United States
Environmental Protection
Agency
EPA-600/9-86/C
July 1986
Research and Development
U-S. EPA Workshop to Develop
A Simple Model for
Acid Deposition
Final Report f
U.S. Environmental Protection Agency
Office of Research and Development
Atmospheric Sciences Research Laboratory
Research Triangle Park, NC 27711
-------
EPA-600/9-86/0
July 1986
U.S. EPA Workshop to
Develop a Simple Model
for Acid Deposition
Final Report
Professor Harvey Jeffries, Workshop Chairman
Department of Environmental Sciences and Engineering
University of North Carolina
Chapel Hill, NC 27514
Contract No. 68-02-3839
Project Officer
Basil Dimitriades
Atmospheric Sciences Research Laboratory
Research Triangle Park, NC 27711
Atmospheric Research Sciences Laboratory
Office of Research and Development
U.S. Environmental Protection Agency
Research Triangle Park, NC 27711
-------
Disclaimer
Although the workshop described in this report has been funded wholly by the U.S. Environmental
Protection Agency under Contract 69-02-3839 to Northrop Services, Inc., the report has not been
subjected to the Agency's required peer and policy review and therefore does not necessarily reflect
the views of the Agency and no official endorsement should be inferred. Mention of trade names or
commercial products does not constitute endorsement or recommendation for use.
-------
Abstract
The Acid Deposition Planning Staff in the Office of Acid Deposition/EPA requested that the
Atmospheric Sciences Research Laboratory undertake the development of a low-computational-
demanding model suitable for educational use in understanding the linear or nonlinear nature of
the acid deposition process and that could assist in constructing a "what if" analysis for a particular
situation. In response, ASRL held a workshop, attended by EPA and non-EPA scientists, to discuss
the design of such a model. In addition, a five-person working committee of non-EPA scientists was
established to write a statement of work for the construction of such a model.
This report summarizes the planning activities that occurred before the workshop, the two
days of discussion held at the workshop, and the three days of working committee meetings after
the workshop, and includes the resulting statement of work for the creation of a Comprehensive
Chemistry Acid Deposition Model (CCADM).
As described in the statement of work, the CCADM should be capable of estimating changes in
acid deposition downwind at a receptor site as a function of changes in emissions in a source region.
The model would estimate these changes along a fixed trajectory for a fixed set of meteorology-
related variables. The model would use the most advanced gas- and aqueous-phase chemistry kinetics
modules and an appropriate parameterization of cloud physics. Output from more complex regional-
scale Eulerian models could be used as input to establish the context for a CCADM simulation. In
addition, the model would include appropriate graphical displays to illustrate and explain acid
deposition sensitivity to changes in emissions.
The model is not designed to produce quantitative assessments of acid deposition.
111
-------
IV
-------
Contents
1 Introduction 1
Origin of Workshop 1
Workshop Approach 2
Organization of Workshop 4
2 Need and Purpose 5
Existing or Planned Acid Deposition Models 5
Regional Acid Deposition Model (RADM) 5
NCAR Engineering Models 6
Need for A Simple Model 6
Model Objectives 7
Concept of Hierarchical Application of Models 7
Scenario for Assessment 9
What Will Be Calculated by the Simple Model? 9
Expected Use of Model Results 10
Ultimate Products or Actions Anticipated 10
Resource and Time Constraints for the Simple Model 11
3 Model Approach 12
Model Components: Approaches and Problems 12
Chemistry 12
Source-Receptor Relationship 14
Mixing 16
Clouds and Storms 16
Emissions 18
Boundary Conditions 21
Display of Output 21
Approach to the Simple Model . . . : 22
-------
4 Refinement of Approach 24
Discussion of Problems 24
Summary and Conclusions 26
5 Working Committee Meeting 27
Evolution of the Simple Model Description 27
Inclusion of Important Factors 28
6 Statement of Work 31
Title 31
Period of Performance 31
Background 31
Purpose 32
Specifications of Model 33
Conceptual Framework 33
Processes To Be Represented in Model 34
Entrainment and Dilution Processes 36
Boundary Processes 36
Model Inputs 36
Model Outputs 37
Statement of Work 38
Task 1-Develop and Submit Detailed Work Plan 38
Task 2-Develop and Test Major Model Components 38
Task 3-Integrate Model Components and Develop Input Processor 38
Task 4-Exercise Model 38
Task 5-Assess Feasibility of Deriving Long-term Deposition Estimates 39
Task 6-Develop Display Technique 39
Task 7-Propose Procedures for Testing and Evaluating Model 40
Task 8-Documentation of Model 40
Technical Proposal Instructions 41
Minimum Qualifications 41
Key Personnel 41
Computer Programs 41
Computer Programming Requirements 41
vi
-------
7 Appendix 42
A. Discussion Leaders and Topics 42
B. List of Participants 47
VII
-------
Acknowledgments
The workshop planners appreciate the time and effort given by the participants and presenters
at this workshop. Their contribution made the workshop possible.
