United States
Environmental Protection
Agency
Office of Air Quality
Planning and Standards
Research Triangle Park, NC 27711
EPA-454/R-93-015
April 1993
Air
» EPA
INTERAGENCY WORKGROUP ON
AIR QUALITY MODELING (IWAQM)
PHASE 1 REPORT: INTERIM
RECOMMENDATION FOR MODELING
LONG RANGE TRANSPORT AND
IMPACTS ON REGIONAL VISIBILITY
-------
EPA-454/R-93-015
INTERAGENCY WORKGROUP ON AIR
QUALITY MODELING (IWAQM)
PHASE 1 REPORT:
INTERIM RECOMMENDATION FOR
MODELING LONG RANGE TRANSPORT
AND IMPACTS ON REGIONAL
VISIBILITY
U.S. Environmental Protection Agency
Technical Support Division (MD-14)
Research Triangle Park, North Carolina 27711
National Park Service
Air Quality Division
Denver, Colorado 80225
USDA Forest Service
Office of Air Quality
Fort Collins, Colorado 80526
U.S. Fish and Wildlife Service
Air Quality Branch
Denver, Colorado 80225
April 1993
-------
NOTICE
The information in this document has been reviewed in its
entirety by the U.S. Environmental Protection Agency (EPA), and
approved for publication as an EPA document. Mention of trade
names, products, or services does not convey, and should not be
interpreted as conveying official EPA approval, endorsement, or
recommendation.
-------
FORWARD
This document is being released as a publication of the
Environmental Protection Agency (EPA) in response to a request
from members of the Interagency Workgroup on Air Quality
Modeling (IWAQM). Members include representatives from the
Environmental Protection Agency, U.S. Forest Service, National
Park Service, and U.S. Fish and Wildlife Service. The document
includes recommendations on how to estimate air quality impacts
associated with prevention of significant deterioration due to
sources farther than 50 km from a Class I area. Impacts on
visibility and other air quality related values at all downwind
distances are also addressed. IWAQM recommends that the
MESOPUFF-II model be used for these analyses in a somewhat
different mode than previously suggested by EPA.
The recommendations of IWAQM contained in this document
should be considered interim until more suitable techniques can
be developed and tested. Implementation of these
recommendations is a matter for the appropriate regulatory
agencies and should be done in consultation with the applicable
EPA Regional Office.
11
-------
PREFACE
The Interagency Workgroup on Air Quality Modeling (IWAQM)
was formed to provide a focus for development of technically
sound, regional air quality models for regulatory assessments
of pollutant source impacts on Federal Class I areas. Meetings
were held with personnel from interested Federal agencies, viz.
the Environmental Protection Agency, the U.S. Forest Service,
the National Park Service, and the U.S. Fish and Wildlife
Service. The purpose of these meetings was to review
respective regional modeling programs, to develop an
organizational framework, and to formulate reasonable
objectives and plans that could be presented to management for
support and commitment. The members prepared a memorandum of
understanding (MOU) that incorporated the goals and objectives
of the workgroup and obtained signatures of management
officials in each participating agency. Although no States are
signatories, their participation in IWAQM functions is
explicitly noted in the MOU.
This Phase 1 Report is published by the IWAQM in an effort
to provide the sponsoring agencies and other interested parties
information on appropriate "off-the-shelf" methods for
estimating long range transport impacts of air pollutants on
Federal Class I areas and impacts on regional visibility. The
IWAQM members anticipate issuing additional publications
related to progress toward meeting the IWAQM goals and
objectives, the results of model evaluation studies, proposed
and final recommendations on modeling systems for regulatory
applications, and other topics related to specific objectives
in the MOU.
111
-------
ACKNOWLEDGEMENTS
The members of the IWAQM acknowledge the special efforts
of John Vimont of the National Park Service for composing the
contents of this document and conducting most of the technical
work. The IWAQM would also like to acknowledge Alan Cimorelli
and John Irwin of the U.S. Environmental Protection Agency;
Richard Fisher of the U.S. Forest Service; Elwyn Rolofson of
the U.S. Fish and Wildlife Service; Patrick Hanrahan of the
State of Oregon, Department of Environmental Quality; Kenneth
McBee and James Browder of the Commonwealth of Virginia,
Department of Air Pollution Control; and Mark Scruggs of the
National Park Service for their input and suggestions on
assembling this document and their subsequent review. IWAQM
also thanks Brenda Cannady for her help in preparing and
proofreading the final text.
IV
-------
TABLE OF CONTENTS
EXECUTIVE SUMMARY ES-1
PHASE 1 RECOMMENDATION SUMMARY ES-5
1. INTRODUCTION 1-1
2. EXISTING MODEL COMPARISONS AND EVALUATIONS 2-1
3. CANDIDATE MODELS 3-1
3.1 MESOPUFF-II 3-1
3.2 Acid Rain Mountain Mesoscale Model (ARM3) .... 3-2
3.3 Model Comparison and Further Technical
Assessment 3-4
3.3.1 Initial Air Quality Model Comparisons . 3-4
3.3.2 Meteorological Processor Comparisons . 3-6
3.3.2.1 Wind Fields 3-7
3.3.2.2 Mixing Height 3-19
3.3.2.3 Meteorological Field Discussion 3-24
3.3.3 Further Air Quality Model Comparisons 3-26
3.3.3.1 Model Comparison Discussion . . 3-28
4. REGULATORY CONSIDERATIONS FOR THE USE OF LONG RANGE
TRANSPORT MODELS 4-1
5. INTERIM RECOMMENDATIONS 5-1
5.1 Protocol for Level I Long Range Transport
Analysis 5-2
5.1.1 Level I Long Range Transport
Techniques for Analyzing
Increment and NAAQS 5-4
5.1.2 Level I Analysis Technique for
Evaluating the Effects of Long
Range Transport and Regional
Visibility 5-4
5.1.3 Level I Analysis of Long Range
Transport and Depositional
Impacts 5-5
5.2 Protocol for Level II Long Range Transport
Analysis 5-7
5.2.1 Increment and NAAQS Cumulative Long
Range Transport Analyses .... 5-10
5.2.2 Visibility Analyses 5-12
5.2.3 Analysis of Other AORVs 5-13
5.2.4 Analysis of Primary Emissions of Fine
Particulates 5-13
6. SUMMARY 6-1
REFERENCES R-l
v
-------
IWAQM RECOMMENDATIONS FOR RUNNING THE MESOPUFF-II MODELING
SYSTEM A-l
Spatial Scale A-l
Spatial Resolution A-2
Temporal Scale A-2
Precipitation and Upper Air Meteorological Processors A-2
User Instructions - Preprocessor Programs .... A-2
READ56/READ62 Upper Air Preprocessors . . . A-2
PXTRACT Precipitation Data Extract Program . A-6
PMERGE Precipitation Data Preprocessor . . A-10
MESOPAC Input Fields A-14
MESOPUFF-II Input Fields A-19
METHOD FOR CALCULATING REGIONAL VISIBILITY IMPAIRMENT . . . B-l
VI
-------
LIST OF FIGURES
Figure 1 - Modeling domain used in comparison analysis.
Surface meteorological stations are indicated by
place name. Rawinsonde stations are indicated by
station number. Elevations contours are in meters. . 3-5
Figure 2a - Wind Vector Plot for July 21, 1984, 1200 LST.
Level 1 winds are the layer average between the
surface and the mixing depth 3-10
Figure 2b - Wind Vector Plot for July 21, 1984, 1200 LST.
Level 2 winds are between the mixing depth and 700
mb 3-11
Figure 3a - ARM3 wind field for level 1, 10 meters, for
July 21, 1984 3-15
Figure 3b - ARM3 wind field for level 3, 300 meters, for
July 21, 1984 3-16
Figure 3c - ARM3 wind field for level 6, 2400 meters, for
July 21, 1984 3-17
Figure 4a - MESOPUFF-II Mixing Heights for July 21, 1984,
1200 LST 3-21
Figure 4b - MESOPUFF-II Mixing Heights for July 21, 1984,
0000 LST 3-21
Figure 5a - ARM3 Mixing Heights for July 21, 1984 1200
LST 3-22
Figure 5b - ARM3 Mixing Heights for July 21, 1984 0000
LST 3-22
Figure 6 - Plot of hourly average mixing depths (m),
plotted every third hour 3-23
Figure B-l - Correction factor to adjust for the effects
of relative humidity on light extinction calculations
(Tang et al. , 1981) B-4
LIST OF INSETS
Inset 1 - Method for calculating concentrations of SO^ and
N03 from S02 and NOX 5-5
Inset 2 - Description of method to estimate deposition
from S02 and NOX concentrations 5-6
VII
-------
EXECUTIVE SUMMARY
The need for a consistent, technically credible approach
for evaluating the impacts of sources of air pollution located
more than 50 kilometers from Class I wilderness areas and
national parks, on those areas, has been identified. The
Interagency Workgroup on Air Quality Modeling (IWAQM),
consisting of representatives from the agencies responsible for
managing the wilderness and national park resources [the U.S.
Forest Service (USFS), the National Park Service (NFS), and the
U.S. Fish and Wildlife Service (FWS)], and the Environmental
Protection Agency (EPA), was formed to develop regional
analysis techniques to evaluate such impacts. The major charge
of the IWAQM is to develop a modeling approach for the
permitting of new and modified air pollution sources which
impact these Federal Class I areas. To this end the IWAQM has
developed a multi-year workplan (EPA, 1992) which is to be
implemented in three phases. Recognizing the immediate need
within the permitting community, the first phase of the
workplan called for an interim recommendation, by October of
1992. Given the time constraints and the practical limitations
of resources and hardware, Phase 1 was designed to provide the
best approach from existing "off-the-shelf-techniques." This
report documents the work performed and conclusions reached in
support of the Phase 1 recommendation (stated below).
Therefore, the IWAQM is proposing a technique, which will
satisfy the above listed need, to be used in the interim, until
a more refined technique is recommended in Phase 2. The
recommended Phase 1 approach is to use the Lagrangian puff
model, MESOPUFF-II (Scire et al., 1984b), to evaluate the
impacts of pollutants from sources located more than 50
kilometers from Class I areas, up to several hundred kilometers
from Class I areas. The impacts of concern are the allowable
Class I increases in pollutants (increments), the National
ES-1
-------
Ambient Air Quality Standards (NAAQS), and Air Quality Related
Values (AQRVs). AQRV impacts include such effects as
visibility degradation and acidic deposition. The recommended
modeling technique is suitable for conducting single source
impact analyses, as well as, cumulative impact analyses. The
results from this technique will frequently need to be combined
with the results from techniques used to estimate
concentrations from sources closer than 50 kilometers to a
receptor area.
It is important to note that by restricting the models
considered for Phase 1 to "off-the-shelf" techniques, the IWAQM
recognizes certain limitations in the suggested techniques.
These include limits in considering the effects of terrain on
the long range transport and dispersion, an underestimation of
the conversion of S02 to SO^ when polluted air interacts with
clouds, and an overestimation of particulate nitrate when a
limited number of sources is considered. Furthermore, the
estimations of the impacts of sources on regional visibility
are simple and do not account for all of the processes
important to regional visibility. Nonetheless, the IWAQM
considers the techniques, suggested herein, to be a significant
improvement over those previously used, in that previous
techniques ignored many of the processes important to the
assessment of air quality impacts in Class I areas. Under some
circumstances, the concentrations of sulfates in the atmosphere
may be underestimated, and hence the impacts on regional
visibility, due to the inability of the model to treat in-cloud
processes. The IWAQM, including the representatives of the
land management agencies, recognize these limitations and
consider the suggested techniques to be technically superior to
simply assuming that there are no impacts on regional
visibility. As the IWAQM work continues, these limitations
will be addressed, to the extent possible.
ES-2
-------
The IWAQM assessed two models for this recommendation, the
MESOPUFF-II model and the Acid Rain Mountain Mesoscale Model
(ARM3). It was believed that the transport and dispersion
portions of both models could be consistent with requirements
outlined in the Guideline on Air Quality Models (Revised) (EPA,
1986). Upon careful examination of both models, however,
coding errors were discovered in the ARM3, which potentially
invalidated its previous evaluations. Therefore, the MESOPUFF-
II is being recommended, since it satisfies requirements for
Class I area evaluation.
The meteorological preprocessor which is used by the
MESOPUFF-II model (MESOPAC) does not account for terrain
influences on the wind field. Also, the IWAQM has shown that
MESOPAC produces discontinuities in the mixing height field.
The IWAQM has reasoned that the possible errors introduced by
the shortcomings of MESOPAC are outweighed by the immediate
need for the Phase 1 recommendation. Therefore, the
recommendation, as stated below, is being made at this time.
However, the IWAQM has identified an existing meteorological
preprocessor which could be used, in place of MESOPAC. This
preprocessor utilizes a technique for smoothing the mixing
height fields and accounts for terrain through the use of a
diagnostic wind model. It is the IWAQM's intention to revise
the Phase 1 recommendation, to substitute this processor for
MESOPAC, as soon as it has been adequately tested within the
MESOPUFF II structure.
Recommendations for running the MESOPUFF-II model are
provided for increment, NAAQS, and AQRV analyses. Methods are
also provided for combining its results with the results from
steady-state, Gaussian plume models, which are generally used
for calculating impacts from sources closer than 50 kilometers
from receptors. A technique for evaluating regional haze
ES-3
-------
impacts from a single source or from a number of sources is
also provided.
ES-4
-------
PHASE 1 RECOMMENDATION SUMMARY
Until the Phase 2 work of the IWAQM is complete the IWAQM
recommends the following modeling approach be used under
circumstances which require the analysis of Class I area
impacts for sources more than 50 kilometers and up to several
hundred kilometers away. This recommendation is interim in
that certain technical compromises were made in order to
satisfy the immediate need for a workable modeling approach.
I. LEVEL I ANALYSIS (PLUME MODEL)
A. PSD INCREMENT AND STANDARDS
(1) For conditions other than extended
stagnation or known conditions of pollutant
recirculation, a steady-state, Gaussian
plume model may be used for all sources.
(2) Mass removal model options for either
chemical transformation or deposition
should not be employed.
(3) Where recirculation or stagnation is known
to be important the applicant should use
the Level II analysis only.
(4) If the Level I analysis indicates an
exceedance then a complete Level II
analysis should be performed.
B. VISIBILITY
The applicant should use the same approach as is
described in I.A. with the following additions:
(1) Assume that all of the emitted S02 and NOX
has been converted to SO^ and N03
respectively.
(2) The concentrations of SO^ and N03 should
then be used in conjunction with the
techniques presented in Appendix B to
estimate impacts on Class I area
visibility.
C. OTHER AQRVs
The applicant should use the same approach as is
described in I.A. with the following additions:
(1) Assume that all of the emitted S02 remains
as S02 and that the NOX has been converted
to HN03.
(2) Use appropriate deposition velocities to
estimate the deposition of the pollutants.
(See Inset 2.)
ES-5
-------
II. LEVEL II ANALYSIS (MESOPUFF-II)
A. PSD INCREMENT AND STANDARDS
(1) For sources > 50 km (and up to several
hundred km) from all Class I area receptors
MESOPUFF-II should be used.
(2) For sources < 50 km from all Class I area
receptors, models recommended for use in
the EPA Modeling Guideline should be used.
(3) For those sources located such that some
Class I receptors are < 50 km and others
are > 50 km the applicant may either
(a) model all receptors with a Guideline
model, or
(b) model those receptors which are > 50
km with MESOPUFF-II and those which
are < 50 km with a Guideline model.
(4) Concentrations from all sources should be
summed hour-by-hour, receptor-by-receptor
and pollutant-by-pollutant.
B. VISIBILITY
(1) All sources being analyzed, regardless of
their distance from the Class I area,
should be modeled with MESOPUFF-II
following the procedures set forth in
Appendix A.
(2) Using the predicted concentrations of SO^
and N03, regional haze calculations should
be made in accordance with the procedures
set forth in Appendix B.
(3) If it is determined that plume blight
analyses need to be made, the
recommendations regarding use of VISCREEN
and PLUVUE II in the Guideline on Air
Quality Models (Revised) should be
followed.
C. OTHER AQRVs (Depositional Loading)
(1) All sources being analyzed, regardless of
their distance from the Class I area,
should be modeled with MESOPUFF-II
following the procedures set forth in
Appendix A.
(2) Outputs of SO^ and N03 deposition should be
used, as necessary, to quantify the impact
to aquatic and terrestrial ecosystems.
Close coordination with the Federal Land
Manager will be necessary in determining
the appropriate averaging times for this
analysis.
ES-6
-------
III. MESOPUFF-II
The following applies to all applications of MESOPUFF
II within the context of the Phase 1 recommendation.
A. Follow the recommendations found in Appendix A.
B. The cross over distance for the time dependent
dispersion curves should be set to 10 km.
C. Both wet and dry deposition options should be
employed.
D. The model's chemical transformation algorithms
should be employed.
IV. METEOROLOGY
A. PERIOD OF RECORD (Applies to both MESOPUFF-II
and Guideline models)
(1) A five year National Weather Service (NWS)
meteorological data record should be used
when the applicant source is either > 50 km
from the Class I area or is within 50 km
and does not have at least one year of on-
site data.
