extending above
100 m, a 100 m tower is suggested unless the stack top is significantly
above 100 m (200 m or more). For cases with stacks 200 m or above, the
Regional Meteorologist should determine the appropriate measurement
height on a case-by-case basis. Remote sensing may be a feasible
alternative.
For routine tower measurements and surface measurement the wind
speed should be measured using an anemometer and the wind direction
measured using a horizontal vane. The specifications for wind measuring
instruments contained in the "Ambient Monitoring Guidelines for Pre-
vention of Significant Deterioration (.PSD)," EPA 450/4-80-012, November
1980 should be followed. Wind direction should be reported to the
nearest degree.
A-l
-------
Temperature
The temperature should be measured at or near standard instrument
shelter height. Ambient temperature can be reliably measured using good
quality linear thermistors or platinum resistance devices.
Stability Category
The Pasqui 1-1-Gifford (P-G) stability categories, as originally
defined, incorporate subjectively determined insolation assessments
based on hourly cloud cover observations. In lieu of such observations
it is recommended that the P-G stability category be estimated using
Table A-I. Use of this table requires the direct measurement of the
elevation angle of the vertical wind direction. Measurements of ele-
vation angle are difficult to make without a substantial commitment in
maintenance, hence it is recommended that a. be determined using the
9
transform:
a = a /u
U)
Where: a. = the standard deviation of the vertical wind direction
fluctuations averaged for a one-hour period;
a = the standard deviation of the vertical wind speed fluc-
0) r
tuations observed for a one-hour period;
u = the average horizontal wind speed for a one-hour period.
It is recommended that a vertically mounted anemometer be used to
measure the vertical wind speed fluctuations. The instrument should
meet the specifications given in the Ambient Monitoring Guidelines
referenced above. The instrument should compute a directly, one value
each, hour using 3600 to 360 values, based on a recommended readout
interval of 1 to 10 seconds.
A-2
-------
If a. is computed using the output of the anemometer by other than
9
direct application of the formula for a variance, the method should be
demonstrated to be equivalent to direct computation.
Both the vertical wind speed fluctuations and the horizontal wind
speed should be measured at the same level. Moreover, these measure-
ments should be made at a height of 10 m for valid use in estimating the
PG stability category. Trees or land use might preclude measurements as
tr
low as 10 m and in such cases the measurements will have to be made at
heights above 10 m
If on-site measurements of either a. or a are not available,
0)
stability categories may be-determined using the horizontal wind direc-
*
tion fluctuation, o_, as outlined by Mitchell and Timbre . This method
o
uses the NRC Safety Guide 1.23 categories of a. listed in Table A
y „
as an initial estimate of the P-G stability category. This relationship
is considered adequate for daytime use. During the nighttime (one hour
prior to sunset to one hour after sunrise) the adjustments given in
Table A-III should be applied. As with o,, a. should be adjusted for
-------
cover and cloud heights from some suitable NWS site (or obtained on-
site) using the CRSTER preprocessor. However, the a, categories and the
9 "
modified a. categories are anticipated to be better correlated to the
o
actual dispersion, especially in complex terrain, than employing either
estimates or measurements of insolation to estimate the P-G stability
category.
A-4
-------
Table A-I
P-G Stability Category Versus
Vertical Wind Direction Fluctuation, a.
P-G Stability
Category
A
B
C
D
E
F
*Standard Deviation of
Vertical Wind Direction, aj
2
> 12°
10° - 12°
7.8° - 10°
5° - 7.8°
2.4° - 5°
< 2.4°
From: Smith, T. B. and S. M. Howard, "Methodology for Treating
Diffusivity," in MRI-72 FR-1030 6 September 1972.
* The table values of a^ should be adjusted for surface roughness
' 02
by multiplying each of the values in the table by (z0/15cm) *
where z0 is the average surface roughness length within a 3 km
radius of the source.
A-5
-------
Table A-II
PG Stability Categories Versus
Horizontal Wind Direction Fluctuations, a
e
22
17
12
7
3
.5
.5
.5
.5
.8
>
>
>
>
>
°e
ae
ae
ae
0
a.
>
>
>
>
>
>
22
17
12
7
3
.5
.5
.5
.5
.8
PG Stability Range of Standard
Category Deviation, Degrees*
A
B
C *
D
E
F
o ~
Adapted from Nuclear Regulatory Commission CNRC) Regulatory Guide 1.23, 1972.
* The table values a. should be adjusted for surface roughness by
02
multiplying each of the values in the table by (z /15 cm) * where
z is the average surface roughness length within a 3 km radius
of the source. A personal communication with Dr. R. Londergan of
TRC indicates that in his evaluation of the use of a. over very
o
smooth surfaces (less than 10 cm) the adjustment for the roughness
length produced values in disagreement with the PGT determined
stabilities. Therefore, use of the roughness length correction
with aQ when the surface is smooth, should take this finding into
9
consideration.
A-6
-------
Table A-III
Night Time P-G Stability Categories Based on o
If the a Then the
stability And if the 10m wind speed, u, is stability class
class is m/s mi/hr is
A u<2.9 u<6.4 F
2.9
-------
APPENDIX B
AIR QUALITY ANALYSIS CHECKLIST*
1. Source location map(s) showing location with respect to:
o Urban areas**
o PSD Class I areas within 100 km
o Nonattainment areas**
o Topographic features (terrain, lakes, river valleys, etc.)**
o Other major existing sources**
o Other major sources subject to PSD requirements
o NWS meteorological observations (surface and upper air)
o On-site/local meteorological observations (surface and upper
air)
o State/local/on-site air quality monitoring locations**
o Plant layout on a topographic map covering a 1-km radius of
the source with information sufficient to determine GEP
stack heights
2. Information on urban/rural characteristics:
o Land use within 3 km of source classified according to
Auer, A. H. (1978): Correlation of land use and cover with
meteorological anomalies, J. of Applied Meteorology, Vol. 17
p. 636-643.
o Population
- total
- density
o Based on current guidance determination of whether the area
should be addressed using urban or rural modeling methodology
* The "Guidelines for Air Quality Maintenance and Analyses," Volume 10
(Revised), EPA-450/4-77-001, October 1977 (OAQPS No. 1.2-029R) should
be used a screening tool to determine whether modeling analyses are
required. Screening procedures should be refined by the user to be
site/problem specific.