This document was written primarily by Harvey Jeffries with significant help from many sources.
In producing the meeting background material, Basil Dimitriades, Dennis Trout, Marcia Dodge, Jack
Durham and John Clarke, all EPA personnel, participated in several cycles of discussion, review,
and revisions of draft material. In addition to Jeffries, the working committee that produced the
statement of work consisted of Akula Venkatram, Environmental Research and Technology; Chris
Walcek, National Center for Atmospheric Research; Leonard Peters, University of Kentucky; and
Hiromi Niki, Ford Motor Co. We are very grateful to the members of this working committee for
lending their special expertise and time to this project. In writing the middle chapters of this report,
Jeffries used tape recordings of the meeting provided by Barry Mangum, Northrop Services, Inc., as
well as extensive notes taken by Marcia Dodge during the meeting. These items were necessary for
the accuracy and completeness of the report and we appreciate their availability.
Ms. Linda Cooper, Northrop Services, Inc., made all arrangements for the meeting and with the
attendees. Her editorial assistance in producing this report was essential and is greatly appreciated.
Vlll
-------
Introduction
Origin of Workshop
The Acid Precipitation Act of 1980 (P.L. 96-294) calls for a comprehensive and
coordinated program to identify causes and effects of acid precipitation, as well
as actions needed to control harmful effects. An Interagency Task Force on Acid
Precipitation was established in 1980 to manage this program, with members from
12 Federal agencies, including EPA as a co-chair member. The Task Force has
10 working-level groups, one of which is responsible for Assessment and Policy
Analysis research activities. The responsibility of this group includes development
of periodic assessments in which available scientific information is focused on policy
issues. The information is examined for gaps and uncertainties to identify actions
and new research needed. Of crucial importance in such assessments are the avail-
ability and use of source-receptor models, that is, models that predict changes in
acid deposition from changes in precursor emissions.
Development of source-receptor models for such assessment uses is the respon-
sibility of EPA, and within EPA, of the Atmospheric Sciences Research Laboratory
(ASRL) in the Office of Acid Deposition, Environmental Monitoring, and Quality
Assurance (OAD). There is a long history of process and receptor model develop-
ment activities within ASRL that include
• an urban and regional scale three-dimensional Eulerian-type physicochemical
model for O3;
• a simpler, nomogram-type chemical mechanistic model for urban Oa (OZIPM/
EKMA);
• a comprehensive, regional, three-dimensional Eulerian model for acid deposi-
tion, and
-------
Introduction . Workshop Approach
• an "engineering" model, a simpler three-dimensional Eulerian model for acid
deposition.
While the latter two models are intended and expected to fulfill the basic needs of
the assessment effort, the EPA-OAD Headquarters staff perceived an urgent need
for a simpler, nomograph-type model that would assist in further understanding
potential state-level emissions trade-offs for source-receptor relationships in acid
deposition assessment issues.
On recommendation from Headquarters, therefore, ASRL undertook the re-
sponsibility for completing, by the fall of 1986, an effort that would lead to the
development of such a model. This ASRL effort consisted of two tasks. The first,
under the direction of Dr. Basil Dimitriades, was to organize and conduct a work-
shop. In this workshop, alternative development approaches would be identified
and discussed with the objective of recommending an approach and producing a
scope of work for its implementation. The second ASRL task, under the direction
of Dr. Marcia Dodge, was to contract the subsequent development effort for the
model.
Workshop Approach
The workshop planning was primarily the responsibility of Dr. Jeffries, a consultant
to Northrop Services, Inc., working in conjunction with Dr. Dimitriades, ASRL staff,
and members of the Acid Deposition Planning Staff (ADPS) in the Office of Acid
Deposition in Washington, D.C. In planning the meeting, Dr. Jeffries chose to follow
an approach similar to that in EPA's Data Quality Objectives program, (see EPA
Memorandum, 1984, "The Development of Data Quality Objectives") but to adapt
these so that they were applicable to the production of Model Quality Objectives.