(2) For an applicant source located within 50
km of a Class I area, all sources being
modeled should use a representative data
record which corresponds to the time period
of the on-site data. On-site data can not
be used unless it covers at least one full
year. Furthermore, if more than one year
of on-site data exists it should be used up
to the most recent 5 years.
B. SELECTION OF DATA BASES
(1) GUIDELINE MODEL: It may be desirable to
divide the analysis domain into
meteorologically similar areas and use area
specific representative meteorological data
to model all sources' impacts in that area.
The use of multiple meteorological data
bases is not the normal practice with
Guideline models and should be approved on
a case-by-case basis by the appropriate
regulatory authority.
(2) MESOPUFF-II: The number and location of
the NWS meteorological data bases to be
used in the MESOPUFF-II analysis should be
determined on a case-by-case basis,
generally using all available,
representative data.
ES-7
-------
1. INTRODUCTION
Under the Clean Air Act, special protection from adverse
air quality impacts is afforded certain national parks and
wilderness areas, through the Prevention of Significant
Deterioration (PSD) program. These areas have been designated
as Class I areas, and as such, increases of pollutant levels in
these areas are strictly limited. Furthermore, the Federal
Land Manager (FLM) of the Class I area is given an affirmative
responsibility to ensure that Air Quality Related Values
(AQRVs) are not adversely impacted. [The FLMs of the Class I
areas are the U.S. Forest Service (USFS), the National Park
Service (NPS), and the U.S. Fish and Wildlife Service (FWS).]
Air quality models are one of the primary tools used to assess
the impacts from sources of air pollution on both the
established PSD increments and the AQRVs. Steady-state models
are generally used for PSD analyses, however, as the PSD
program has developed, the need for more sophisticated models
to assess air quality impacts in Class I areas, from sources at
relatively greater distances from the Class I areas, has
arisen. In some areas, the FLMs have asserted that Class I
areas have been adversely affected by air pollution and that
new sources of pollution over a broad area are further harming
the resource. The absence of any recommended long range
modeling techniques has left permitting authorities without the
means to assess the assertions of the FLMs. The Environmental
Protection Agency (EPA) and the FLMs have undertaken various
model development efforts to address the air quality impacts of
pollution transported over relatively long distances. The
Interagency Workgroup for Air Quality Modeling (IWAQM) was
formed to coordinate the independent modeling efforts of the
EPA and the FLMs so that a consistent, technically credible
approach can be recommended and used.
1-1
-------
The IWAQM work plan (EPA, 1992) describes a phased
approach to satisfy the modeling needs described above.
Phase 1 consists of reviewing EPA guidance and "off-the-shelf-
technology" for recommending a modeling approach to meet the
immediate need for a regional scale model for ongoing
permitting activity. It is important to note that in order to
satisfy this immediate need, the IWAQM restricted itself to
"off-the-shelf-technology." Phase 1, described herein, is
based on current EPA guidance and existing models, which have
been further reviewed by the IWAQM. During Phase 2, the
workgroup will augment Phase 1 with a review of other available
models and make a recommendation of the most appropriate
modeling techniques. The Phase 2 recommendation will represent
a compromise between the current modeling state-of-science and
best available operational computer capabilities. The IWAQM
recognizes this later recommendation may change the initial,
first phase, interim recommendation. More advanced modeling
techniques will be considered in Phase 3.
Models used to evaluate the impact of sources of air
pollution on the PSD increments and National Ambient Air
Quality Standards (NAAQS), are required to follow the Guideline
on Air Quality Models (Revised) (EPA, 1986). (Hereafter,
referred to as the Guideline.) For many situations, preferred
models, considered generally applicable under a variety of
circumstances, are defined. When a physical situation exists
for which there are no preferred models, criteria are
established in the Guideline to use appropriate methods. These
criteria are:
1. The model can be demonstrated to be applicable
to the problem on a theoretical basis, and
2. the data bases which are necessary to perform
the analysis are available and adequate, and
1-2
-------
3a. performance evaluations of the model in similar
circumstances have shown that the model is not
biased toward underestimates, or
3b. after consultation with the EPA Regional Office,
a second model is selected as a baseline or
reference point for performance and the interim
procedures are then used to demonstrate that the
proposed model performs better than the
reference model.
One such situation is long range transport, that is,
transport of pollution beyond distances of 50 km. Therefore,
in order for any Phase 1 recommendation to be viable, from a
regulatory point of view, it will need to satisfy criteria I.,
2., and 3a. above. It is recognized that justification of an
approach under 3b. is beyond the scope of most projects.
The processes which become important in the transport of
pollution over long distances include the spatial and temporal
variability of the winds which transport and disperse air
pollutants in the presence of various terrain and water
features, the chemical transformation of the pollutants as they
travel, and the deposition of the pollutants along the way.
There are existing long range transport models available which
meet some, but not all of these needs and some which meet these
needs, but have not been sufficiently tested.
One of the primary goals of the IWAQM is to evaluate
existing modeling codes and either recommend one as an accepted
approach or combine the better elements of several of the
existing codes, creating a new modeling construct. Either of
the above approaches will require full testing and evaluation.
Creating a model with all of the desired features and testing
it and evaluating it requires time. There is, however, an
immediate need for assessing the impacts of long range
transport of pollutants into Class I areas. Therefore, the
IWAQM decided to review a limited set of existing long range
1-3
-------
transport models and recommend a specific model, which meets
the Guideline criteria, for long range transport analysis, in
the interim, until a more comprehensive solution can be
formulated and tested. In addition to the recommendations of a
specific long range transport model the Phase 1 recommendation
also specifies how this model, in conjunction with existing
regulatory models, should be used to provide those analyses
necessary for Class I PSD permitting.
By restricting the models considered for Phase 1 to "off-
the-shelf" techniques, the IWAQM recognizes certain limitations
in the suggested techniques. These include a lack of
consideration of the effects of terrain on the long range
transport and dispersion, an underestimation of the conversion
of S02 to SOI when polluted air interacts with clouds, and an
overestimation of particulate nitrate when a limited number of
sources are considered. Furthermore, the estimations of the
impacts of sources on regional visibility are simple and do not
account for all of the processes important to regional
visibility. Nonetheless, the IWAQM considers the techniques,
suggested herein, to be a significant improvement to those
previously used, in that previous techniques ignored many of
the processes important to the assessment of air quality
impacts in Class I areas. Under some circumstances, the
concentrations of sulfates in the atmosphere may be
underestimated, and hence the impacts on regional visibility,
due to the inability of the model to treat in-cloud processes.
The IWAQM, including the representatives of the land management
agencies, recognize these limitations and consider the
suggested techniques to be technically superior to simply
assuming that there are no impacts on regional visibility. As
the IWAQM work continues, these limitations will be addressed,
to the extent possible.
1-4
-------
Two models were selected for consideration for an interim
approach, the Acid Rain Mountain Mesoscale Model (ARM3) (Morris
et al., 1988) and the MESOPUFF-II model (Scire et al. , 1984b) .
The MESOPUFF-II model was considered by the EPA for inclusion
in its list of refined models in the Guideline, but was
subsequently suggested for inclusion only in Appendix B of the
Guideline, the section reserved for models which could be
considered for regulatory use, but not generically preferred.
The NPS has been evaluating the ARM3 for use in its program of
evaluating the impacts of air pollution in the national parks.
As part of this evaluation, they chose to compare some of the
ARM3 results against MESOPUFF-II because of MESOPUFF-II's
availability and its consideration by the EPA. These two
models both contain features considered desirable in a model
for use in long range transport to Class I areas, particularly
the ability to consider the chemical transformation of S02 and
NOX to SO^ and N03 and the removal of chemical species through
deposition. In addition, the ARM3 contains algorithms for
considering the effects of terrain on dispersion and on the
transport flow. Furthermore, both of these models have been
compared against other, similar models and have performed
somewhat better than those other models relative to measured
tracer data (Carhart et al., 1989; Moore et al., 1990).
The IWAQM recognizes that there are certain risks involved
with recommending an interim long range modeling approach.
From a regulatory perspective, it is generally desirable to use
an interim model which will yield somewhat higher impact
calculations than a more refined, preferred approach. In the
case of steady-state air quality models for example, this can
be relatively easily ensured because of the independence of the
concentration calculations from one hour to the next. In the
case of the Lagrangian long range transport models under
consideration here, the concentration calculations for a given
hour will be explicitly dependent on the spatially and
1-5
-------
temporally varying wind field from that hour and previous
hours. Therefore, the exact behavior of a given modeling
system relative to a similar but different modeling system can
not be predicted with certainty.
1-6
-------
2 . EXISTING MODEL COMPARISONS AND
EVALUATIONS
There have been a number of surveys of long range
transport models which may be suitable for estimating the
concentrations of pollutants which degrade visibility and/or
contribute to acid deposition. One such survey (Thompson et
al., 1987) used a series of screenings and rankings to narrow
the field of Lagrangian, Eulerian, hybrid, and statistical
models which they would consider for application in western
Canada. The model characteristics this study considered
important were:
a. domain - 0 to 500 km and up to one year
b. resolution - 1 to 10 km and event to seasonal
c. predictands - ambient air concentrations of S02, SO^,
HF, metals, oxidants and NOX
d. processes - convective and frontal storms, flow in
complex terrain, rain and snow scavenging, influences
of soil particles, cloud physics and chemistry
e. design - modular
f. accuracy - ±30 percent for sulfate concentration and
deposition
g. chemistry - nonlinear.
The desire for a model which exhibits these characteristics is
also shared by the IWAQM. A relatively small number of the
potentially available models were identified through this
process. The MESOPUFF-II model was among the models identified
as meeting the criteria of the survey.
Another review of models was conducted for the EPA as part
of the Rocky Mountain Acid Deposition Model Assessment (Morris
and Kessler, 1987). The conclusion of this study was that
"...no one meteorological or acid deposition model is
significantly superior to the others; all the candidate models
2-1
-------
contained different features that would be desirable attributes
in an acid deposition model for the Rocky Mountain region.
Hence, the conceptual design of the mesoscale acid deposition
model uses modules selected from various existing
meteorological and acid deposition models." This ultimately
lead to the development of the Acid Rain Mountain Mesoscale
Model (ARM3).
While there have been a number of reviews and surveys of
models and modeling features which could potentially address
long range transport and visibility and acid deposition
effects, there have been relatively few model evaluation
efforts against field data. One such effort examined eight
short-term, long-range transport models (Carhart et al., 1989).
The models were tested against two tracer data bases. One of
the data bases was collected in Oklahoma from perfluorocarbon
tracer releases upwind of sampling arcs placed at 100 and 600
km. The second data base was collected at the Savanna River
Plant from the release of Kr85 gas from a 62 m stack. The
samplers from this experiment were at distances from 28-144 km
downwind. The main method used in the evaluation of the
performance of the models was the application of the American
Meteorological Society (AMS) Statistics. Additional statistics
were added to the AMS recommended list in order to assist in
interpreting the results. In addition, graphical analyses were
used to supplement the statistical comparisons in order to shed
light on the causes of model performance trends identified by
the statistics. The data bases both involve an inert tracer,
therefore, the evaluations only deal with the transport and
dispersion algorithms of the models and not the deposition or
chemical conversion algorithms. The field experiments were not
designed to evaluate dispersion in complex terrain. Some of
the important conclusions of this study were:
2-2
-------
The causes of model/data discrepancies can be largely
traced to inadequate wind field modeling that leads
to an incorrect temporal and spatial positioning of
the plume, and the use of the Turner curves to
downwind distances beyond which they can accurately
represent the scales of atmospheric turbulence. The
use of multilayer wind field models and the use of
the Heffter formula for lateral plume dispersion
close to the source appear to improve model
accuracies.
The above model evaluation study was being conducted about
the time that the Acid Rain Mountain Mesoscale Model (ARM3) was
being completed. As the final portion of the Rocky Mountain
Acid Deposition Model Assessment Project, the ARM3 was
evaluated against the same data bases, using the same
statistics. The results of that evaluation were that the
overall performance of the ARM3 was similar to that of
MESOPUFF-II (Moore et al., 1990). Again, it should be noted
that the data bases used in these analyses were not designed to
stress the models' ability to simulate transport and dispersion
in complex terrain.
The MESOPUFF-II model was also evaluated against data
collected during the Cross-Appalachian Tracer Experiment
(CAPTEX) (Godowitch, 1989). This experiment did include some
transport and dispersion over complex terrain. This study
concluded that any bias in the model estimates was toward over-
prediction of measured concentrations.
2-3
-------
3 . CANDIDATE MODELS
For the interim recommendation, the IWAQM only considered
models which could meet the Guideline criteria for the use of
alternative models, described above. Due to the results of the
above evaluations, the model features, the availability of the
models, and the relative familiarity of the MESOPUFF-II model
and the ARM3 to the IWAQM, these models seemed the logical
choice to consider for this interim recommendation. The IWAQM
considered that either model was applicable on a theoretical
basis, that the available evaluation data bases, referred to
above, are adequate, although not ideal for the purposes cited
herein, and that the evaluations of the models indicated that
there was not a systematic bias toward underestimation. While
these models meet the Guideline criteria, the IWAQM recognizes
that there are potentially other models which might be better
suited for a particular application, but for general long range
transport modeling, the aforementioned models should be
adequate.
3.1 MESOPUFF-II
The following is an excerpt from the abstract of the
Development of the MESOPUFF-II Dispersion Model, (Scire et al.,
1984a), which provides a good summary of the nature of and
features of the MESOPUFF-II model:
...MESOPUFF-II is a Lagrangian variable-trajectory
puff superposition model suitable for modeling the
transport, diffusion and removal of air pollutants
from multiple point and area sources at transport
distances beyond the range of conventional straight-
line Gaussian plume models (i.e., beyond ~ 10-50 km)
It is an extensively modified version of the
MESOscale PUFF (MESOPUFF) model (Benkley and Bass,
1979). Major additions and enhancements include:
use of hourly surface meteorological data and twice-
3-1
-------
daily rawinsonde data; separate wind fields to represent
flow within and above the boundary layer; parameterization
of vertical dispersion in terms of micrometeorological
turbulence variables; parameterization of S02, to SO^ and
NOX to N03 conversion, including the chemical equilibrium
of the HN03/NH3/NH4N03 system; resistance modeling of dry
deposition, including options for source or surface
depletion; time- and space-varying wet removal; and a
computationally efficient puff sampling function...
One of the limitations of the model, with respect to the
calculation of pollutant concentrations in Class I areas, which
are frequently located in complex terrain areas, is the absence
of any complex terrain treatment either on the generation of
the meteorological fields or on the dispersion. The
shortcoming of the meteorological fields is overcome to the
extent that the meteorological observations, which are used to
generate the wind fields in the MESOPAC meteorological
processor, represent the influence of terrain. The lack of
influence of complex terrain on the dispersion is somewhat
obviated by the fact that at the downwind distances of the
receptors from the sources, envisioned by the use of this
model, the puff will generally be uniformly mixed throughout
the depth of the mixed layer. Therefore, in most applications,
it is not expected that these shortcomings will overwhelmingly
bias the results of this model.
3.2 Acid Rain Mountain Mesoscale Model (ARM3)
The following brief description of the ARM3 is taken from
the preface to the ARM3 users guide (Morris et al., 1988):
...The ARM3 model is a Lagrangian trajectory model
with simplified chemistry applied to the discrete
plume parcels...
The ARM3 model consists of mesoscale
meteorological modules and acid deposition/air
3-2
-------
quality modules applied to the plume parcels transported
with the winds in the Lagrangian frame...
A three-dimensional diagnostic wind model is
used to calculate the spatially and temporally
varying wind fields. Kinematic, blocking and
deflection, and thermodynamic effects are accounted
for through simple parameterizations. The wind model
is designed to generate wind fields within regions
with sparse data; thus, the validity of the wind
field is highly dependent on the quality of the
observations and their applicability to the
interpolation applied between the observations. Each
interpolation of the observed temperature, dew point,
and precipitation amounts contains an orographic
adjustment based on limited climatological data from
the Rocky Mountain region... Mixing height,
stability classification, friction velocity,
convective velocity, Monin-Obukhov length, and
surface pressure are all estimated at each grid cell
using appropriate algorithms with interpolated
observations...
The acid deposition/air quality modules treat
the plume parcels along their trajectories. The
height of the parcel can be set either as terrain-
following or reduced relative to the difference
between the elevation of the terrain at the parcel
location and elevation at the stack base...
There are three options for determining
dispersion rates. The use of the Pasquill-Gifford
dispersion rates provides the minimum dispersion.
The other options provide higher dispersion rates
that may be appropriate over regions of complex
terrain. The dry deposition algorithm is based on
the resistance approach. A dry deposition velocity
is calculated based on the land-use type at the plume
parcel location. The algorithm in the ARM3 is
comparable to those in other models that use this
approach. The wet deposition algorithm uses the
scavenging coefficient approach. The precipitation
rate for the grid cell containing the centroid of the
Lagrangian parcel is used...
Chemical transformation of S02 and N02 to sulfate
and nitrate can be calculated in the ARM3 using one
of two highly-parameterized options in the ARM3
model. They are the methods adopted from the RIVAD
[Regional Impact in Visibility and Acid Deposition
Model] and the MESOPUFF-II models...