**Within 50 km or distance to which source has a significant impact,
whichever is less.
B-l
-------
3. Emission inventory and operating/design parameters for major
sources within region of significant impact of proposed site (same as-
required for applicant:
o Actual and allowable annual emission rates (g/s) and opera-
ting rates*
o Maximum design load short-term emission rate (g/s)*
o Associated emissions/stack characteristics as a function of
load for maximum, average, and nominal operating conditions
if stack height is less than GEP or located in complex
terrain. Screening analyses as footnoted on Bl or detailed
analyses, if necessary, must be employed to determine the
constraining load condition (e. g., 50%, 75%, 100% load)
to be relied upon in the short-term modeling analysis.
- location (UTM's)
- height of stack (m) and grade level above MSL
- stack exit_diameter Cm)
- exit velocity (m/s)
- exit temperature (°K)
o Area source emissions (rates, size of area, height of area
source)* . -
o Location and dimensions of buildings (plant layout drawing)
- to determine GEP stack height
- to determine potential building downwash considerations
for stack heights less than GEP
o Associated parameters
- boiler size (megawatts, pounds/hr. steam, fuel consump-
tion, etc.)
- boiler parameters (% excess air, boiler type, type of
firing, etc.l
- operating conditions (pollutant content in fuel, hours of
operation, capacity factor, % load for winter, summer,
etc.)
- pollutant control equipment parameters (design efficiency,
operation record, e.g., can it be bypassed?, etc.)
o Anticipated growth changes
*Particulate emissions should be specified as a function of particulate
diameter and density ranges.
B-2
-------
4. Air quality monitoring data:
o Summary of existing observations for latest five years
(including any additional quality assured measured data
which can be obtained from any state or local agency or
company)*
o Comparison with standards
o Discussion of background due to uninventoried sources and
contributions from outside the inventoried area and descrip-
tion of the method used for determination of background
(should be consistent with the Guideline on Air Quality
Models)
5. Meteorological data:
o Five consecutive years of the most recent representative
sequential hourly National Weather Service (NWS) data, or
one or more years of hourly sequential on-site data
o Discussion of meteorological conditions observed (as applied
or modified for the site-specific area, i.e., identify
possible variations due to difference between the monitoring
site and the specific site of the source)
o Discussion of topographic/land use influences
6. Air quality modeling analyses:
o Model each individual year for which data are available with
a recommended model or model demonstrated to be acceptable
on a case-by-case basis
- urban dispersion coefficients for urban areas
- rural dispersion coefficients for rural areas
o Evaluate downwash if stack height is less than GEP
o Define worst case meteorology
o Determine background and document method
t
- long-term
- short-term
o Provide topographic map(s) of receptor network with respect
to location of all sources
*See ** on page 81 of checklist.
B-3
-------
o Follow current guidance on selection of receptor sites for
refined analyses
o Include receptor terrain heights (if applicable) used in
analyses
o Compare model estimates with measurements considering the "
upper ends of the frequency distribution
o Determine extent of significant impact—provide maps
o Define areas of maximum and highest, second-highest impacts
due to applicant source (refer to format suggested in Air
Quality Summary Tables)
- long-term
- short-term
7. Comparison with acceptable air quality levels:
o NAAQS
o PSD increments -
o Emission offset impacts if nonattainment
8. Documentation and guidelines for modeling, methodology:
o Follow guidance documents
- Guideline on Air Quality Models, EPA-45Q/2-78-027.
April 1978
- Workbook for Comparison of Air Quality Models, EPA-450/2-
78-028a.b, May 1978
- Guidelines for AQMA, Vol. 10(R), EPA-450/4-77-001, October
1977
- Technical Support Document for Determination of Good
Engineering Practice Stack Height (Draft), EPA, July 1978
- Ambient Air Monitoring Guidelines for PSD, EPA-450/2-78-
019. May 1978
- Requirements for the Preparation, Adoption and Submittal of
Implementation Plans; Approval and Promulgation of Implemen-
tation Plans, Federal Register, Volume 43, No. 118, pp 52676-
52748, August 1980.
B-4
-------
AIR QUALITY SUMMARY
For New Source Alone
** *•*
Pollutant
Highest Highest - Highest Highest Annual
2nd High 2nd High
Concentration Due to
Modeled Source (yg/m3)
Background Concentration
(yg/m3)
Total Concentration (yg/m3)
Receptor Distance (Km)
(or UTM Easting)
Receptor Direction (°)
(or UTM Northing)
Receptor Elevation (m)
Wind Speed (m/s)
Wind Direction (°)
Mixing Depth (m)
Temperature (°K)
Stability
Day/Month/Year
of Occurrence
*Use separate sheet for each pollutant (SCL, TSP, CO, NO , HC, Pb,
Hg, Asbestos, etc.) c x
**List all appropriate averaging period (1-hr, 3-hr, 8-hr, 24-hr,
30-day, 90-day, etc.) for which an air quality standard exists
Surface Air Data From Surface Station Elevation (m)
Anemometer Height Above Local Ground Level (m)
Upper Air Data From
Period of Record Analyzed
Model Used
Recommended Model
B-5
------- |