Model Quality Objectives are qualitative and quantitative statements of the
quality of model output needed to support specific decisions or regulatory actions.
They are the mechanism for balancing the conflicting demands of time and resource
constraints and the need for the model to have acceptable quality and utility.
The objectives development process consisted of a series of exchanges among
headquarters decision makers and technical staff to reach an understanding of what
was needed and why. In the idealized situation, this process would consist of three
iterated phases:
1) the decision maker supplies initial information, from his perspective, in as much
detail as possible, about what is needed and why;
-------
Workshop Approach Introduction
2) the technical staff refines the initial concepts through discussion among staff
and the decision maker to reach a common understanding of the problem and
its constraints; and
3) various approaches are proposed and evaluated for meeting the necessary func-
tions within the constraints.
We began the development process with a verbal explanation of what was de-
sired by ADPS staff. After discussion among ASRL staff, a draft statement was
prepared and circulated to all parties. Using a series of written exchanges and dis-
cussions, the basic requirements for the model were established. In the latter stages
of this process, members of the scientific modeling community were contacted and
asked to participate actively in the meeting planning process, and four individuals
were selected to form, with the meeting chairman, the working committee. This
committee had the primary responsibility for creating the main workshop product,
a statement of work for the development of the model.
The four members of the working committee assisting Dr. Jeffries were
• Dr. Leonard Peters, Department of Chemical Engineering, University of Ken-
tucky, coauthor of the STEM-II, a sulfur transport Eulerian model;
• Dr. Chris Walcek, National Center for Atmospheric Research, a member of the
Regional Acid Deposition Model development team;
• Dr. Akula Venkatram, Environmental Research and Technology, a member of
the ERT team developing ADOM, the Acid Deposition Oxidant Model, for use
in Canadian assessment studies; and
• Dr. Hiromi Niki, Ford Motor Company, a chemistry and kinetics specialist.
-------
Introduction Organization of Workshop
Organization of Workshop
The workshop was divided into eight 1.5-hour sessions, and each session had one
or more discussion leaders. These discussion leaders supplied background informa-
tion and introduced proposals, hypotheses, and questions. The leaders facilitated
discussion and information exchange among participants and assisted the working
committee in introducing information needed to develop a statement of work. The
workshop was designed so that, initially, researchers active in the acid precipitation
field would present material about existing models and about alternative approaches
that might be acceptable for meeting the needs expressed by the model quality ob-
jectives. Later in the meeting, the various approaches were debated and discussed.
Finally, the strengths and weaknesses of each approach were discussed.
The two-day workshop was followed by a three-day meeting of the working
committee. In these meetings, an approach was settled upon and a statement of
work for its development was produced.
-------
Need and Purpose
This chapter provides the initial background and setting in which the workshop oc-
curred. It presents material concerning the model objectives process as discussed by
the Atmospheric Sciences Research Laboratory staff and Acid Deposition Planning
Staff before the meeting. This material was available to all participants before the
workshop.
Existing or Planned Acid Deposition Models
Regional Acid Deposition Model (RADM)
This National Center for Atmospheric Research (NCAR) model is
"an Eulerian regional acid deposition model suitable for assessing source-
receptor relationships. It will be a modeling framework that integrates a
mesoscale meteorological model with a transport and chemical transforma-
tion model. All components of this model system are expected to include
state-of-the-art representations of the relevant physical and chemical pro-
cesses. This model, with its comprehensive description of all the currently
recognized major physical and chemical processes, makes RADM a suitable
tool for research and a scientifically defensible application model. For policy
and assessment studies, however, the RADM would need to be streamlined for
computational efficiency and to have the representation of the physical and
chemical details suitably simplified. A synthesis of the comprehensive RADM
modeling result would be required ... to construct a set of fast-turnaround
"engineering" models for application."
— from Jan. 1985 NCAR Status Report on the First-Generation
Eulerian Acid Deposition Model
-------
Need and Purpose NCAR Engineering Models
RADM is an episodic model—12 synoptic cases have been selected to represent
the annual variations—and statistical techniques are being investigated to construct
the annual and seasonal deposition.