3-3
-------
To treat the aqueous-phase formation of sulfate,
both mechanisms assume a linear oxidation rate that
depends only on the S02 concentrations. RIVAD uses a
constant rate of 0.2 percent/hr MESOPUFF uses a rate
that ranges from 0.2 to 3 percent/hr depending on the
relative humidity...
The ARM3 is of a similar nature to the MESOPUFF-II model,
except that it has algorithms which explicitly treat complex
terrain. These include a diagnostic wind model which treats
kinematic and blocking effects of terrain on the air flow,
dispersion parameters for complex terrain, and a correction for
plume height as a plume passes over terrain. These
enhancements should, ostensibly, make it more suitable for
calculations in complex terrain.
3.3 Model Comparison and Further Technical Assessment
Rather than proceed with recommending a model strictly
based upon its reported technical merits, a series of
comparison runs were conducted to test the manner in which the
models function under varying input conditions. The workgroup
also examined the results of the meteorological processors of
the two models under consideration, to appraise the credibility
of the fields produced.
3.3.1 Initial Air Quality Model Comparisons
The first step undertaken was to run the ARM3 and
MESOPUFF-II for a hypothetical point source located in south-
central Virginia and calculate the concentrations of pollutants
which might reach Shenandoah National Park. These analyses
used available National Weather Service data for July 1984
(Figure 1). The results of these analyses were unexpected.
3-4
-------
725
X
5£
AAAf\
ftet*t\J
4400
4360
4320
4280
4240
A or\rt
4200
4160
4120
4080
Af\AC\
*rU*rU
5£
20
Modeling Domain
10 620 650 680 710 740 770 800 830 860 890 920 950 980
j i i j'i; ; i i : '/'V : ' ' ' '/ ' r ' rrt i i i,t-J ' L
L. ,' /-, 1 * '\'C'\ ' '' --' 1
1 »f -' '. 1 , ' '" *'%'l\\l'' ,' 1
- / \ to" ' 4§i '' £*'' i ' ',' ' /'*
-- t i 'v* ^ " ** ^ v^'/ ' ' ^
*' ' * *f ' '* ' * \ ' f' /' C^
-> ' \' ! '"'"''I i"' ' N
i i 'i .,' ' : ' '
- 1 I1*.-"1 s .' '.| ~
-'» X '* Lj' ,"l II,"" -
-/ /' jS1 ' ' / ' ,'/ 72403
^-.P* $::'';?*&* * -
I ri;;^' ^' # ^
:\/X'^''};' /^^'/ :
-"V%^--'v^_ /' -
~/>x\'"'\/*i/ ~
»/'''»/'' 'i *"*** ''/' -
~/\/*x**^ ""
i ,'' ,y i
'' .''
I X ""
L « I
f X
^ /
S> *
^ S "
~i i Ti i i i i i i i i i i i i i i i i i i i i i i i i i i i i i i i i i i i r
10 620 650 680 710 740 770 800 830 860 890 920 950 980
72317
X
7465
AAAf\
*fr*fr*frU
44.no 72407
"T*TUV/
X
4360
4320
4280
4240
'2402
A *)f\f\
4200
4160
4120
4080
4O4O
*rU*rU
}5
Figure 1 - Modeling domain used in comparison analysis.
Surface meteorological stations are indicated by place name.
Rawinsonde stations are indicated by station number.
Elevations contours are in meters.
The concentrations of pollution calculated by the ARM3
were approximately an order of magnitude higher than those
calculated using MESOPUFF-II. Since this result was not
expected, the IWAQM undertook a series of model runs to
determine whether some of the fundamental differences in the
3-5
-------
air quality model formulations were responsible for the
dramatic concentration differences or whether differences in
the generation of the meteorological fields were responsible.
First, both models were run in an inert mode; that is, the
options for calculation of chemical transformation and
deposition were turned off. The dissimilarities in results,
were again, essentially the same. It was considered that
differences in the results stemmed from the treatment of
complex terrain in the ARM3, either in the wind fields or the
plume dispersion and transport algorithms. Complex terrain
potentially has two effects when considering concentration
calculations from the ARM3 air quality model. First, under the
options selected for this series of tests of ARM3, the
dispersion of pollutants is enhanced by the effects of the
complex terrain. The second effect was the influence terrain
has on bringing the receptor closer to the plume elevation.
The first effect would have a tendency to lower the
concentration estimates, while the latter could potentially
increase the concentration estimates. Therefore, it was
decided to run the ARM3 without the plume height to receptor
correction included on the original runs. The removal of this
option had little effect on the concentrations calculated by
the model; this was not the expected result. Furthermore,
selecting the option within the ARM3 to use the MESOPUFF-II
dispersion parameters did not bring the modeled concentrations
appreciably closer.
3.3.2 Meteorological Processor Comparisons
Since different options in the air quality models, which
should force them to be nearly the same, could not account for
the discrepancies in the concentrations calculated in the
initial runs, the meteorological fields generated by the
models' respective processors were examined. Each model treats
3-6
-------
the meteorological inputs somewhat differently. The MESOPUFF-
II uses a two layer representation of the mesoscale winds. The
lower layer (Level 1) is an average wind defined between the
surface and the mixing depth, while the upper level wind (Level
2) is an average between the mixing depth and an arbitrarily
defined upper bound, usually 700 mb. The ARM3 allows the
selection of the number of layers to represent the winds. In
the test cases run, pursuant to this discussion, six vertical
layers were chosen. An average wind for each of these layers
is calculated by the ARM3 meteorological processor. The
methods used to generate the mixing heights in the two models
are somewhat different. This will be discussed further below.
3.3.2.1 Wind Fields: The wind fields generated by the
MESOPUFF-II processor are spatial and temporal interpolations
of the surface and upper air observations. The method for
calculating the mixed layer wind at each point follows (Scire
et al., 1984a):
A representative rawinsonde sounding (00 or 12
GMT) is selected based upon the stability class
at the nearest surface station to the grid point
and the time of day. Neutral/unstable and
stable conditions are assumed to be represented
by the 00 GMT and 12 GMT sounding, respectively.
Using the sounding selected in Step (1),
vertically averaged u (easterly) and v
(northerly) wind components are computed through
the layer from the surface to the grid point
mixing height.
The ratio, R, of the layer-averaged wind speed
to the surface wind speed at the rawinsonde
station, and the angular difference in the wind
direction, A©, between the layer averaged and
surface winds are calculated.
3-7
-------
(4) The hourly surface wind data are used to
calculate spatially interpolated surface wind
components (us,vs) at each grid point. Data from
all surface stations within a user-specified
'scan-radius' of a grid point are used to
compute (us,vs) according to
where:
us,vs are the easterly and northerly
components of the surface wind at grid
point (i,j),
a£
r"
k
uk,vk are the easterly and northerly
components of the surface wind at
surface station k,
rs is the distance from the surface station to
grid point (i,j), and
as is an alignment weighting factor
(as = 1-0.5 | sin cf>s | , where cf>s is the angle
between the observed wind direction and the
line from the surface station to the grid
point).
For equal values of rs, alignment weighting
causes winds at a station directly upwind or
downwind of a grid point to be weighted twice as
heavily as the winds for a station at right
angles to the grid point.
The mixed layer averaged wind at the grid point
is calculated by multiplying the surface wind
speed at the grid point from Step (4) by the
wind speed ratio, R, at the nearest rawinsonde
site. Similarly, the surface wind direction is
adjusted by the wind direction factor, AO.
Vertically averaged winds from the mixing height
to the 850 mb, 700 mb, or 500 mb levels are
computed in the following manner. The 00 GMT
and 12 GMT winds at each rawinsonde station are
3-8
-------
first interpolated in time, and then vertically
averaged through the layer from the grid point
mixing height to the selected level (e.g., 700
mb). The winds at grid point (i,j) are obtained
from the previous equation, with the summation
over rawinsonde stations instead of surface
stations. Only rawinsonde stations within a
'scan-radius' of the grid point are considered.
3-9
-------
Allen/Beth
72520
Pittsburgh
Mesopuff Level 1
4440
4400
4360
4320
4280
590 620 650 680 710 740 770 800
72407
4080
4040
590 620 650 680 710 740 770 800 830 860 890 920 950 980
Greensboro
80
'040
Rol/Durhom
74695
Figure 2a - Wind Vector Plot for July 21, 1984, 1200 LST.
Level 1 winds are the layer average between the surface and the
mixing depth.
3-10
-------
Figures 2a and 2b are examples of the wind fields generated by
the MESOPUFF processor for July 21, 1984 at 1200 LST. The
effects of the 'scan-radius' and the influence of a deviant
surface station on the calculation of mixed layer average
(Level 1) winds can be seen in the vicinity of Richmond. There
are also some aberrant winds generated in the vicinity of
72520
Pittsburgh
Allen/Beth
Mesopuff Level 2
4440
4400 -
590 620 650 680 710 740
4080
4040
590 620 650 680 710 740 770 800 830 8!
72317
Greensboro
*
Rol/Durhom
4040
74695
Figure 2b - Wind Vector Plot for July 21, 1984, 1200 LST.
Level 2 winds are between the mixing depth and 700 mb.
3-11
-------
Philadelphia. It appears, however, that the model is
generating a sea breeze along the coastal areas. The Level 2
winds are generally fairly smooth and uniform.
The ARM3 takes a somewhat different approach to
calculating the wind field for the region of interest. The
ARM3 meteorological processor was designed to account for the
influence of terrain on the wind fields over a data sparse
area. The ARM3 processor computes a three dimensional wind
field. The ARM3 air quality model can use the vertical
velocities, generated by the meteorological processor, to
transport puffs. However, vertical velocities generated by
diagnostic wind models over complex terrain are highly suspect
and are not recommended for use in the air quality model.
Therefore, the remaining discussion will only pertain to the
formulation of the horizontal components of the wind, although
some of the procedures for calculating the vertical component
can influence the calculation of the horizontal components.
The ARM3 processor uses "domain mean" parameters to
initialize the entire modeling domain, then computes the
effects of terrain on that flow and finally incorporates the
observations into the terrain modified flow in the vicinity of
the observations. The rationale for this is the supposition
that the domain mean winds, modified for the terrain effects,
better represent the flow in data sparse areas than relatively
distant observations. In the ARM3 processor, the domain mean
wind is defined from the sounding at the "central-most"
station. The domain mean horizontal flow is modified to
account for the effects of slope flows and blocking effects.
The observations are incorporated into the modified mean flow
through a user-specified weighting factor, such that grid cells
near an observation give the observation relatively greater
weight and cells more distant from any observations are
weighted more heavily toward the modified mean flow.
3-12
-------
A slope flow vector is calculated for each grid cell and
added to the gridded domain mean (i.e., horizontally uniform)
wind field. The slope flow parameterization does not account
for any nonlinear interaction of slope flow with ambient flow.
For any slope angle, a, the speed, S, of the parameterized
slope flow is defined as:
where:
S0\ I = The slope flow amplitude based on
the domain-scale temperature lapse rate
f1(t) = ±Idepending on the time of day
(-1 for downs lope, +1 for upslope)
f2(oc) = variability of the slope speed with slope angle
Blocking effects are calculated from the gridded
horizontal wind field, the available atmospheric stability
information, and the gridded terrain heights. A local Froude
number is calculated at each grid point. The Froude number is
defined as:
Fr =
(N&h)
where:
S = the grid-point wind speed
N = the Brunt-Vaisala frequency, defined as-.
\\ where g= acceleration of gravity &
. . e/V dz/
}=potential temperature
= the effective terrain height at the grid-point
If Fr is less than a critical Froude number, Frc, usually
equal to 1, and the horizontal wind at a given grid point has
an uphill component, the horizontal wind is adjusted so that
3-13
-------
flow is in a terrain-tangent direction with no change in speed.
If Fr is greater than Frc, then flow is not adjusted.
After the domain mean flow is adjusted for both the
kinematic effects of terrain and thermodynamic blocking
effects, the available observational data are combined with
this field to produce a final gridded wind field. This
involves interpolation, smoothing of the analyzed field,
computation of a vertical velocity field and a minimization of
the three-dimensional divergence. The discussion here will
only focus on the interpolation of the observational data into
the modified domain mean wind field, since this feature will
most affect future considerations of the ARM3 wind fields.
The procedure for interpolating both the surface and upper
air data is a modified inverse weighting scheme. The
interpolation is carried out separately for each model level.
Unless otherwise specified, all surface wind observations are
incorporated into the lowest model level. Upper-air
observations are first vertically and temporally interpolated
to model levels and desired simulation times. The terrain
adjusted, domain mean horizontal components of the wind at each
grid point, (u,v)i; are modified to yield the observationally
interpolated wind, (u,v)' as follows:
where .
rk = horizontal distance station k to grid-point
R = weighting factor for the diagnostic wind field
i
n controls the relative influence of the observations
vj . = the observed wind at station k
0 K
This procedure weights the step 1 wind field, (u,v)i;
heavily in regions far removed from observations. The degree
3-14
-------
of influence exerted by the step 1 wind field is inversely
related to the value of parameter R^. The exponent, n,
controls the relative influence of observations distant from a
given grid point.
72520
ARM3 Level 1
590 620 650 680 710 740 770 800 830 860 890 920 950 980
4440 rrr
4400
rrrrivi i 444°
' 4400 72407
X
4360
4320
4280
_, 4240
^2402
j 4200
4160
4120
4080 -
-! 4080
4040
i I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I l~l
4040
590 620 650 680 710 740 770 800 830 860 890 920 950 980
72317
74695
Figure 3a - ARM3 wind field for level 1, 10 meters, for July
21, 1984.
3-15
-------
The wind fields generated by the ARM3 meteorological
Allen/Beth
72520
Pittsburgh
ARM3 Level 3
590 620 650 680 710 740 770 800 830 860
-I'fjTfffff-fffff
4080
4040
590 620 650 680 710 740 770 800 830 860 890 920 950 980
4040
72317
Greensboro
*
Rol/Durhom
74695
72407
Figure 3b - ARM3 wind field for level 3, 300 meters, for July
21, 1984.
3-16
-------
Allen/Beth
72520
Pittsburgh
ARM3 Level 6
590 620 650 680 710-^746
4080
4040
590 620 650 680 710 740 770 800 830 860 890 920 950 980
4040
72317
Greensboro
*
Pol/Durham
74695
Figure 3c - ARM3 wind field for level 6, 2400 meters, for July
21, 1984.
processor for July 21, 1984, at 1200 LST (Figures 3a, 3b, and
3c) exhibit markedly different features than those generated by
the MESOPUFF-II processor. First, it must be noted that the
MESOPUFF-II winds are mixed layer averaged winds, whereas the
ARM3 generated winds represent thinner layers of fixed
thickness, above ground level. Although the vector fields of
3-17
-------
Figure 2 represent somewhat different layers than those in
Figure 3, some similarities might be expected. First the 10
meter field of Level 1 exhibits a relatively slow flow field.
The influence of the interpolation scheme around surface
stations is quite evident, particularly in the vicinity of
Richmond, where the vectors are at about 90° from the mean flow
around Richmond. Other than some of the local station
influences, there is little resemblance between the ARM3 and
MESOPUFF-II Level 1 wind fields. The 300 meter winds generated
by the ARM3 meteorological processor (Figure 3b) might be
expected to be more similar to the MESOPUFF-II Level 1 field
(Figure 2a), since it represents a mixed layer average and
since the ARM3 300 meter winds should be relatively
representative of the mixed layer around 1200 LST; they are,
however, quite disparate. Also, 300 meters might be the
expected plume height of a relatively large source with a
moderately tall stack. The MESOPUFF-II winds reach speeds of
18.9 m/s in the northeast corner of the domain. The ARM3
generated winds only reach 9.4 m/s. Both models have an area
of stronger winds along the coastal area, but the wind
directions are shifted by approximately 90°. The inland winds
of the ARM3 Level 3 are relatively uniform from the NE (see
Figure 3b), with the exception of the southwest corner of the
modeling domain where they are light from the SW. The
MESOPUFF-II winds are much more variable. Again the wind
directions are shifted by about 90°. The MESOPUFF-II winds
above the mixed layer and the ARM3 2400 meter winds are very
similar. The wind directions are about the same and the
overall magnitude of the winds is about the same.
As previously noted, the two models handle the generation
of winds above the surface somewhat differently. The MESOPUFF-
II approach uses the deviation between the surface and upper
air wind speeds and directions at the time of the soundings and
the current hour's surface data to calculate the wind speeds
3-18
-------
and directions within the surface layer. Whereas, the method
used by the ARM3, above the first layer, is to use spatially
and temporally interpolated upper air data to perform its
calculations. Furthermore, the ARM3 interpolations of data are
not performed until after a uniform "first guess" wind field is
modified for the effects of terrain and the modified first
guess field is given higher priority in the interpolations in
areas more removed from observations. Therefore, the general
directional features of the flow fields between the respective
models' lowest levels are similar, since both make use of the
surface station data. However, with the six layer
representation of the atmosphere used in ARM3, the winds at
levels above the first level, and below the height of the
mixing depth, are quite different than those calculated by
MESOPUFF-II. The method for calculating winds above the mixed
layer, however, is similar between the two models in that both
use only spatially and temporally interpolated winds for their
respective calculations. The ARM3 still uses the modified
first guess field, but at higher levels there is generally much
more uniformity to the overall flow field.