NCAR Engineering Models
These "engineering" models are simpler, but still three-dimensional Eulerian mod-
els. They are presently described as lower resolution models derived from the RADM
by simplification and parameterization of the physical and chemical processes im-
portant in acid deposition. These EMs will still be large models, having 3 to 5
vertical levels and 160 to 320-km grids, and requiring a small mainframe or large
supermini computer for their solution. These models require a series of technical
studies before development, including such subjects as parameterization of the non-
linear chemistry; evaluation of computational costs; evaluation of suitable databases
for initialization, execution, and evaluation; and testing of statistical ensemble pro-
cedures for generation of seasonal and annual averaged deposition.
Need for A Simple Model
The RADM Eulerian model is expected to provide scientifically credible answers to
many source-receptor relationship issues such as the degree to which the entire sys-
tem is "linear" or "nonlinear," and, in selected scenarios, the response of deposition
to changes in emissions.
The engineering models, which are closely coupled in development to the full
RADM, have simplified representations of many nonlinear physical and chemical pro-
cesses and thus will not require the large computer resources of the RADM. Many
more assessment strategies can be explored with these models than with the more
comprehensive, and presumably more correct, RADM model.
Both of these models will be used by EPA to make assessments for sulfur emis-
sions reduction on a state-wide basis. The ADPS is concerned, however, for the
time and space scales "where things are linear or nonlinear" and where the states
may have trade-off options among VOC, NOX, and SO2 emissions. Understanding how
these options work and exploring possible alternatives are highly desirable objectives
from the viewpoint of the decision maker.
Thus, a tool is needed that would allow a decision maker to explore "what if"
and trade-off options. This tool is not seen as an alternative to RADM or engineering
models, rather, these options would be variations upon the reductions of sulfur
emissions that were predicted by these larger regional-scale models. The basic goal
-------
Model Objectives
Need and Purpose
Peak
Weekly
SO2
Emission
Isoiines of Total S Deposition
to the Adirondack*
20 40 60 80 kg/ha/yr
PeakDaiiyO3fromEKMA
HYPOTHETICAL NO MO GRAM FOR A STATE
SO2/NOX/VOC TRADE OFFS
LOCAL OR LONG-RANGE
SCREENING MODELS
Figure 1.
would be to investigate linearities or nonlinearities in the chemistry, that is, to
explore the coupling among NOX, VOC, Os, and SOX chemistry. Ideally, relationships
might be portrayed in an EKMA-type nomogram method: one example supplied
by the ADPS is shown in Figure 1. The simplicity and intuitive understanding
associated with such a pictorial representation of the options is considered highly
desirable.
Model Objectives
In pre-meeting discussion among EPA staff and the workshop chairman, a concept
of how these different models would be used was described. This concept was further
refined into a set of objectives and expected uses for the simple model.
-------
Need and Purpose Concept of Hierarchical Application of Models
Concept of Hierarchical Application of Models
Each of the models was seen as having a role and scope in its application. The role
and scope of RADM follow:
> to provide complete, state-of-the-art treatment for 12 synoptic events;
o to be a relatively high resolution regional model;
> to have intermediately complex chemistry;
> to be very resource intensive;
> to calculate seasonal and annual averaged values with statistical ensemble tech-
niques;
> to be used to verify assessment strategies derived from lower level models; and
> to provide a foundation for simpler engineering models.
The role and scope of the engineering models follow:
> to provide faster computation on smaller machines, allowing more strategies to
be tested;
> to be low resolution regional models;
> to have parameterized chemistry;
> to be derived from and checked by comparison to full RADM; and
> to be the primary tools used to calculate state assessments.
The role and scope of the simple model follow:
> to provide variations of a single state's impact on receptor site for a given
episode;
> to most likely require complex chemistry vs. transport because it must be
capable of demonstrating any nonlinearities in local options;
> to begin with regional model's predictions of needed S-deposition;
> to provide fast computation on generally available machines;
> to provide effects of local control on changes in deposition.
-------
What Will Be Calculated by the Simple Model? Need and Purpose
Scenario for Assessment
To understand better how these three models might work together, a scenario for a
potential assessment was developed for discussion. This scenario has three steps:
• Step 1: an engineering model would be used to calculate regional sulfur depo-
sition. As currently envisioned this would consist of
> making calculations for 12 synoptic episode conditions;
> computing seasonal deposition by statistical ensemble procedure;
> performing cause-and-effect analysis of deposition by
o analyzing source-receptor relationships for high deposition cases,
o identifying "worst-case" contributions of sources to key sites, and
o describing the processes by which each state contributes to sites;
c> calculating reductions for each state to achieve key site deposition goals;
i> conducting, and iterating if needed, reduced conditions in the engineering
models;
> assigning needed sulfur reductions to states.
• Step 2: the simple model would be used to calculate local emissions trade-offs.
In this calculation
> each key site that would be impacted by state would be treated separately;
> conditions would be those giving "worst-case" state contribution;
> other states would be assumed to meet their reduction goals; and
> the ultimate product would be a set of alternative emissions for VOC, NOX,
SO2.
• Step 3: the engineering model and/or the RADM model would be used to verify
alternatives. In this step:
> options for VOC, NOX, and S(>2 reductions would be input to the regional
model;
> a new set of calculations would be performed with the RADM or engineering
models;
> estimated local reductions would be adjusted, if needed, and new values
would be calculated.
What Will Be Calculated by the Simple Model?
The model will calculate variations on a particular relationship between state emis-
sions and receptor site. In these calculations, the following information would be
-------
Need and Purpose Expected Use of Model Results
obtained:
• The context of the source-receptor relationship would be supplied by regional
model data. The initial source-receptor relationship would be supplied to the
simple model by the "baseline" run of the regional model. In calculating the
possible trade-offs available to a geographical subregion, it would be assumed
that the other states or regions that impacted on the site in the same "worst
case" episode would reduce their S-emissions as predicted by the EPA assess-
ment.
All other factors, however, would be the same as the "worst case" baseline run
of the regional model. That is, the VOC emissions, NOX emissions, precipitat-
ing and non-precipitating cloud conditions, etc., along the path between the
state and the receptor site would be exactly the same as those in the engineer-
ing or RADM model "worst-case" episode conditions; the S emissions along the
path and outside the state, however, would be reduced according to the EPA
assessment for that state.
• The model will provide quantitative predictions of total S-deposition at recep-
tor. These would be produced as a function of VOC, NOX, and S emissions.
Ideally, the model would also provide other items such as total acidit, SO2 and
Os concentration, nitrate deposition, and ammonia.
• The model would provide a graphical display illustrating effects of emissions
changes.
Expected Use of Model Results
The first use of the model results is expected to be as an educational tool to illustrate
the extent to which source-receptor relationships may be "nonlinear" and the extend
to which "emissions trade-off" might be possible. A later use of the model might
be to provide regulatory assistance in which a decision maker would perform "what
if" analysis for individual states or regions. Eventually, depending upon its utility,
the model might be to explore options for trade-offs available to a given state.
Ultimate Products or Actions Anticipated
By exploring factors affecting S deposition at key target receptors for each sub-
regional area, the ADPS expects that decision makers would gain a significantly
increased understanding of the acid deposition problem.
In addition, because there might be many options available for reducing S de-
position impact at a distance site, and because the Federal assessment will only be
at a whole-state level, tools are needed to assist in fine tuning the control options.
10
-------
Resource and Time Constraints for the Simple Model Need and Purpose
The ultimate products or actions, therefore, are expected to be an improved
understanding of the problem and an evaluation of the control options needed,
taking into account the individual situations in each subregion under analysis.
Resource and Time Constraints for the Simple Model
There are important resource and time constraints for this model. In routine appli-
cations of the model, it should be able to perform 4 to 10 assessments per day with
total computer costs for each assessment being between $50 and $200. Constraints
in the development of the model are the necessary completion of full documention
and peer review by November 1986. The total cost for developing the model should
be less than $240,000.
11
-------
Model Approach
Model Components: Approaches and Problems
In this portion of the meeting, discussion leaders presented information about model
components, describing the factors that would need to be treated in various ap-
proaches and discussing the problems likely to arise in each approach.
Chemistry
Dr. Allen Lloyd, ERT, led a discussion of the chemistry component. Jeffries asked
Lloyd, "Is the chemistry understood well enough to develop a chemistry component
for this model?"
Lloyd responded that gas- and aqueous-phase chemistry are very complex prob-
lems and may not be well represented by simple models. He listed areas of gas-
and aqueous-phase chemistry that may present potential problems in constructing
chemical models (see Table 1). Even the chemistry models in RADM may be quite
simple compared to the complexities of the actual chemistry. He stated that the
only method for assessing the accuracy of a model was to test it against real-world
data.