3.3.2.2 Mixing Height: Mixing heights are another parameter
which could potentially result in dramatically different
concentrations calculated by the two air quality models.
MESOPUFF-II and the ARM3 both calculate a mechanical mixing
depth for the nighttime hours and a mechanical and convective
mixing height during the daytime and use the greater of the two
as the mixing depth. The two models use similar, but somewhat
different, algorithms to calculate the mixing depth, which
yields different results.
The convective mixing depth algorithm in the MESOPUFF-II
meteorological processor assumes that during daylight hours,
solar radiation reaching the ground produces an upward flux of
sensible heat and the development of a well-mixed adiabatic
3-19
-------
layer. If the hourly sensible heat, H, is known, the mixed
height z± at time t+1 can be estimated from time t in a
stepwise manner.
,2 2H(l+B)At 2(A6)t(z.)t
(A6)t+1 =
2i|/1E'HAt
pc
p I
where:
il^ = the potential temperature lapse rate
in the layer above z±
At = the time step (3600 s)
E = a constant (-0.15)
A6 = the temperature discontinuity at the
top of the mixed layer
The sounding at the nearest rawinsonde station to the grid cell
is used to determine the lapse rate i\i1.
The daytime mechanical mixing depth is calculated from:
Bu
Zi =
(fNB) 2
where:
f = the Coriolis parameter
B = a constant (^2)
u+ = the friction velocity
NB = the Brunt-Vaisala frequency in the stable
layer aloft
The nighttime mechanical mixed layer is determined from:
= 2400u,
3-20
-------
Examples of a daytime and nighttime MESOPUFF-II generated
mixing depth field are shown in Figures 4a and 4b. The daytime
mixing depths generated by MESOPUFF-II exhibit some extreme
discontinuities from one grid cell to the next. This is a
result of using only the nearest rawinsonde station sounding to
calculate the convective mixing depth, rather than an
interpolated field. The nighttime mixing depths make a much
smoother transition from one cell to the next. The daytime
mixed depth values range from 552 meters to 1759 meters,
whereas the nighttime mixed depth values range from 10 meters
to 1360 meters.
Figure 4a - MESOPUFF-II Mixing
Heights for July 21, 1984,
1200 LST
Figure 4b - MESOPUFF-II
Mixing Heights for July 21,
1984, 0000 LST
In the ARM3 system, the daytime convective mixed depths
are determined as the height of the intersection of the hourly
surface potential temperature and the morning potential
temperature sounding. Cold or warm air advection is accounted
for by adjusting the hourly surface potential temperature
values according to an advection rate. The advection rate is
determined from the difference in potential temperature between
3-21
-------
the afternoon and morning sounding at a height above the
convective mixing height.
The ARM3 mechanical mixing depth is the same for both
daytime and nighttime conditions.
mech Free Stream
where: U,
Free Stream
at 3000 meters
is taken to be the wind speed
Figure 5a - ARM3 Mixing
Heights for July 21, 1984 1200
LST.
Figure 5b - ARM3 Mixing
Heights for July 21, 1984
0000 LST.
The ARM3 generated fields (Figures 5a and 5b) show
relatively smooth fields for both daytime and nighttime. The
actual heights, however, are much lower during the daytime than
those calculated for MESOPUFF-II, ranging from 125 meters to
596 meters. During the nighttime, the values range from 185 to
593. So while the ARM3 mixing height fields do not exhibit the
discontinuities from one cell to the next that the MESOPUFF-II
3-22
-------
daytime fields show, they do not show any appreciable diurnal
variation.
To further illustrate this, domain average mixing heights
were calculated for each model and plotted (Figure 6). The
mesopuff
Figure 6 - Plot of hourly average mixing depths (m), plotted
every third hour.
MESOPUFF-II heights show diurnal variation over the entire
time-span plotted, while the ARM3 mixing heights show almost no
relationship to the diurnal cycle. This can be partially
explained by the method used to calculate the mechanical mixing
depth. The choice of the 3000 meter wind speed as an indicator
of the free stream wind will almost always force the wind to be
at least on the order of 5 m/s. If the term U in the ARM3
mechanical mixing equation is set to 5, then the minimum mixing
3-23
-------
depths will generally be on the order of 265. In order for
reasonable nighttime mixing heights to be calculated, the 3000
meter wind would have to be more on the order of 1 m/s. At
that altitude, however, a wind speed of 1 m/s is not likely to
occur very often. The daytime mixing depth calculations,
generated by the ARM3, are generally too low. The reason for
this has not been investigated at this time.
3.3.2.3 Meteorological Field Discussion: The wind fields
generated by the two models are each based on distinctly
different approaches to interpolating surface and upper air
observations. Both methods have some significant problems.
For the stated purpose of IWAQM, the MESOPUFF-II approach lacks
the ability to treat the effects terrain will have on the mean
flow, except as much as local observations represent the mean
flow in the terrain. The mixed depth average wind speed and
direction used by MESOPUFF-II can be both a strength and a
weakness. The strength of the system is that it provides some
other information to the interpolation scheme between 12 hour
soundings by using the relationship between the surface and
upper air data at the time of the twelve hour soundings. One
problem with this, however, is that it is based upon the
assumption that the surface and upper air data are indeed
coupled. It is quite possible that under some circumstances,
particularly in complex terrain, that the surface and upper air
winds are independent from each other. Hence, the use of the
relationship between the surface and upper air winds may yield
spurious results. Another factor to consider with respect to
the mixed depth average winds is the behavior of the mixed
depth, as illustrated in Figures 4a and 4b. In the northwest
corner of the 1200 LST MESOPUFF-II mixing height field (Figure
4a), the mixing depth jumps from a height of around 500 meters
to approximately 1600 meters in adjoining grid cells. Thus,
the mixed layer average wind will represent very different
quantities between those adjoining cells.
3-24
-------
The ARM3 wind fields ostensibly meet the IWAQM criteria of
accounting for the effects of terrain on the wind fields. The
method the ARM3 uses to account for terrain effects warrants
some discussion. The wind field generating portion of the
model is called the Diagnostic Wind Model (DWM). The order of
analysis that the DWM uses is to first generate a horizontally
uniform "first-guess wind field." This first-guess field is
defined from the central-most sounding found in the modeling
domain. The effects of terrain, blocking and kinematic
effects, are applied to this first-guess, mean flow field. The
remaining surface and upper air observations are then applied
with the weighted interpolation scheme described earlier. The
strength of this approach is that it can yield a more
reasonable flow field in complex terrain where meteorological
observations are sparse. It does, however, introduce some
problems when generating a regional scale flow field.
If a wind field is to be generated over a relatively small
air basin, which includes complex terrain, where one may have
only one sounding within the domain, the aforementioned use of
a first-guess wind field is probably valid. When one is
generating a wind field over a larger domain, however, the
assumption of a first guess-field, based on one sounding, is
probably not appropriate. If, for example, a major topographic
barrier runs through the domain, it is quite likely that the
air flow on the opposite sides of the barrier may be very
different. Blocking, for example, will only occur on the
windward side of the barrier. If only one sounding is used,
this blocking and subsequent turning of the wind will only
occur on one side of the barrier, where in reality there may be
upslope flows on both sides of the barrier, with subsequent
terrain modifications to the flow. Thus, while one of the
strengths of the ARM3 wind generation model is its ability to
treat flows in complex terrain, its implementation may
ultimately lead to the generation of spurious winds on the
3-25
-------
sides of barriers opposite the station used to generate the
first-guess field.
The generation of mixing depths by the two models is
dismal. The implementation of the MESOPUFF-II algorithms
yields large discontinuities in adjoining grid cells, while the
ARM3 implementation of the calculation of mixing depths did not
reproduce any expected diurnal variability. As noted earlier,
the reasons for the behavior of the two algorithms are
generally understood. The MESOPUFF-II algorithms only rely
upon data from the nearest sounding. There has been no attempt
to use a weighted interpolation between soundings. Therefore,
if adjoining grid cells are nearer to different soundings, the
calculated mixing depths may be quite different. The only
independent variable in the ARM3 mechanical mixing depth
calculations is the 3000 meter wind speed at that grid cell.
At an altitude of 3000 meters, the wind speed will generally be
relatively high, thus leading to spuriously high mixing depths
at night. The daytime mixing depth calculations, generated by
the ARM3, are too low. The reasons for this have not been
investigated at this time; methods which overcome the problems
with both modeling systems' mixing depth algorithms have been
identified and should be implemented in Phase 2.
3.3.3 Further Air Quality Model Comparisons
After reviewing the mixing height fields and the wind
fields, a likely possibility for the higher concentrations
calculated by the ARM3 model was the tendency for the mixed
layer height to stay at relatively low levels over extended
periods. To test this hypothesis, the previously run test case
was run with the MESOPUFF-II generated mixing depth fields and
the ARM3 generated wind fields as input to the ARM3 air quality
model. This resulted in little appreciable change in the
calculated concentrations; the ARM3 maximum concentrations were
3-26
-------
still considerably higher than those generated by MESOPUFF-II.
Similarly, when the MESOPUFF-II air quality model was run using
the mixing depths generated by the ARM3 meteorological
processor, the calculated maximum concentrations did not change
appreciably.
The wind field from the MESOPUFF-II model was substituted
as input to the ARM3 air quality model. This produced
concentrations quite similar to the MESOPUFF-II air quality
model, using the same wind field. Since this change in the
wind fields generated such different results, it was
anticipated that there might be a region of very light winds
generated by the ARM3 processor. Examination of the plots of
the gridded wind fields of the two models for the time
preceding the maximum concentration calculation, did not elicit
any obvious reason for the vastly different concentration
calculations. The examinations focussed on the Level 2 and 3
fields from ARM3 and the Level I field from MESOPUFF-II, since
these were expected to be the respective transport levels for
the hypothetical source being modeled in the test cases. To
check which levels were being selected by the ARM3 to transport
puffs, the code was examined. This lead to the discovery of a
fundamental coding error.
The ARM3 calculates the center of mass of a puff and uses
the wind level at the altitude of the center of mass to
transport the puff. A subroutine is called to extract the wind
data for the appropriate grid cell and level. The algorithm
defined the transport altitude as a fraction of the total model
domain height. This fraction was then compared against the
absolute magnitude of the height of the various wind levels to
select the level of the wind to use. This resulted in a
normalized transport altitude, which would always be less than
or equal to 1, being compared with an array of heights ranging
from 10 to 2400 meters. Thus, the model would always choose
3-27
-------
the 10 meter wind level for all of its calculations. The 10
meter winds did exhibit some very light, almost calm conditions
between the hypothetical source and the receptor area in the
test case. Since the model was always selecting the 10 meter
winds, when the MESOPUFF-II mixed layer average winds were
substituted into appropriate levels in the ARM3 input files,
the ARM3 would always be picking the MESOPUFF-II mixed layer
winds as its 10 meter winds and thus was yielding similar
results.
Subsequent to the discovery of the coding error, the code
was patched, and the ARM3 meteorological and air quality
modeling system was again run. The concentrations from this
run were of the same order as those calculated by the MESOPUFF-
II.
3.3.3.1 Model Comparison Discussion: Until a more refined
technique can be evaluated, neither of the models under
consideration by the IWAQM to use in the interim are totally
satisfactory. With respect to the meteorological processors,
the MESOPUFF-II processor should incorporate a weighted
interpolation algorithm in the calculation of mixing depths to
avoid the extreme gradients encountered in the current
formulation. The ARM3 processor, on the other hand, should
employ different algorithms for calculating mixing depths and
should change the order of analysis in the diagnostic wind
field calculations to start out with a field interpolated from
the observations as the first-guess field, upon which the
kinematic and blocking effects of terrain should be applied.
The MESOPUFF-II model does not explicitly include any
treatment for the effects of terrain on long range transport
and dispersion. This is a relatively major deficiency with
respect to the needs of the IWAQM. While the ARM3 incorporates
3-28
-------
algorithms to treat such effects, the model has been found to
have a number of coding errors which render its use suspect.
One of the rationales for limiting the interim modeling
choices to either the MESOPUFF-II or the ARM3 was their
performance relative to other similar models. After the coding
error, described above, was discovered in the current version
of the ARM3 code, an examination was made of the originally
released version of the code. The code was found to contain
the same fundamental error found in the current version. Thus,
the evaluations of the ARM3 were carried out using only the
surface wind fields. For the case of the Oklahoma data, this
is not necessarily a major problem, since it was based on a
surface tracer release. However, the Savannah River evaluation
involved an elevated release into a wind level above the
surface level. If the surface level and the upper level were
very similar, the ARM3 evaluation results may be valid, but
unless the corrected model is re-run with the Savannah River
data, confidence should not be placed in the Savannah River
comparisons.
3-29
-------
4. REGULATORY CONSIDERATIONS FOR THE USE
OF LONG RANGE TRANSPORT MODELS
The Guideline explicitly identifies certain steady-state,
Gaussian plume models which are preferred for calculating
concentrations of inert pollutants for source-receptor
distances of less than 50 km. The MESOPUFF-II model has been
shown to replicate steady-state, Gaussian plume results, under
many conditions, when it is run with steady-state meteorology
and Pasquill-Gifford dispersion coefficients (Scire et al.,
1984a). When MESOPUFF-II is run in an operational mode,
however, the results could be quite different, since the
meteorology will be time varying and the simulated pollutant
transport can be quite different than the steady-state
approximation. Having two acceptable techniques (i.e.,
MESOPUFF II and a Guideline model) which can give different
results for the same application is untenable for regulatory
purposes. Therefore, while it would be desirable to run only
one model, it will be necessary to run both a preferred,
Guideline model and the recommended long range transport model
for many permitting situations.
For estimating impacts on AQRVs, the approach need not be
as operationally complicated. Since all AQRV analyses, covered
by this recommendation, involve secondary pollutants, the use
of MESOPUFF-II for all distance ranges was felt to be
acceptable. Although the use of MESOPUFF-II for sources
located close to the Class I area is questionable, their impact
should be quite small due to the small travel times involved.
Also, based on the very stringent Class I increments, any
sources locating close to a Class I area will usually be unable
to emit large quantities of primary pollutants.
4-1
-------
Similarly, there are likely to be occasions when long
range transport is to be considered, when it would be desirable
to run a simpler model, for screening purposes, than the
recommended long range transport model. In past applications,
steady-state, Gaussian plume models have been used. Comparison
runs of the MESOPUFF-II, run in a steady-state mode, and the
ISCST2 model indicate that the ISCST2 will frequently produce
similar or higher concentrations, but not under all
circumstances. Since the MESOPUFF-II will use a spatially and
temporally varying wind field when run in an operational mode,
the concentrations it produces will generally be lower than
those produced by the ISCST2 model. If, however, there are
prolonged periods of near stagnation conditions or if
recirculations occur, the ISCST2 will not necessarily produce
conservative concentrations.
4-2
-------
5. INTERIM RECOMMENDATIONS
From the review of the ARM3 and MESOPUFF-II models, the
IWAQM found neither model to be totally satisfactory. However,
due to the immediate need for a modeling system to estimate
impacts from sources farther than 50 km from a receptor and to
estimate the impacts on visibility and other AQRVs from all
distances, the IWAQM is recommending that the MESOPUFF-II model
be used for these analyses. This recommendation is to be
considered interim, until a more suitable technique can be
developed and tested. The IWAQM is recommending that MESOPUFF-
II be run in a somewhat different mode than was previously
suggested by EPA (EPA, 1988), and is recommending some
approaches to integrating the long range modeling system with
regulatory modeling requirements. A relatively simple method,
using steady-state, Gaussian plume models, is suggested as a
way to provide a first estimate of concentrations from long
range transport. The recommendations herein are the
suggestions of the IWAQM; implementation of these
recommendations is a matter for the appropriate regulatory
agencies and should be done in consultation with the
appropriate EPA regional office.
The following recommendations are divided into two
categories. The first is referred to as a "Level I" analysis.
This relatively simple analysis is expected to provide a
conservative estimate of concentrations due to long range
transport. If the Level I estimates indicate that adverse
conditions may be caused in the Class I area of concern, then a
"Level II" analysis, using the MESOPUFF-II modeling system,
should be undertaken.
5-1
-------
5.1 Protocol for Level I Long Range Transport Analysis
The IWAQM is suggesting that a steady-state, Gaussian
plume model may be acceptable as a first level long range
transport analysis technique under many conditions. It is
anticipated that this first level technique will, under most
conditions, yield a conservative estimate of ambient air
quality concentrations. However, under conditions of extended
stagnation or where recirculation patterns or convergence zones
are known to occur, this Level I technique should not be used.
If a Level I analysis indicates that there are exceedances of
the appropriate parameters, the more refined analyses, using
the Level II long range transport models, should be run for all
meteorological conditions, not just those identified through
the Level I technique. Level I analyses will consist of
running the appropriate steady-state, Gaussian plume models
with five years of meteorological data. Steady-state Gaussian
plume models used for Level I long range transport analyses
should utilize bivariate plume dispersion factors. It is
generally anticipated that plume impaction models, utilizing
uniform horizontal plume dispersion, would not be appropriate
for a long range transport analysis.