In the case of radical initiation, Lloyd said,. "Although 03 work shows that
we have a reasonable understanding of these processes, experimental data, under
careful controlled conditions, do not give us very much faith that we fully know what
is going on." A similar statement could be made for radical termination processes,
which are represented only approximately in many mechanisms for Os. This is
because, until now, we have not been concerned about predicting the products that
arise from these termination processes. The major source of hydrogen peroxide
(Hz02) is a termination reaction of HO2 radicals. Lloyd believes that the ability of
mechanisms to predict H2<>2 concentrations must be tested with real-world data.
12
-------
Chemistry Model Approach
Table 1. Potential Problem Areas In Chemical Models.
gas-phase aqueous-phase
radical initiation role of H2(>2, 03, trace metals
radical termination free radical chemistry
nighttime chemistry lifetime of droplet
biogenic HC liquid water content
slowly reacting species mass transfer
VOC, NOX, SOZ temperature-rates
relationships and solubilities
temperature and
pressure
Nighttime chemistry also has not been of much concern because of the typical
one-day O3 applications of the models, but for multi-day simulations an adequate
representation of nighttime chemistry will be vital. The role of biogenic HC has
been seen as small for one-day O3 predictions. On a regional scale, however, these
sources may be very important in producing H202 or organic peroxides. Thus, they
could be very important regional sources of oxidation. Precise information is not
known about these processes.
For the most part, many slowly reacting species have been neglected for the O3
problem and these can be very important in a regional oxidation scheme. These
also may not be well represented in the current chemical mechanisms.
With regard to VOC, NOX, and SOj; relationships, the VOC-NOX-O3 relationships
are again reasonably well understood, but the relationship of other oxidants to
SOZ is not as well denned. There may be major uncertainties that will make the
development of trade-off strategies difficult.
Finally, for the gas-phase, pressure and temperature effects vary as a function
of altitude, and, for the multi-layer models, may represent potential difficulties.
In the area of aqueous-phase chemistry, Lloyd said, "Many in this room think
that these are completely understood—this is not the case. There are very few
experimental data to check the assumptions."
The relative roles of H2C»2, O3, and trace metals are still undergoing analysis
and simulation. 03 itself, or free radicals for which 03 may be a surrogate, are only
13
-------
Model Approach Source-Receptor Relationship
partially understood. Radical sticking coefficients are needed. Models have to use
profiles and assumptions to represent lifetimes of droplets and liquid water content
of clouds.
Data are lacking for many parameters such as gas/aqueous-phase mass transfer
and the effect of temperature on solubility and reaction rates.
Clearly, the aqueous-phase chemical component of models must be designed for
change. A protocol for testing and providing quality assurance (QA) for rapidly
developing chemical submodels needs to be developed.
We need a whole range of sensitivity studies to determine effects of uncertainties
in the aqueous chemistry components. In addition, mechanism comparison studies
are needed. If different mechanisms are used in each of the models and these
mechanisms have different sensitivities, how do we know that, in the same scenario,
we are going to come up with the same results. A testing protocol and QA guidance
are needed to minimize such problems.
Other sources of potential differences in the chemistry are dry deposition veloc-
ities, particle size distributions, and biogenic emissions.
Finally, is there a "best approach" to mechanism condensation, and what will
be the method used in RADM?
Lloyd concluded by saying, "The tool you design must be able to accommodate
change."
Source-Receptor Relationship
Dr. Akula Venkatram, ERT, led a discussion about establishing a source-receptor
relationship. Jeffries asked, "Is there a source-receptor relationship that is quantit-
fiable, and if you had a model that gave you such a relationship, how would you
use it?"
Venkatram said that if you defined a source-receptor relationship in an oper-
ational manner as "How much does the receptor concentration decrease when the
source is removed?", then the concept does not make much sense in a nonlinear sys-
tem. He argued that in a nonlinear system, deposition may not be very dependent
upon S(>2 source strength, but might be oxidant limited. Thus, in an experiment
with the ADOM model, when SOj emissions were reduced uniformly by 50%, land
regions only saw a 10% decrease in wet sulfate (SO^) deposition. Far from the
14
-------
Source-Receptor Relationship Model Approach
sources, over the ocean, there was a 50% decrease in wet SO^ deposition. The rea-
son given was that the system was oxidant limited, not sulfur limited, and there
was excess S(>2 to start.
Workshop participants objected to the limited definition of source-receptor re-
lationships proposed and suggested instead that any source-receptor relationship
would be highly context sensitive. The desired source-receptor relationship would
be determined by "How many molecules of S deposited at the receptor site came
from the source?" In the example used by Venkatram, one source-receptor rela-
tionship affected other source-receptor relationships. That is, removing S from one
source increased the source-receptor relationships for other sources. This is because
oxidizing species that were used by the original source were, in its absence, available
for use on other sources, and the system was limited by the availability of these ox-
idizing species. Thus, any particular source-receptor relationship will be influenced
by interactions of other sources along the way to the receptor. This is why Eulerian
models are used: they are best able to represent this source-source interaction.
Jeffries asked, "Can the engineering model be used to determine source-receptor
relationships?" Venkatram replied that if it was done by removing sources and
examining effects at receptors, then no. Walcek said that NCAR was "assuming
that the engineering model can apportion sources and receptors at each point in the
current model framework' and that "we are assuming that we can define a worse
case." Walcek went on, "This apportionment will require many repeated runs of
the engineering model to determine the effect of a given source on a given receptor.
That is, just because you know that Ohio was contributing 20% of the sulfur to New
York, does not tell you what reducing emissions in Ohio will do to reduce sulfur
deposition in New York. The model must be run over and over."
Lamb pointed out that what was desired was knowing that 20% of the sulfur
atoms in New York came from Ohio, not that changing some source resulted in a
20% change in sulfur deposition in New York. Venkatram said, "But knowing that
20% of the sulfur comes from Ohio is not very useful for a control decision. This is
because removing Ohio's sulfur emissions may not change the sulfur deposition in
New York at all because the transformation process may be highly nonlinear!"
Walcek again stated that many runs of the engineering model would be required
to determine needed controls.
Discussion suggested that the parameterization of the chemistry in the engi-
neering model would have to mimic any nonlinearities that occurred in the use of
15
-------
Model Approach Mixing
the full RADM. There were concerns that this might be a very difficult task. Clarke
explained that there are several engineering models envisioned, each designed to
work for a different range—there could be 8-10 engineering models. In any case,
these models are the tools EPA will used to perform apportionment.
Mixing
Dr. Jack Shannon, ANL, led a discussion on the mixing process in models.
Mixing is caused by turbulence and this is represented differently in all models.
The treatments range from almost no mixing to instantaneous mixing. Mixing will
be very simple no matter which model approach is used.
Sub-grid processes are very difficult to represent and are very important from a
chemical viewpoint. Lamb said that, in the Regional Oxidant Model, the results are
entirely different depending upon whether 18-km or 80-km grids are used. (RADM
will use 80-km grids.) For example, Shannon said that, in the real world, a point
source of SC>2 and a point source of VOC might never meet, but in the model they
will meet if they are in the same grid, as illustrated in Figure 2 from a 1983 NCAR
report.
A second mixing problem is with representing the effects of vertical mixing and
vertical shear. Walcek said that tests on the meteorology model in RADM showed
that vertical mixing was critical to the correct calculation of multi-day trajectories.
Clouds and Storms
Dr. Chris Walcek, NCAR, led the discussion on cloud and storm processes.
Walcek said that clouds have four important effects on regional processes:
1) they are very effective vertical mixers of pollutants,
2) they have important radiative effects for photochemical reactions,
3) they provide an aqueous chemical reaction system, and
4) they are effective at scavenging pollutants.
The spatial scale of processes for clouds covers about seven orders of magnitude
from microns for droplets to several kilometers for whole clouds. The effects of
clouds can be nonlinear.
In describing clouds as vertical mixers of pollutants, Walcek said that the up-
draft velocities in clouds are several orders of magnitude larger than free air ve-
locities, thus clouds can mix emissions (lower layers) with oxidants (upper layers).
16
-------
Clouds and Storms
Model Approach
•100 km-
100 km True Reaction =» 0
Computer Simulation
of Reaction
True Reaction *
Simulated Reaction
Figure 2. Effect of Grid Size on Chemical Reactions (ADMP, NCAR, 1983).
17
-------
Model Approach Emissions
In the aloft layers, horizontal transport occurs quickly compared to the boundary
layer.
Clouds represent a sizable volume of water and thus they can provide an aqueous
chemical reactor system. Figure 3 is a summary of the RADM cloud processes routine.
Figure 4 is an example calculation of the effect of an individual convective cloud on
the distribution of sulfur. It represents one hour of processing of a 6-km deep cloud
that covered about 10% of a grid cell area. The numbers attached to the boxes are
percent of total sulfur originally below the cloud base. About 25% of the sulfur was
processed by the cloud, about 5% was deposited, and 10% was vented above the
mixed layer.