While the IWAQM is suggesting a relatively simple, Level
I, technique, it should not be construed as an endorsement of
that approach. With limited testing, the suggested Level I
technique appears to yield higher ambient pollutant estimates
than the Level II procedures. However, if there is any
question about the technical veracity of using the Level I
technique in a given situation, then it is completely
appropriate to proceed directly to a Level II analysis.
In general, the steady-state Gaussian plume models will
yield a high estimate of concentrations at distances beyond 50
kilometers, due to the spatially varying paths of puffs in the
5-2
-------
recommended long range transport model. As mentioned
previously, this is not always the case. Therefore, it is
expected that application of the Level I techniques will need
to be reviewed by those with sound professional judgement.
Under light wind conditions or under recirculation conditions,
the MESOPUFF-II model may yield higher concentrations than
steady-state, Gaussian plume models. Therefore, if these
conditions occur, the Level I analysis should be adequately
assessed. These conditions are not necessarily generally
defined; the IWAQM is not attempting to define them further.
It is anticipated that as more experience is gained with the
Level I and II techniques, further resolution on when either
method is appropriate can be better elucidated.
Since many of Class I areas are located in complex
terrain, the issue of the appropriate methods for applying
Level I long range transport modeling techniques to these areas
arises. Each analysis will have unique characteristics which
must be evaluated on a case-by-case basis. In many cases,
however, some of the following considerations will apply. As
noted previously, models incorporating bivariate plume
dispersion parameters are suggested for use in a Level I
analysis. This suggestion is based on the presumption that
after a plume has travelled 50 km or farther, it will have been
affected by a variety of processes, land use, and terrain
intervening along the trajectory of the plume. Therefore, it
may not be appropriate to use a plume impaction model for these
circumstances. Similarly, under many conditions, when long
range transport is involved in moving pollutants to a Class I
area, the plumes will be traversing steadily rising terrain.
Under the constraints of existing, steady-state, Gaussian plume
models, receptor heights are either restricted to be no
greater than the height of the stack which emitted the
pollutants, or in the case of some of the plume impaction
models, the concentrations are considered to be zero if the
5-3
-------
elevation of the receptor is sufficiently far above the
calculated plume height. Under conditions of steadily rising
terrain, the air flow, and plumes imbedded in that flow, will
tend to follow the terrain and become well mixed after a
distance of 50 km. Therefore, the various simple terrain
models, outlined in the Guideline, should generally provide a
reasonable approximation of transport and dispersion under
these conditions.
The results of the Level I technique will be analyzed
somewhat differently for increment and NAAQS analyses than for
visibility and other AQRV analyses. The methods for performing
these first level long range transport analyses are discussed
below.
5.1.1 Level I Long Range Transport Techniques for Analyzing
Increment and NAAQS
If Level I methods are to be utilized for increment and
NAAQS analysis, then all sources, whether closer or farther
than 50 km should be analyzed assuming no conversion of S02 to
S0=4 and NOX to N03 and no deposition. If that analysis
indicates that the increments or NAAQS are being exceeded, then
a full analysis, using the recommended long range transport
model in combination with the steady-state Gaussian model, as
described below, should be employed.
5.1.2 Level I Analysis Technique for Evaluating the Effects
of Long Range Transport and Regional Visibility
If a Level I approach is desired, it should be assumed
that all S02 has been converted to SO^ and that all NOX has been
converted to N03 in the analysis. Refer to Inset 1 for the
appropriate method for converting S02 and NOX to SO^ and N03.
5-4
-------
The procedures in Appendix B should be used to estimate the
visibility impacts.
Calculation of SO^ & NO^ For Visibility Screening
1. Run appropriate long range transport screening model.
2. Assume no conversion of S02 or NOX to other species.
(i.e. assume all NOX is emitted as N02 and that all S02
remains inert at this step.)
3. Multiply the hourly concentrations of S02 and NOX by
the ratios of the molecular weights of the secondary
species to the primary species.
Note: The molecular weights of S02 and SO^ are 64
and 96 and the molecular weights of N02 and
N03 are 46 and 62. Thus multiplying the
concentration of S02 by 1.5 will yield the
concentration of SO^ and multiplying the
concentration of N02 by 1.35 will yield the
concentration of N03.
4. The averaging time of interest is generally 1-hour.
Inset 1 - Method for calculating concentrations of SO^ and N03
from S02 and NOX.
5.1.3 Level I Analysis of Long Range Transport and
Depositional Impacts
If a Level I approach is to be taken for depositional
impacts, it should be assumed that concentrations of S02 and NOX
are deposited as S02 and HN03 (see Inset 2). Since the steady-
state, Gaussian plume models do not actually remove any mass
from the plume, when run in their recommended modes, this will
provide a conservative deposition estimate.
5-5
-------
Calculation of Deposition For Level I analysis
1. Run appropriate model for Level I analysis.
2. Assume no conversion of S02 or NOX to other species.
(i.e. assume all NOX is emitted as N02 and that all S02
remains inert at this step.)
3. Multiply the concentrations of NOX (ug/m3) , if
applicable by the ratio of the molecular weights of
the secondary species (HN03) to the primary
species (N02) .
Note: The molecular weights of N02 and HN03 are 46
and 63. Thus multiplying the concentration
of N02 (ug/m3) by 1.37 will yield the
concentration of HN03 (ug/m3) .
4. The majority of sulfur will be deposited as S02, so no
conversion is necessary.
5. The averaging times for deposition will generally
require a long term value (monthly, seasonal, or
annual) and short term value (1, 3 or 24-hour). Since
the Level I models will produce average values, they
must be converted to total rates.
Multiply the concentration of S02 or HN03 by the number
of seconds in the averaging time of interest to obtain
a total rate. (3.1536 x 107 seconds/year, 86400
seconds/day, 10800 seconds/3-hours, or 3600
seconds/hour)
6. Multiply the result of step 5 by the deposition
velocity for the appropriate pollutant. (0.005 m/s for
S02 or 0.05 m/s for HN03) . This will result in a
deposition value in units of ug/m2.
7. To convert to kg/hectare, multiply the result of step
7 by 10
-5
Inset 2 - Description of method to estimate deposition from S0
and NOX concentrations.
2
5-6
-------
5.2 Protocol for Level II Long Range Transport Analysis
The IWAQM is recommending that the MESOPUFF-II model be
run for source-receptor distances of greater than 50 km, up to
several hundred km, when predicting the concentrations for
criteria pollutants. For impacts on AQRVs, the IWAQM is
recommending the use of MESOPUFF-II for all source-receptor
distances. It is also recommended that the analyst generally
follow the recommendations in A Modeling Protocol for Applying
MESOPUFF-II to Long Range Transport Problems (EPA, 1988); the
primary exceptions to that protocol are the distance at which
time dependent dispersion curves are used, the use of the
chemical transformation and deposition algorithms (both wet and
dry), and the use of five years of meteorological data. The
IWAQM recommendations for model options and input parameters
can be found in Appendix A.
The two areas where the previous protocol and the IWAQM
protocol diverge are in the specification of the distance where
time dependent dispersion curves are used to calculate puff
dispersion and the use of the chemical transformation and
deposition options. The previously suggested protocol for the
use of MESOPUFF-II (EPA, 1992) stated that the distance at
which the time dependent dispersion curves should be use was
beyond 100 km. It was indicated that the reason for this
choice was that the 100 km distance was used during the
performance evaluations (EPA, 1986), which identified MESOPUFF-
II as one of the better performing models. The reference to
100 km appears to have been erroneous. A summary article,
describing the performance evaluations (Carhart et al., 1989)
stated:
The Turner curves were established from experiments
carried out over distances of 0-1 km from the source,
and yet some models use the curves out to 100 km. In
the MESOPUFF-II model, the transition from the Turner
to the Heffter formulation is made at 10 km from the
5-7
-------
source (a recommended user-input value adopted in
this study). On the other extreme are MESOPUFF and
MESOPLUME, which specify 100 km as the transition
distance.
Therefore, it seems clear that the performance evaluations
were indeed carried out with the transition between distance
and time dependent formulations set at 10 km. Thus the IWAQM
is recommending that the model be run with the 10 km setting.
The previous protocol also recommended that the model be
run assuming no chemical conversion or deposition of pollutants
(either wet or dry). For the purposes of IWAQM, namely to
calculate visibility impacts, secondary pollutants, such as
SO^, are the contributing pollutants. After reviewing the
algorithms used in the MESOPUFF-II code for calculating the
chemical conversion and deposition, the IWAQM considered them
simple, but adequate. Therefore, the IWAQM is recommending
that they be used, recognizing the following limitations.
First, the treatment of the aqueous phase conversion of S02 to
SO^ is likely to be greatly underestimated. Field measurements
have indicated that when a plume passes through a non-
precipitating cloud that the conversion of S02 to SO^ can be as
high as 100% per hour. The assumed conversion of 3% per hour
is, therefore, expected to be an underestimate. The model,
however, does not adequately treat the occurrence of non-
precipitating clouds. Therefore, the tendency will be to
underestimate SO^ formation when non-precipitating clouds would
be present, with a commensurate overestimation of the primary
S02 concentration.
Ultimately, when examining Air Quality Related Values
(AQRVs), one of the parameters of interest is frequently
deposition of SO^ and N03. Also, even when trying to estimate
impacts on visibility, an accurate assessment should include
the removal of these pollutants from the atmosphere.
5-8
-------
Therefore, the IWAQM is recommending that transformation and
deposition be actively modeled in order to reasonably estimate
the fate of pollutants undergoing long range transport.
The IWAQM recognizes that the chemical and depositional
algorithms have not been rigorously tested against field data.
Chemical and depositional algorithms do not readily lend
themselves to field evaluation due to their dependence on the
ambient conditions into which they are emitted and the
infeasibility of producing and releasing a unique, chemically
active tracer to assess chemical conversion and deposition on a
source by source basis. Given these limitations, it is not
likely that there will be such data available in the near
future. Therefore, proposing to use a technically credible
method to estimate the formation and deposition of secondary
pollutants, seems appropriate, and is the course recommended by
the IWAQM.
To be consistent with the current modeling guidance and to
attempt to adequately capture year-to-year variation in
meteorological conditions, the IWAQM is recommending that five
years of meteorological data be run with the MESOPUFF-II
analyses. There are some exceptions to this recommendation,
noted below.
One of the weaknesses of the MESOPUFF-II modeling system,
identified by the IWAQM, is the lack of treatment of terrain on
the air flow and the discontinuities in the mixing height
field. A meteorological processor, which uses a diagnostic
wind model and smooths out the mixing depth fields, has been
identified, but has not yet been thoroughly tested with the
MESOPUFF-II, and thus, is not being recommended at this time.
When this processor is ready for distribution, Appendix A will
be updated to include these enhancements. Until such time, the
MESOPUFF-II meteorological processor (MESOPAC) should be used.
5-9
-------
5.2.1 Increment and NAAQS Cumulative Long Range Transport
Analyses
Sources at distances greater than 50 km from a given
receptor should be analyzed with the MESOPUFF-II model, run as
described in Appendix A. This includes using the chemical
transformation and deposition algorithms. Sources within 50 km
of a receptor should be modeled using the appropriate model and
model options, as recommended in the Guideline. The
concentrations from these two analyses should be added on an
hour-by-hour, receptor-by-receptor, and pollutant-by-pollutant
basis.
There will be occasions when a source will be both nearer
and farther than 50 km from receptors in an analysis. It is
not intended that such a source be analyzed with both models.
In general, the nearer receptors will yield the controlling
concentrations for that source, therefore, the appropriate
Guideline model should generally be run for all receptors, for
that source. If, however, it is determined that a particular
source, modeled with a steady-state, Gaussian model, is
possibly causing an exceedance at a receptor more than 50 km
from the source, it is obviously appropriate to model its
impacts at that receptor with the long range transport model,
to more accurately estimate its impacts. Again, it is expected
that a fair degree of professional judgement will need to be
exercised in these analyses.
When analyzing impacts in a large Class I area it may be
appropriate to divide the steady-state, Gaussian plume analyses
into several source groups and use meteorological data
appropriate to each source group. For example, if a Class I
area is 150 km long, it is unlikely that sources located at
opposite ends of the area will experience the same meteorology.
Therefore, it may be appropriate to use representative
5-10
-------
meteorological data sets for each end of the Class I area,
particularly since the results will be combined with a model
which uses spatially and temporally varying wind fields. The
use of multiple meteorological data sets in a steady-state,
Gaussian plume modeling analysis is not the recommended
approach in the Guideline; therefore, this is a matter for the
appropriate regulatory agencies and should only be considered
on a case-by-case basis in consultation with the appropriate
EPA regional office.
As noted previously, it is recommended that both the long
range transport model and the steady-state Gaussian model be
run with five years of meteorological input data. The
exception to this is when the source being permitted is within
50 km of the Class I area and has collected at least one year
of on-site meteorological data. If a cumulative impact
analysis, which includes sources beyond 50 km, is required, the
meteorological period of analysis for both analyses should
correspond to the period of on-site meteorological data. If,
however, the source being permitted is beyond 50 km from the
affected Class I area, then it is recommended that the full
five years of data should be run for all analyses.
Combining the results of steady-state Gaussian models with
a Lagrangian puff model, such as the MESOPUFF-II, will produce
some contrived results. The steady-state model assumes
instantaneous transport, whereas the Lagrangian puff model
simulates the actual transport time. It may be physically
impossible for the emissions from a source modeled with the
steady-state model to reach a receptor at the predicted time,
given the wind speed. However, steady-state models do not
generally accurately predict the time and location of maximum
concentrations, but are routinely applied as if they do provide
such information. Therefore, while combining the results of
two fundamentally different modeling systems is somewhat
5-11
-------
contrived, it is the only workable approach under the
regulatory framework.
5.2.2 Visibility Analyses
The MESOPUFF-II model will be run for all sources, whether
nearer or farther than 50 km from a receptor, to provide
concentrations of SO^ and N03 for visibility calculations. A
complete description of the visibility calculations is
described in Appendix B.
SO^ has been identified as the primary constituent of
visibility degradation in the eastern U.S.; organic aerosols
and N03 are less important contributors, but a significant
species, nonetheless. Therefore, it is critical to have
estimates of SO^ in order to estimate visibility impacts.
Estimates of nitrates are desirable and can be obtained from
the long range transport modeling system; estimates of organic
aerosols, while more important to total scattering than
nitrates in most cases, are not readily estimated by the
modeling system. Therefore, organic aerosol concentrations
will only be considered as a background concentration for
determining regional visibility impacts. The MESOPUFF-II model
is capable of producing concentrations of SO^ and N03, whereas
the preferred models, listed in the Guideline are not.
Therefore, the IWAQM is recommending that when a cumulative
visibility impact analysis is desired, all sources be modeled
with MESOPUFF-II in order to estimate regional haze impacts.
It is also important to note that although the application of
MESOPUFF-II for short transport distances is questionable the
contribution for such sources is relatively unimportant. In
many permitting analyses, it is likely that only the source
being permitted will be analyzed. Therefore, the MESOPUFF-II
analysis will be greatly simplified.
5-12
-------
The primary purpose of this analysis is to identify
regional haze impacts. Plume blight analyses may need to be
performed (EPA, 1988), as well, depending on the location of a
source with respect to the Class I area. For sources located
more than 50 kilometers from a Class I area, the regional haze
analysis will generally be more appropriate. The methods,
outlined in Appendix B, are based upon analysis of ambient,
speciated fine particulate data, correlated with visibility
parameters. Therefore, in order to utilize this method,
concentrations of particulate species need to be obtained. The
MESOPUFF-II model can provide estimates of 30= and N03, but
assumptions need to be included with respect to the cations
associated with these radicals in the fine particulate, the
amount of primary carbon particles in the atmosphere, and the
formation of secondary organic particles. These are documented
in Appendix B.
5.2.3 Analysis of Other AQRVs
The modeling system, described herein, will only provide
estimates of the ambient concentrations of pollutants or the
depositional loadings at given locations. This modeling system
is adequate for providing such estimates, which then may be
used to assess the total or incremental impacts of these
concentrations or depositional loadings on aquatic and
terrestrial ecosystems. The methods for estimating effects on
these other AQRVs are not provided; such methods should be
determined in consultation with the appropriate Federal Land
Manager.
5.2.4 Analysis of Primary Emissions of Fine Particulates
The MESOPUFF-II model is not set up to directly simulate
emissions of primary particulates. It can, however, be used
for such an analysis, although it must be run as an independent
5-13
-------
simulation from other pollutants of interest. The method for
simulating primary emissions of fine particulates follows.
The MESOPUFF-II dispersion model includes the option for
modeling primary emissions of SO^. Hence, since these are in
particulate form, SO^ can be used as a surrogate for the
emissions of other fine particulate species. If the model is
run simulating only SO^ emissions, with the chemistry options
turned off, an estimate of the ambient concentrations of
primary particulate species can be obtained. The deposition
algorithms for particulate SO^ can be used for other fine
particulates, as well. Since the MESOPUFF-II does not include
a provision for gravitational settling, ambient concentrations
of the course particle fraction (diameters > 2.5 urn) may be
overestimated over transport distances greater than 50 km.
5-14
-------
6. SUMMARY
The IWAQM considered two Lagrangian puff models for use in
regional scale modeling analyses, the MESOPUFF-II and the ARM3.
Both models have known limitations. After comparison and
examination of the models, it was determined that the ARM3
contained a number of fundamental coding errors which rendered
its use suspect. Although the errors are correctable, they
invalidate previous evaluation analyses completed with the
ARM3. The MESOPUFF-II has been shown to be applicable, on a
theoretical basis, for long range transport; has been evaluated
with adequate data bases; and has been shown to not be biased
toward underestimation. Therefore, the IWAQM is recommending
the use of the MESOPUFF-II for long range transport analyses,
in the interim, until a more suitable model construct can be
developed.
It would be desirable to be able to recommend a single
modeling approach for estimating regional impacts on the NAAQS
and PSD increments, that is, have a single model which could
treat source-receptor distances of less than and greater than
50 km. Given the need for regulatory consistency, it was
determined that for calculating impacts on quantities regulated
under the Clean Air Act, techniques identified in the Guideline
on Air Quality Models (Revised) (EPA, 1986) should be used for
source-receptor distances of less than 50 km. The MESOPUFF-II
model should be run for source-receptor distances greater than
50 km and the results of these two techniques should be added
together.
A technique to estimate visibility impacts, based upon
statistical relationships between observed fine particle
concentrations and visibility parameters, has been provided.
The most important constituents in these calculations are the
6-1
-------
concentrations of sulfate and the relative humidity. The
modeling techniques, recommended above, will yield
concentration estimates of sulfate and nitrate and the
meteorological fields, used as input to the long range
transport model, contain the relative humidity. Therefore, the
recommended long range transport model can be used to estimate
the impacts of pollutant sources on regional visibility
degradation.
The recommendations for the use of the long range
transport model include the calculation of the chemical
conversion of S02 and NOX, and deposition. This is being
recommended whether the model is being applied to a PSD
increment analysis or to an AQRV analysis. It is recognized
that including these processes will decrease the amount of S02
calculated by the model, and hence, reduce the calculated
increment concentration. However, these processes are known to
occur over the long range transport distances being modeled
here and their inclusion in the modeling is considered to be
critical for an accurate estimation of impacts. Furthermore,
in an AQRV analysis, the secondary pollutants generated and
removed through these processes are frequently of most concern.
A number of limitations were identified in the recommended
modeling system. As improvements become available, it is
intended that they be incorporated into the system.
The IWAQM recognizes that by proposing the use of a
Lagrangian puff model for use in the interim, it is not
possible to guarantee that the results will be more restrictive
than more sophisticated techniques to be proposed during Phase
2 of the IWAQM work plan. The IWAQM still asserts, however,
that the interim system is the best of currently available
alternatives for simulating the impacts of long range transport
on the PSD increments, NAAQS, and AQRVs.
6-2
-------
REFERENCES
Benkley, C.W. and A. Bass. 1979. Development of Mesoscale Air
Quality Simulation Model: Vol. 3. User's Guide to
MESOPUFF. EPA-600/7-80-058. U.S. Environmental
Protection Agency, Atmospheric Research and Exposure
Assessment Laboratory, Research Triangle Park, NC.
Carhart, R.A., A.J. Policastro, M. Wastag, and L. Coke. 1989.
Evaluation of Eight Short-term Long-Range Transport Models
using Field Data. Atmospheric Environment, 23(1) :85-105.
EPA. 1986. Guideline on Air Quality Models (Revised). EPA-
450/2-78-027R. U.S. Environmental Protection Agency,
Office of Air Quality Planning and Standards, Research
Triangle Park, NC.
EPA. 1986. Evaluation of Short-term Long-Range Transport
Models, Volumes I and II. EPA-450/4-86-016a and b. U.S.
Environmental Protection Agency, Office of Air Quality
Planning and Standards, Research Triangle Park, NC.
EPA. 1992. A Modeling Protocol for Applying MESOPUFF-II to
Long Range Transport Problems. EPA-454/R-92-021. U.S.
Environmental Protection Agency, Office of Air Quality
Planning and Standards, Research Triangle Park, NC.
EPA. 1992. Workbook for Plume Visual Impact Screening and
Analysis (Revised). EPA-454/R-92-023. U.S. Environmental
Protection Agency, Office of Air Quality Planning and
Standards, Research Triangle Park, NC.
EPA. 1992. Interagency Workgroup on Air Quality Modeling
(IWAQM) Workplan Rationale. EPA-454/R-92-001. U.S.
Environmental Protection Agency, Office of Air Quality
Planning and Standards, Research Triangle Park, NC.
Godowitch, J.M. 1989. Evaluation and Sensitivity Analyses
Results of the MESOPUFF-II Model with CAPTEX Measurements.
EPA-600/3-89-056. U.S. Environmental Protection Agency,
Atmospheric Research and Exposure Assessment Laboratory,
Research Triangle Park, NC.
Moore, G.E., R.E. Morris, S.G. Douglas, and R.E. Kessler.
1990. Rocky Mountain Acid Deposition Model Assessment:
ARM3 Model Performance Evaluation. EPA-600/3-90-024.
U.S. Environmental Protection Agency, Atmospheric Research
and Exposure Assessment Laboratory, Research Triangle
Park, NC.
R-l
-------
Morris, R.E. and R.C. Kessler. 1987. Rocky Mountain Acid
Deposition Model Assessment: Review of Existing Mesoscale
Models for Use in Complex Terrain. EPA-600/3-87-013.
U.S. Environmental Protection Agency, Atmospheric Research
and Exposure Assessment Laboratory, Research Triangle
Park, NC.
Morris, R.E., R.C. Kessler, S.G. Douglas, K.R. Styles, and G.E.
Moore. 1988. Rocky Mountain Acid Deposition Model
Assessment: Acid Rain Mountain Mesoscale Model (ARMS).
EPA-600/3-88-042. U.S. Environmental Protection Agency,
Atmospheric Research and Exposure Assessment Laboratory,
Research Triangle Park, NC.
Scire, J.S., F.W. Lurmann, A. Bass, and S.R. Hanna. 1984a.
Development of the MESOPUFF-II Dispersion Model. EPA-
600/3-84-057. U.S. Environmental Protection Agency,
Atmospheric Research and Exposure Assessment Laboratory,
Research Triangle Park, NC.
Scire, J.S., F.W. Lurmann, A. Bass, and S.R. Hanna. 1984b.
User's Guide to the MESOPUFF-II Model and Related
Processor Programs. EPA-600/8-84-013. U.S. Environmental
Protection Agency, Atmospheric Research and Exposure
Assessment Laboratory, Research Triangle Park, NC.
Thomson, R.B., R.P. Angle, and S. Sakiyama. 1987. Selecting
Air Quality-Acid Deposition Models for Mesoscale
Application. J. Air Pollution Cont. Assoc., 37(3):260-
265.
R-2
-------
Appendix A
IWAQM Recommendations for
Running The MESOPUFF-II
Modeling System
-------
IWAQM RECOMMENDATIONS FOR RUNNING THE
MESOPUFF-II MODELING SYSTEM
The MESOPUFF-II modeling system was designed to include
flexibility in its use, so it could be used to simulate a wide
variety of conditions. Therefore, a number of user specifiable
parameters were included as input to the system. Furthermore,
the MESOPUFF-II modeling system is capable of simulating
varying spatial and temporal scales, which also need
specification. The purpose of the recommendations contained in
this appendix is to provide consistent, technically credible
methods for operating the MESOPUFF-II modeling system for
regulatory applications. These recommendations refer to model
specific variables and options without detailed definition of
the options. The user is referred to the MESOPUFF-II user's
manual for further information, where needed.
The primary focus of these recommendations is for the
evaluation of air pollution impacts on Class I areas, both
increment consumption and the impacts on Air Quality Related
Values (AQRVs), from sources located more than 50 kilometers
from the potentially affected area. The general procedures
outlined could also be used for the assessment of pollution
impacts, for source-receptor distances greater than 50 km,
outside of Class I areas. The applicability of such techniques
should, however, be confirmed with the appropriate regulatory
authorities.
Other documents which may be useful to the users of these
recommendations include:
User's Guide to the MESOPUFF-II Model and Related
Processor Programs, (EPA-600/8-84-013).
A Modeling Protocol for Applying MESOPUFF-II to Long Range
Transport Problems, (EPA-454/R-92-021).
Development of the MESOPUFF-II Dispersion Model, (EPA-
600/3-84-057).
Spatial Scale
The MESOPUFF-II modeling system is generally applicable to
source-receptor distances greater than 50 kilometers. The grid
dimensions used in the MESOPUFF-II system, upon which the
meteorological fields are defined and the puff calculations are
performed, should not exceed 1000 km in the east-west direction
or 600 km in the north-south direction (EPA-454/R-92-021).
With grid distances greater than this, significant errors can
be introduced through the orthogonal nature of the modeling
grid, superimposed on the curved surface of the earth. The
computational grid size should be such that sources and
receptors of interest are not too close to the edge of the
grid, since once puffs leave the grid, they are eliminated from
A-l
-------
the computations; concentrations may be significantly
underestimated for sources or receptors too close to the edge
of the computational grid.
Spatial Resolution
The various grid systems used in the MESOPUFF-II modeling
system are all relative to the initially defined meteorological
grid. Therefore, the resolution (grid spacing) of the
meteorological grid is of prime importance. Since the
meteorological fields, generated by the MESOPAC processor, are
defined from the interpolation of available observations, the
practical resolution of those fields will depend on the
distance between observation stations. Therefore, the maximum
recommended resolution is | the median distance between
observation stations. Finer resolutions can be used, but at
the cost of some computation time. If an area in the domain is
considered very important and has relatively dense
meteorological observations, then the resolution should be
based on this area of more refined observations.
In general, all available meteorological stations within
the initially defined grid system should be included in the
analysis. Stations relatively near to the boundaries,
particularly upper air stations, should also be included, as
they will improve the representativeness of the wind fields
generated by the interpolation.
Temporal Scale
In order to capture year-to-year meteorological
variability and the effect that can have on air pollution
concentrations, five years of meteorological data should be run
with the MESOPUFF-II modeling system.
Precipitation and Upper Air Meteorological Processors
The version of MESOPAC, the meteorological processor for
the MESOPUFF-II modeling system, being distributed, can make
use of upper-air meteorological data in either a TD-5600 format
or the newer TD-6201 format. Processors for both of these data
types are provided with the modeling system. Precipitation
data is now distributed in a TD-3240 format. Descriptions and
information on running these processors are provided below.
User Instructions - Preprocessor Programs
READ56/READ62 Upper Air Preprocessors
READ56 and READ62 are preprocessing programs which extract
and process upper air wind and temperature data from standard
A-2
-------
NCDC data formats into a form required by the MESOPAC
meteorological model. READ56 operates on the older TD-5600
data format. Although this format is not currently used by
NCDC, many historical data sets contain data in this format.
READ62 processes data in the current TD-6201 format.
Although the data inputs are different, the user inputs to
the program are identical as is the processed output file. In
the user input file, the user selects the starting and ending
dates of the data to be extracted and the top pressure level.
Also selected are processing options determining how missing
data are treated. The programs will flag or eliminate sounding
levels with missing data.
If the user selects the option to flag (rather than
eliminate) levels with missing data, the data field of the
missing variables are flagged with a series of nines. If the
option to eliminate levels with missing data is chosen, only
sounding levels with all values valid will be included in the
output data file.
Although MESOPAC allows missing values of wind speed, wind
direction, and temperature at intermediate levels (i.e., levels
other than the surface and model top), the user is cautioned
against using soundings with significant gaps due to missing
data. For example, adequate vertical resolution of the morning
temperature structure near the surface is especially important
to the model for predicting daytime mixing heights. It should
be kept in mind that the model will fill in missing data by
assuming that a straight-line interpolation between valid
levels is appropriate. If this assumption is questionable, the
sounding should not be used with the model.
Two input files are required by the preprocessor: a user
input control file and the NCDC upper air data file. Two
output files are produced: a list file summarizing the user
option selected and missing data monitored and the processed
data file in MESOPAC format. Table A-l contains a listing of
the input and output files for READ56 and READ62.
The READ56/READ62 control file consists of two lines of
data entered in FORTRAN free format. A description of each
input variable is shown in Table A-2. A sample input file is
shown in Table A-3.
The output data file (UP.DAT) produced by READ56/READ62 is
a formatted file containing the pressure, elevation,
temperature, wind speed, and wind direction at each sounding
level. The contents and format of the UP.DAT file are
discussed in Section A.2.3.
A-3
-------
Table A-l
(a) READ56 Input and Output Files
Unit
5
6
8
9
File Name
READ56.INP
READ56 .LST
TDF56.DAT
UP . DAT*
Type
input
output
input
output
Format
formatted
formatted
formatted
formatted
Description
Control file containing user
inputs
List file (line printer
output file)
Upper air data in NCDC
TD-5600 format
Output file containing
processed upper air data in
format required by MESOPAC
(b) READ62 Input and Output Files
Unit
5
6
8
9
File Name
READ62 . INP
READ62.LST
TD6201 .DAT
UP . DAT*
Type
input
output
input
output
Format
formatted
formatted
formatted
formatted
Description
Control file containing user
inputs
List file (line printer
output file)
Upper air data in NCDC TD-
6201 format
Output file containing
processed upper air data in
format required by MESOPAC
'Should be renamed UP1.DAT (for upper air station #1), UP2.DAT (for station
#2), etc. for input into the MESOPAC model.
A-4
-------
Table A-2
READ56/READ62 Control File Inputs
RECORD 1. Starting and ending date/hour, top pressure level to extract.
Columns
*
*
*
*
*
*
*
Variable
IBYR
IBDAY
IBHR
IEYR
IEDAY
IEHR
PSTOP
Type
integer
integer
integer
integer
integer
integer
real
Description
Starting year of data to extract (two
digits)
Starting Julian day
Starting hour (00 or 12 GMT)
Ending year of data to extract
(two digits)
Ending Julian day
Ending hour (00 or 12 GMT)
Top pressure level (mb) for which data is
extracted (possible values are 850 mb, 700
mb, or 500 mb) . The output file will
contain data from the surface to the
" PSTOP" -mb pressure level.
* Entered in FORTRAN free format
Table A-2 (Concluded)
READ56/READ62 Control File Inputs
RECORD 2. Missing data control variables
Columns
Variable
Type
Description
LHT
logical
Height field control variable. If LHT =
T, a sounding level is eliminated if the
height field is missing. If LHT = F, the
sounding level is included in the output
file but the height field is flagged with
a "9999", if missing.
LTEMP
logical
Temperature field control variable. If
LTEMP = T, a sounding level is eliminated
if the temperature field is missing. If
LTEMP = F, the sounding level is included
in the output file but the temperature
field is flagged with a "999.9", if
missing.
A-5
-------
LWD
logical
Wind direction field control variable. If
LWD = T, a sounding level is eliminated if
the wind direction field is missing. If
LWD = F, the sounding level is included in
the output file but the wind direction
field is flagged with a "999", if missing.
LWS
logical
Wind speed field control variable. If LWS
= T, a sounding level is eliminated if the
wind speed field is missing. If LWS = F,
the sounding level is included in the
output file but the wind speed field is
flagged with a "999", if missing.
* Entered in FORTRAN free format
Table A-3
Sample READ56/READ62 Control File (READ56.INP, READ62.INP)
79, 365, 00, 80, 002, 00, 500. -- Beg. yr, day, hr(GMT), Ending yr, day, hr, top pressure
level
.TRUE., .TRUE., .TRUE., .TRUE. -- Eliminate level if height, temp., wind direction, wind
direction, wind speed missing ?
PXTRACT Precipitation Data Extract Program
PXTRACT is a preprocessor program which extracts
precipitation data for stations and time periods of interest
from a fixed length, formatted precipitation data file in NCDC
TD-3240 format. Hourly precipitation data is available from
NCDC in large blocks of data sorted by station. For example, a
typical TD-3240 file for California may contain data from over
100 stations statewide in blocks of time of 30 years or more.
Modeling applications require the data sorted by time rather
than station, and usually involve limited spatial domains of
tens of kilometers or less and time periods of one year or
less. PXTRACT allows data for a particular model run to be
extracted from the larger data file and creates a set of
station files that are used as input files by the second-stage
precipitation preprocessor, PMERGE (see PMERGE section)
NOTE: If wet removal is not to be considered by the
MESOPUFF-II dispersion model, no precipitation processing needs
to be done. PXTRACT (and PMERGE) are required only if wet
removal is an important removal mechanism for the modeling
application of interest. In addition, if wet removal is a
factor, the user has the option of creating a free-formatted
precipitation data file that can be read by MESOPAC. This
option eliminates the need to run the precipitation
preprocessing programs for short MESOPAC runs (e.g., screening
runs) for which the input data can easily be input manually.
A-6
-------
The input files used by PXTRACT include a control file
(PXTRACT.INP) containing user inputs, and a data file
(TD3240.DAT) containing the NCDC data in TD-3240 format. The
precipitation data for stations selected by the user is
extracted from the TD3240.DAT file and stored in separate
output files (one file per station) called xxxxxx.DAT, where
xxxxxx is the station identification code. PXTRACT also
creates an output list file (PXTRACT.LST) which contains the
user options and summarizes the station data extracted. Table
A-12 contains a summary of PXTRACT1s input and output files.
The PXTRACT control file contains the user-specifled
variables which determine the method used to extract
precipitation data from the input data file (i.e., by state, by
station, or all stations), the appropriate state or station
codes, and the time period to be extracted. A sample PXTRACT
control file is shown in Table A-13. The format and contents
of the file are described in Table A-14.
The PXTRACT output list file (PXTRACT.LST) contains a
listing of the control file inputs and options. It also
summarizes the station data extracted from the input TD-3240
data file, including the starting and ending date of the data
for each station and the number of data records found. The
PXTRACT output data files consist of precipitation data in
TD-3240 format for the time period selected by the user. Each
output data file contains the data for one station.
Table A-12
PXTRACT Input and Output Files
Unit
1
2
6
7
File Name
PXTRACT. INP
TD3240 .DAT
PXTRACT . LST
idl .DAT
(idl is the
6-digit
station
code for
station
#1, e.g.,
040001)
Type
input
input
output
output
Format
formatted
formatted
formatted
formatted
Description
Control file containing user
inputs
Precipitation data in NCDC
TD-3240 format
List file (line printer output
file)
Precipitation data (in
TD-3240) format for station #1
for the time period selected
by the user
A-7
-------
id2.DAT
(id2 is the
6-digit
station
code for
station
#2, e.g.,
040002)
output
formatted
Precipitation data (in
TD-3240) format for station #2
for the time period selected
by the user
(Up to 200 new precipitation data files are allowed by PXTRACT).
Table A-13
Sample PXTRACT Control File (PXTRACT.INP)
2
5
040001
040002
040003
040004
040005
80 01 01 01 80 01 02 24
Selection code, ICODE
Number of states or stations
State or station code
Starting yr, month, day, hour(01-24),
ending yr, month, day, hour
Table A-14
PXTRACT Control File Inputs (PXTRACT.INP)
RECORD 1. Data selection code.
Columns
*
Variable
ICODE
Type
Integer
Description
Selection Code:
1 = extract all stations within state
or states requested
2 = input a list of station codes of
stations to extract
3 = extract all stations in input file
with data for time period of
interest
* Entered in Fortran free format
A-8
-------
Table A-14 (Continued)
PXTRACT Control File Inputs (PXTRACT.INP)
RECORD 2. Number of state or station codes
(This record is included only if ICODE = 1 or 2)
Columns
*
Variable
N
Type
Integer
Description
If ICODE = 1:
Number of state codes to follow
If ICODE = 2:
Number of station codes to follow
* Entered in Fortran free format
Table A-14 (Continued)
PXTRACT Control File Inputs (PXTRACT.INP)
RECORD 3, 4,
2+N. State or station codes of data to be
extracted
Columns
1-6
Format
16
Variable
I DAT
Description
If ICODE=1: State code (two digits)
If ICODE=2: Station code (six digits)
consisting of state code
(two digits) followed by
station ID (four digits)
A-9
-------
Table A-14 (Concluded)
PXTRACT Control File Inputs (PXTRACT.INP)
NEXT RECORD. Starting/ending dates and times
Columns
1-2
4-5
7-8
10-11
13-14
16-17
19-20
22-23
Format*
12
12
12
12
12
12
12
12
Variable
IBYR
IBMO
IBDAY
IBHR
IEYR
IEMO
IEDAY
IEHR
Description
Beginning year of data to process (two
digits)
Beginning month
Beginning day
Beginning hour (01-24 LST)
Ending year of data to process (two
digits)
Ending month
Ending day
Ending hour (01-24 LST)
'Record format is (8(i2,lx)
Table A-16
Sample TD-3240 Format Precipitation Data File (040001.DAT)
HPD04000100HPCPHI19820100010010100 00002
HPD04000100HPCPHI19820100010010200 00002
HPD04000100HPCPHI19820100010010300 00004
HPD04000100HPCPHI19820100020010700 00000
HPD04000100HPCPHI19820100030011300 OOOOOM
HPD04000100HPCPHI19820100120010400 OOOOOM
HPD04000100HPCPHI19820100120010500 00000
PMERGE Precipitation Data Preprocessor
PMERGE reads, processes and reformats the precipitation
data files created by the PXTRACT program, and creates an
unformatted data file for input into the MESOPAC meteorological
model. The output file (PRECIP.DAT) contains the precipitation
data sorted by hour, as required by MESOPAC, rather than by
station. The program can also read an existing unformatted
output file and add stations to it, creating a new output file.
PMERGE also resolves "accumulation periods" and flags missing
or suspicious data.
Accumulation periods are intervals during which only the
total amount of precipitation is known. The time history of
precipitation within the accumulation period is not available.
For example, it may be known that within a six-hour
accumulation period, a total of a half inch of precipitation
fell, but information on the hourly precipitation rates within
the period is unavailable. PMERGE resolves accumulation
A-10
-------
periods such as this by assuming a constant precipitation rate
during the accumulation period. For modeling purposes, this
assumption is suitable as long as the accumulation time period
is short (e.g., a few hours). However, for longer accumulation
periods, the use of the poorly time-resolved precipitation data
is not recommended. PMERGE will eliminate and flag as missing
any accumulate periods longer than a user-define maximum
length.
PMERGE provides an option to "pack" the precipitation data
in the unformatted output in order to reduce the size of the
file. A "zero packing" method is used to pack the
precipitation data. Because many of the precipitation values
are zero, strings of zeros are replaced with a coded integer
identifying the number of consecutive zeros that are being
represented. For example, the following record with data from
20 stations requires 20 unpacked words:
0.0, 0.0, 0.0, 0.0, 0.0, 1.2, 3.5, 0.0, 0.0, 0.0,
0.0, 0.0, 0.0, 0.7, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0,
These data in packed form would be represented in six
words:
-5. , 1.2, 3.5, -6., 0.7, -6.
where five zero values are replaced by -5., six zero values are
replaced by -6., etc. With many stations and a high frequency
of zeros, very high packing ratios can be obtained with this
simple method. All of the packing and unpacking operations are
performed internally by PMERGE and MESOPAC, and are transparent
to the user. The header records of the data file contain
information flagging the file to MESOPAC as a packed or
unpacked file. If the user selects the unpacked format, each
precipitation value is assigned one full word.
The input files used by PMERGE include a control file
(PMERGE.INP), an optional unformatted data file (PBIN.DAT)
created in a previous run of PMERGE, and up to 150 TD-3240
precipitation station files (e.g., as created by PXTRACT). The
output file consists of a list file and a new unformatted data
file in MESOPAC format with the data for all stations sorted by
hour. Table A-17 lists the name, type, format, and contents of
PMERGETs input and output data files.
The PMERGE control file (PMERGE.INP) contains the user-
specified input variables indicating the number of stations to
be processed, a flag indicating if data is to be added to an
existing, unformatted data file, the maximum length of an
accumulation period, packing options, station data, and time
zone data. PMERGE allows data from different time zones to be
merged by time-shifting the data to a user-specified base time
A-ll
-------
zone. A sample PMERGE control file is shown in Table A-18.
The format and contents of the file are described in Table
A-19.
The PMERGE output list file (PMERGE.LST) contains a
listing of the control file inputs and options. It also
summarizes the number of valid and invalid hours for each
station including information on the number of hours with zero
or non-zero precipitation rates and the number of accumulative
period hours. Additional statistics provide information by
station on the frequency and type of missing data in the file
(i.e., data flagged as missing in the original data file, data
which is part of an excessively long accumulation period, or
data missing from the input files before (after) the first
(last) valid record.
Table A-17
PMERGE Input and Output Files
Unit
3
4
5
6
7
8
File Name
PBIN.DAT
PRECIP.DAT
PMERGE . INP
PMERGE. LST
user input
file name
user input
file name
Type
input
output
input
output
input
input
Format
unformatted
unformatted
formatted
formatted
formatted
formatted
Description
Existing PMERGE data file to
which stations are to be added
(Used only if NBF=1)
Output data file created by
PMERGE (PRECIP.DAT is an input
file to MESOPAC)
Control file containing user
inputs
List file (line printer output
file)
Precipitation data (in TD-
3240) format for station #1.
(Output file of PXTRACT)
Precipitation data (in TD-
3240) format for station #2.
(Output file of PXTRACT)
(Up to 150 new precipitation data files are allowed by PMERGE) .
5 0 12
(614)
040001.dat
040002.dat
040003.dat
040004.dat
040005.dat
Table A-18
Sample PMERGE Control File (PMERGE.INP)
-- No. stations,no. binary files,max. accum. period,base time
zone,ioform(l=binary,2=formatted),pack(0=no, l=yes) --
-- Input file name, time zone
-- (Aid,13)
A-12
-------
82 01 01 01 82 01 03 01
hour (01-24)
-- Starting yr, month, day, hour (01-24), ending yr, month, day,
-- (8(12,IX))
Table A-19
PMERGE Control File Inputs (PMERGE.INP)
RECORD 1. General run information.
Columns
1-4
5-8
9-12
13-16
17-20
Format*
14
14
14
14
14
Variable
NFF
NBF
MAXAP
IOTZ
IOPACK
Description
Number of formatted 80-column NCDC input
files to be processed (up to 150)
Flag indicating if data is to be added
to an existing unformatted precip. data
file (0=no, l=yes)
Maximum allowed length of an
accumulation period (hours) . It is
recommended that MAXAP be set to 24
hours or less.
Time zone of output data (05=EST,
06=CST, 07=MST, 08=PST)
Flag indicating if output data are to be
packed (0=no, l=yes)
'Record format is (514)
Table A-19 (Continued)
PMERGE Control File Inputs (PMERGE.INP)
RECORDS 2, 3, ... 1+NFF. File names and time zone of each station
(Each record has the following format)
Columns
1-10
12-13
Format*
A10
12
Variable
CFFILES
ISTZ
Description
Name of file containing formatted
precipitation data (TD-3240 format)
(PXTRACT output file) . First six digits
of file name must contain station code
(SSIIII) , where SS is the two digit
state code, and IIII is the station ID)
Time zone of station (08=PST)
Record format is (alO,lx,12;
A-13
-------
Table A-19 (Concluded)
PMERGE Control File Inputs (PMERGE.INP)
NEXT RECORD. Starting/ending dates and times
Columns
1-2
4-5
7-8
10-11
13-14
16-17
19-20
22-23
Format*
12
12
12
12
12
12
12
12
Variable
IBYR
IBMO
IBDAY
IBHR
IEYR
IEMO
IEDAY
IEHR
Description
Beginning year of data to process (two
digits)
Beginning month
Beginning day
Beginning hour (01-24 LST)
Ending year of data to process (two
digits)
Ending month
Ending day
Ending hour (01-24 LST)
Record format is (8(i2,lx))
MESOPAC Input Fields
The version of MESOPAC discussed herein, includes some
enhancements from earlier versions. These include an expanded
format for including precipitation data (TD-3240) and the
inclusion of site specific data on the wind speed measurements,
specifically the wind speed measurement height and the surface
roughness length appropriate for the surface station site.
Inclusion of the TD-3240 requires running the precipitation
processors described previously. Some coding enhancements were
implemented to trap artifacts, produced by the model, and treat
them in a consistent manner.
Card Group 1 - TITLE
Columns
1-80
Typ*
CA
Variable
Name
title(20)
Description
80 Character Title
Recommended Value
Appropriate Choice
A-14
-------
Card Group 2 - GENERAL RUN INFORMATION
Columns
1-5
6-10
11-15
16-20
21-25
26-30
31-35
Typ*
I
I
I
I
I
I
I
Variable
Name
nyr
idystr
ihrmax
nssta
nusta
ibtz
npsta
Description
Two Digit Year
Starting Julian Day
Number of Hours in Run
# Surface Stations
# Rawinsonde Stations
Reference Time Zone
# Precipitation
Stations
Recommended Value
Appropriate Choice
Appropriate Choice
Appropriate Choice
< 25
< 10
5=EST 6=CST 7=MST
8 = PST
< 200
Card Group 3 - GRID DATA
Columns
1-5
6-10
11-20
Typ*
I
I
R
Variable
Name
imax
jmax
dgrid
Description
# X grid points
(west-east)
# Y grid points
(south-north)
grid spacing (m)
Recommended Value
< 40
< 40
Vs median distance
between stations
Card Group 4 - OUTPUT OPTIONS
Columns
1-5
6-10
11-15
16-20
21-25
26-30
31-35
36-40
Typ*
L
L
I
L
I
I
I
I
Variable
Name
Isave
Iprint
iprint
Ibd
ndyl
nhrl
ndy2
nhr2
Description
Binary To Disk for
Post -process ing
Print Met fields
Print Interval (hours)
Print Met Input
Julian day to start
printing input
Hour to start printing
input
Julian day to stop
printing input
Hour to stop printing
input
Recommended Value
T
F (can produce
voluminous output)
> 1 (used only if
lprint=T)
F (can produce
voluminous output)
(used only if lbd=T)
(used only if lbd=T)
(used only if lbd=T)
(used only if lbd=T)
A-15
-------
Card Group 5 - GRIDDED LAND USE CATEGORIES
Columns
1-80
Format
(4012)
Typ*
IA
Variable
Name
Ilandu
(40,40)
Description
Land use categories
for each grid point
Recommended Value
As Appropriate
'jmax' cards are required, each card with 'imax' land use categories
(corresponding to X-coordinates 1 to imax) . The first card contains
values for Y = jmax, the second card for Y = jmax-1, etc.
Card Group 6 - DEFAULT OVERRIDE OPTIONS
Columns
1
2
3
4
5
6
7
Typ*
IAE
IAE
IAE
IAE
IAE
IAE
IAE
Variable
Name
iopts (1)
iopts (2)
iopts (3 )
iopts (4)
iopts (5)
iopts (6)
iopts (7)
Description
Use Default Surface
Wind Speed Measurement
Height (Default=10m)
Use Default von Karman
Constant (Def ault=0 . 4)
Use Default Friction
Velocity Constants
(Defaults: Y=4-7-
A=1100)
Use Default Mixing
Height Constants
(Defaults: B=1.41,
E=0.15, Az=200m,
ae/3zmln=0.0010K/m,
N=2400
Use Default Wind Field
Variables (Defaults:
Vertically Averaged
Winds used from Ground
to Mixing Height,
Vertically Averaged
Winds used from Mixing
Height to 700 mb, &
Scan Radius for Wind
Field Interpolation
RADIUS=99.0km)
Use Default Surface
Roughness Lengths
(Determined from Land
Use Categories)
Use Default Heat Flux
Estimates (Can not be
changed)
Recommended Value
0
0
0
0
0
0
0
A-16
-------
8
9
10
IAE
IAE
IAE
iopts (8)
iopts (9)
iopts (10)
Use Default Radiation
Reduction Factors
(Defaults: 1.0, 0.91,
0.84, 0.79, 0.75,
0.72, 0.68, 0.62,
0.53, 0.41, 0.23
Use Default Heat Flux
Constants (Default:
RADC=0.3)
If iopts(10)=l,
starting date of run
is not the beginning
of the meteorological
file, else set
iopts (10) =0
0
0
0 or 1 as
appropriate
Card Groups 7-14 - NEW VALUES TO REPLACE DEFAULT PARAMETERS
The default options are recommended for all regulatory uses of the
MESOPUFF-II modeling system. If the defaults are used, these card groups
do not need to be included. See the user's manual for further
information .
Card Group 15 -
Columns
1-5
6-15
16-25
26-35
36-45
46-50
51-55
cf: e~c
Typ*
IAE
RAE
RAE
RAE
RAE
RAE
IAE
IAE
SURFACE STATION DATA, 'nssta1 cards - one for each
CD144/TD9657 surface station
Variable
Name
idcd
xscoor
yscoor
slat
slong
szone
isunit
idprcp
Description
Surface station ID for
CD144 data (5 digits)
X-coordinate of
station (in grid
units)
Y-coordinate of
station (in grid
units)
Station latitude
(decimal degrees)
Station longitude
(decimal degrees)
Station time zone
(5=EST 6=CST 7=MST
8=PST)
Logical unit number of
CD144 surface data
Surface station ID for
TD9657 data (6 digits)
Recommended Value
As Appropriate
As Appropriate
As Appropriate
As Appropriate
As Appropriate
As Appropriate
As Appropriate
TD9657 data not
implemented in this
version
A-17
-------
66-70
71-75
76-80
IAE
R
R
Card Group 16 -
Columns
1-5
6-15
16-25
26-35
36-45
46-50
51-55
Typ*
IAE
RAE
RAE
RAE
RAE
RAE
IAE
ipunit
zmsurf
zOsurf
Logical unit number of
TD9657 data
(ipunit=999 if TD9657
data is not available
for this station)
Wind measurement
height at station idcd
(m)
Surface Roughness
Length (m) for surface
station site
999
TD9657 data not
implemented in this
version
As Appropriate
As Appropriate
RAWINSONDE STATION DATA, 'nusta1 cards - one for each
TDF5600 or TDF6201 rawinsonde station.
Variable
Name
idtd
xucoor
yucoor
ulat
ulong
uzone
iuunit
Description
Rawinsonde station
identification number
(5 digits)
X-coordinate of
station (in grid
units)
Y-coordinate of
station (in grid
units)
Station latitude
(decimal degrees)
Station longitude
(decimal degrees)
Station time zone
(5=EST 6=CST 7=MST
8=PST)
Logical unit of
processed TDF5600 or
TDF6201 data
Recommended Value
As Appropriate
As Appropriate
As Appropriate
As Appropriate
As Appropriate
As Appropriate
As Appropriate
A-18
-------
Card Group 17 -
Columns
1-5
6-15
16-25
Typ*
IAE
RAE
RAE
PRECIPITATION STATION DATA. 'npsta1 cards - one for
each precipitation station.
Variable
Name
idp
xpcoor
ypcoor
Description
Precipitation station
IDs
X-coordinate (grid
units)
Y-coordinate (grid
units)
Recommended Value
As Appropriate
As Appropriate
As Appropriate
The codes under type correspond to the following:
I Integer Variable
IA Integer Array
IAE Integer Array Element
R Real Variable
RA Real Array
RAE Real Array Element
C Character Variable
CA Character Array
L Logical Variable
MESOPUFF-II Input Fields
The version of the MESOPUFF-II described below contains
some enhancements from earlier versions. These allow for the
initialization of runs from the results of previous runs and
the output of deposition calculations, both wet and dry.
Some modifications to the modeling code were implemented
to make some of the default values consistent with those
recommended here, and to make some of the calculations, such as
plume rise, consistent with the methods implemented in EPA's
preferred models.
Card Group 1 - TITLE
Columns
1-80
Typ*
CA
Variable
Name
title(20)
Description
80 Character Title
Recommended Value
Appropriate Choice
A-19
-------
Card Group 2 - GENERAL RUN INFORMATION
Columns
1-5
6-10
11-15
16-20
21-25
26-30
31-35
36-40
41-45
Typ*
I
I
I
I
I
I
I
I
I
Variable
Name
nsyr
nsday
nshr
nadvts
npts
nareas
nrec
npec
icont
Description
Two digit year of run
Starting Julian day
Starting Hour
Number of hours in run
# point sources
# area sources
# non-gridded
receptors
# of chemical species
to model (=1 S02 =2
S02 & SO; | =3 S02,
SO;, & NOX =5 S02,
so;, NOY, HNO,, & NO;
Continuation run
(0=no, l=yes)
Recommended Value
As Appropriate
As Appropriate
(00-23)
As Appropriate
(< 20)
(< 5)
(< 180)
2 if S02 Source
5 if S02 and NOX
source
As Appropriate
Card Group 3 - COMPUTATIONAL VARIABLES
Columns
1-5
6-10
11-15
16-20
21-25
26-30
31-35
Typ*
I
I
I
L
R
L
R
Variable
Name
iavg
npuf
nsamad
Ivsamp
wsamp
Isgrid
agemin
Description
Concentration
averaging time (hours)
Puff Release Rate
(puffs/hour)
Minimum Sampling Rate
(samples /hour)
Variable Sampling
Option (T or F)
Reference wind speed
for variable sampling
Calculate gridded
concentrations at
sampling grid points
Minimum age of puffs
to sample (seconds)
Recommended Value
1
4
2
T
2.
As Appropriate
(T or F)
900.
Card Group 4 - GRID INFORMATION
Columns
1-5
Typ*
I
Variable
Name
iastar
Description
Element # of the
meteorological grid
defining the beginning
Recommended Value
(l
-------
6-10
11-15
16-20
21-25
26-30
31-35
36-40
41-45
I
I
I
I
I
I
I
I
iastop
j astar
jastop
isastr
isastp
j sastr
j sastp
meshdn
Element # of the
meteorological grid
defining the end of
the computational grid
in the X-direction
Element # of the
meteorological grid
defining the beginning
of the computational
grid in the
Y-direction
Element # of the
meteorological grid
defining the end of
the computational grid
in the Y-direction
Element # of the
meteorological grid
defining the beginning
of the sampling grid
in the X-direction
Element # of the
meteorological grid
defining the end of
the sampling grid in
the X-direction
Element # of the
meteorological grid
defining the beginning
of the sampling grid
in the Y-direction
Element # of the
meteorological grid
defining the end of
the sampling grid in
the Y-direction
Sampling grid spacing
factor
(l
-------
Card Group 5 - TECHNICAL OPTIONS
Columns
1-5
6-10
11-15
16-20
21-25
Typ*
L
L
L
L
L
Variable
Name
Igauss
Ichem
Idry
Iwet
13vl
Description
Vertical Distribution
control (F=uniform,
T=Gaussian)
Chemical
transformation control
Dry deposition control
Wet removal control
Dry removal from
surface layer (T) or
throughout mixed layer
(F)
Recommended Value
T
T
T
T
T
Card Group 6 - OUTPUT OPTIONS
Columns
1-5
6-10
11-15
16-20
21-25
26-30
31-35
36-40
41-45
Typ*
L
L
I
L
I
I
L
L
L
Variable
Name
Isave
Iprint
iprint
Ibd
nnl
nn2
Iwetg
Iwetng
Idryg
Description
Disk/tape output
Print concentrations
Print interval in
hours of
concentrations
Print puff data (puff
height, ay, az,
location,
transformation rate,
etc .
Time step to begin
printing puff data
Time step to stop
printing puff data
Wet flux at gridded
receptors
Wet flux at non-
gridded receptors
Dry flux at gridded
recectors
Recommended Value
Generally true (T)
(allows post-
analysis)
Generally false (F)
(will usually want
some other averaging
times, so will use
results from post-
analysis)
(Used only if
lprint=T)
Generally false (F)
(can produce
voluminous output)
Generally 0
(if lbd=T, then
1 < nnl < nadvts)
Generally 0
(if lbd=T, then
nnl < nn2 < nadvts)
T if gridded
receptors used
T if non-gridded
receptors used
T if gridded
recectors used
A-22
-------
46-50
51-55
56-60
61-65
66-70
L
L
L
I
I
Idryng
Isavef
Iprflx
ires
iint
Dry flux at non-
gridded receptors
Save wet/dry fluxes
Print wet/dry fluxes
Save results for
restart?
Save results every
'iint' hours and the
last hour of the run
T if non-gridded
receptors used
T (allows for post-
processing)
F (can create
voluminous output)
0=no, l=yes
Generally 1
9999 saves only last
hour for restart
Card Group 7
Columns
1
2
3
4
5
6
Typ*
IAE
IAE
IAE
IAE
IAE
IAE
Variable
Name
iopts (1)
iopts (2)
iopts (3 )
iopts (4)
iopts (5)
iopts (6)
Description
Use default dispersion
parameters (ay, by, az,
bz, azt as defined by
Turner and Heffter,
Tm= 10000. jsup=5
(Tm reset in code from
original default value
of 100000)
Use default vertical
diffusivity constants
k, = 0.01, k,= 0.10
Use default S02 canopy
resistance
Use default dry
deposition parameters
Use default wet
removal parameters
Use default chemical
transformation methods
Recommended Value
0
0
0
0
0
0
Card Groups 8-13 - NEW VALUES TO REPLACE DEFAULT PARAMETERS
The default options are recommended for all regulatory uses of the
MESOPUFF-II modeling system. If the defaults are used, these card groups
do not need to be included. See the user's manual for further
information .
A-23
-------
Card Group 14 -
Columns
1-5
6-10
11-15
16-20
21-25
26-30
31-80
Typ*
RAE
RAE
RAE
R
R
R
RAE
Card Group 15 -
Columns
1-5
6-10
11-15
16-20
21-25
26-75
Typ*
RAE
RAE
RAE
RAE
RAE
RAE
POINT SOURCE DATA, 'npts1 cards required - one for each
point source.
Variable
Name
xstak
ystak
htstak
d
w
tstak
emis (1-5)
Description
X-coordinate of point
source (in
meteorological grid
units)
Y-coordinate of point
source (in met grid
units)
Stack height (m)
Stack diameter (m)
Stack exit velocity
(m/s)
Stack gas temperature
(°K)
Emission rates (g/s)
for S02, 30°, NOX,
HNO,, & NO;
Recommended Value
As Appropriate
As Appropriate
As Appropriate
As Appropriate
As Appropriate
As Appropriate
As Appropriate
AREA SOURCE DATA, 'nareas1 cards required - one for
each area source.
Variable
Name
xar
yar
htar
sigyar
sigzar
emisar
(1-5)
Description
X-coordinate of area
source center (in
meteorological grid
units)
Y-coordinate of area
source center (in met
grid units)
Effective height of
area source (m)
Initial ay (m) of area
source emissions
Initial az (m) of area
source emissions
Emission rates (g/s)
for S02, SOJ, NOX,
HNO,, & NO;
Recommended Value
As Appropriate
As Appropriate
As Appropriate
As Appropriate
As Appropriate
As Appropriate
A-24
-------
Card Group 16 -
Columns
1-10
11-20
Typ*
RAE
RAE
NON-GRIDDED RECEPTOR COORDINATES. ' nrec ' cards
required - one for each non-gridded receptor
Variable
Name
xrec
yrec
Description
X-coordinate of non-
gridded receptor (in
meteorological grid
units)
Y-coordinate of non-
gridded receptor (in
met grid units)
Recommended Value
As Appropriate
As Appropriate
The codes under type correspond to the following:
I Integer Variable
IA Integer Array
IAE Integer Array Element
R Real Variable
RA Real Array
RAE Real Array Element
C Character Variable
CA Character Array
L Logical Variable
A-25
-------
Appendix B
Method for Calculating Regional
Visibility Impairment
-------
METHOD FOR CALCULATING REGIONAL VISIBILITY
IMPAIRMENT
The primary sources of anthropogenically induced, regional
visibility degradation (also referred to as regional haze),
measured as light extinction, are fine particles (diameters <
2.5 urn) in the atmosphere. In the eastern U.S., these
anthropogenic particles are composed primarily of sulfate (S04)
compounds, organic compounds, and to a much lesser extent,
nitrate (N0~) compounds. These are important constituents in
other areas of the U.S. as well; their relative importance,
however, changes. For example, in some areas of the Pacific
Northwest, organic aerosols are as, or more, important than S04
aerosols. In some parts of Southern California, N03 aerosols
are the dominant specie. When examining individual source's or
groups of sources' impacts on regional visibility degradation,
primary emissions of fine particulate should also be
considered.
The generally observed sulfate compound is ammonium
sulfate {(NH4)2S04}; although ammonium bisulfate and un-
neutralized sulfuric acid particles have also been measured.
Particles composed of nitrate compounds usually take the form
of ammonium nitrate {NH4N03}. These compounds are generally not
directly emitted from air pollutant sources, but are formed
through a series of chemical reactions in the atmosphere. The
air pollutants, which contribute to the formation of these
particles, are gaseous emissions of oxides of sulfur and
nitrogen (SOX and NOX) , which eventually oxidize to form S04 and
nitric acid (HN03) , as well as other compounds, and
utlimately react with natural and anthropogenic emissions of
ammonia. The formation of NH4N03 is dependent on the
concentrations of ammonia gas (NH3) and nitric acid (HN03) as
well as the concentration of S04. S04 competes with HN03 for
the available NH3. Thus, in the presence of both S04 and HN03,
(NH4)2S04 will be formed preferentially to NH4N03. Essentially,
NH4N03 will only be formed when there is an excess of NH3
available, relative to S04. The MESOPUFF-II modeling system
accounts for the balance between S04, HN03 and NH3. Therefore,
emissions of both S02 and NOX should be modeled in the same run
to account for this balance. This balance will not be
accounted for if Level I methods are used; the contribution of
N03 to visibility degradation may be overestimated in Level I
analyses, but this is consistent with the rationale for Level I.
As noted above, fine particles are the major contributor
to anthropogenically produced visibility degradation, and
sulfates and organics constitute the highest contributions to
measured fine particle concentrations in most areas of the
country. Organic aerosols are generally considered to be
B-l
-------
secondary products of chemical reactions in the atmosphere; the
processes which lead to their formation are not well
understood. The sources of organic aerosols can be both
natural and anthropogenic. Current modeling and analysis
techniques are inadequate for providing an estimate of organic
aerosols.
Thus, for the purposes of calculating regional visibility
degradation due to specific sources of air pollution, the
primary focus will be on the contribution to light extinction
of fine particles of sulfate compounds and nitrate compounds,
expressed as (NH4)2S04 and NH4N03. Once these particles are
formed, however, their size can change, and thus their light
scattering efficiency, due to changes in the relative humidity
of the atmosphere. Therefore, in order to adequately account
for the contribution to light extinction of either (NH4)2S04 or
NH4N03 the mass of these constituents and the relative humidity
of the atmosphere in which these particles reside must be
known. The calculations of the extinction due to primary fine
particulates are assumed to be non-hygroscopic.
Method
1. Apply an appropriate air quality model to obtain hourly
concentrations of S04 and/or N03 and/or primary fine
particulate.
a. If using MESOPUFF-II concentrations of S04 and N03 are
obtained as direct model output (refer to
Appendix A).
1) To obtain primary fine particulate
concentrations, The MESOPUFF-II should be run as
an independent run from the S04 and N03 run,
assuming all of the fine particulate emissions
are S04 emissions and that they are the only
emissions in that run, the chemistry options
should be turned off, and the deposition options
should be turned on. The other options should
be set as described in Appendix A.
b. If running a steady-state model, use the methods,
outlined in Inset 1 of the main body of the report,
to convert SOX and NOX to S04 and N03. The primary
fine particulate emissions can be directly modeled.
2. As noted above, it is assumed that the compounds of
concern are (NH4)2S04 and NH4N03 and primary fine
particulate. Therefore the mass concentrations of S04 and
N03 must be corrected for the presence of NH4. (The
primary fine particulate is modeled as S04, as a surrogate
in MESOPUFF-II; this should not be corrected for NH4.)
a. Multiply the mass concentration of S04 by 1.375 to
obtain (NH4)2S04
B-2
-------
c.
d.
e.
Multiply the mass concentration of N03 by 1.29 to
obtain NH4N031. Obtain the hourly values of
relative humidity, appropriate
for each receptor and
corresponding to the hourly
concentrations calculated in step
1.
These can be obtained from the MESOPUFF-II
meteorological files.
If hourly relative humidity values are not available,
assume that the relative humidity is 95%.
Obtain the relative humidity correction factor
(f(RH)) from Figure B-l (see also Table B-l).
3. Calculate the extinction based on the following equation.
bext. s=0. 003 [part] f(RH)
where '
bext s = The extinction coefficient due to particle scattering (km'
0.003 = a nominal dry scattering efficiency
[part] = The concentration of (NH4) 2S04 or NH4N03 in }ig/m3
or of primary particulate expressed as SO (]ig/m3)
f(RH) = The RH correction factor (see figure B-l)
(The dry efficiency is a consensus value based on Trijonis et
al. (1987) and the relative humidity correction factor is based
on Tang et al. (1981) .)
a. It is only appropriate to compute the extinctions
based on hourly values of [part] and relative
humidity. It is not appropriate to use average
values of these quantities.
b. To calculate the extinction due to primary fine
particulate, use the above equation, but set the
relative humidity correction factor (f(RH)) equal to
1.
Example
If one has a sulfate concentration [S04] of 1.7 ug/m3, then
this would correspond to an ammonium sulfate concentration
[(NH4)2SOJ of 2.34 ug/m3. The extinction due to this
concentration would be 0 . 003x2 . 34x.f (RH) km"1. From Figure B-l,
if the relative humidity is 95%, f(RH) is 11.5. Therefore
b_, = 0.08 km"1.
B-3
-------
Relative Humidity Factor
17
16
15
14
13
12
11
10
9
8
6
0 10 20 30 40 50 60 70 80 90 100
Relative Humidity (%)
Figure B-l - Correction factor to adjust for the effects of
relative humidity on light extinction calculations (Tang et
al., 1981).
B-4
-------
TABLE B-l. Relative humidity factor values versus humidity
values used to construct Figure B-l.
Relative
Humidity
0.0
30
40
50
60
67
70
75
80
92
98
Relative
Humidity
Factor
1
1
1.2
1.35
1.65
1.95
2.3
2.6
3.5
6.5
16
References
Tang, I.N., W.T. Wong, and H.R. Munkelwitz. 1981. The
Relative Importance of Atmospheric Sulfates and Nitrates
in Visibility Reduction. Atmospheric Environment,
15(12):2463-2471.
Trijonis, J.C., M. Pitchford, and M. McGown. 1987.
Preliminary Extinction Budget Results from the RESOLVE
program. In: Visibility Protection Research and Policy
Aspects. P.S. Bhardwaja, ed., Air Pollution Control
Assoc. (Currently Air and Waste Management Assoc.),
Pittsburgh, PA, pp. 872-880.
B-5
------- |