In the aqueous chemistry in the cloud, the production of SO^ is limited by
the oxidant present. Because of oxidant limitations, there can be a very nonlinear
relationship between sulfur and oxidant concentration.
At present, most of the cloud physics must be highly simplified or parameterized.
For example, cloud vertical velocities may have to be specified based on rainfall rate
being slow, medium, and fast. These models generally ignore any feedback effects
between the cloud chemistry and the gas chemistry (other than photochemical rate
changes).
Although simplifications of cloud processes may be possible, Peters pointed out
that, in the VENTEX experiment in 1984, the conversion rates in very similar
clouds varied from 2 to 36% per hour.
Emissions
Dr. Leonard Peters, University of Kentucky, led the discussion on emissions infor-
mation for the simple model.
Peters stated that, based on the desired application of the simple model—to
examine potential emissions trade-offs, this model may need more detailed emissions
data than RADM, and on a smaller grid than 80 x 80 km. Although we can identify
major individual sources, there are many smaller and area plume sources that will
be important. Plume interactions become more important, and diurnal variability
of emissions is more important for the application of this model.
Lamb described model experiments with the Regional Oxidant Model, which
had an urban plume only 10 km wide. In the two-day simulation, a reduction of
32% in HC and 8% in NOX emissions had a much greater effect on the peak O3 than
18
-------
Emissions
Model Approach
Inputs (80 km resolution):
- Rainfall rate (ram/hr)
- Vertical profiles of:
- Temperature (K)
- Water vapor mixing ratio (g/g)
- Pollutant mixing ratio (g/g)
Cloud processes routine:
I
Outputs:
- Compute cloud base, cloud top,
and cloud area coverage.
- Compute cloud vertical
structure for:
- Pollutants
- Temperature
- Liquid water
- Run aqueous chemistry model
- Compute rainout amounts for
each pollutant
- Compute Mixed, scavenged and
grid averaged pollutant profiles
I
DC/DT for each pollutant
DC = DCmix f
DCreaction +
DCremove
Wet deposition amount
for each pollutant
Figure 3. Summary of Cloud Processes in RADM.
19
-------
Model Approach
Emissions
16.5
Figure 4. Effect of Cloud Processes on Sulfur Distribution.
20
-------
Display of Output Model Approach
on the area wide Os average. So, Lamb suggested that, because of the nonlinearity
of the chemistry and because large grids smear oxidants out, the grid resolution
must be about the size of the urban plumes in the areas under study.
Boundary Conditions
Dr. Venkatram also led the discussion on boundary condition issues for the simple
model.
In the pre-meeting material, it was stated that the meteorological and boundary
conditions would be supplied by a model external to the simple model, such as the
RADM or engineering models. Venkatram saw potential problems in this approach.
He questioned whether a box model would work on the scales needed in the simple
model. He felt that a critical factor would be the entrainment of material through
the sides and top of a simple box model. A good representation must consider
vertical motion and wind shear. He felt that to get the initial conditions for the
control simulations with a simple box model, the engineering model would have
to be run in several specific situations, adopting a state-by-state emissions control,
resulting in hopeless combinations of conditions to even approximate the necessary
boundary conditions.
Johnson questioned why the simple model had to tie to RADM under any cir-
cumstances. Why not have a total stand-alone model? Jeffries replied that the
simple model had to give up something to be simple, and if the same emissions,
wind fields, and other operational conditions of RADM were not used in the simple
model, then extraneous causes of differences in predictions may be introduced. A
complex model would be needed to predict transport correctly.
Display of Output
The original request from the ADPS in Washington suggested that some type of
"nomograph" display technique be used to illustrate the possible control options
(see Figure 1).
Although not seen as directly applicable to this simple model, the manner in
which the Os isopleth diagram is produced was thought to be of interest. Dr. Gary
Whitten, SAI, described how multiple simulations—about 65 individual simulations
along the same trajectory—are used to produce an 03 isopleth diagram.
Although the concept of some type.of a graphical display for acid deposition was
appealing, no one could describe what would be displayed as a function of what.
21
-------
Model Approach Approach to the Simple Model
Would SO^ deposition at a single site be displayed for variations in source VOC and
NOX emissions at constant S(>2 emissions? How would the deposition that occurred
in this episode be related to the annual acid deposition at the receptor site?
Approach to the Simple Model
Jeffries presented some observations about likely model approaches based upon pre-
meeting discussion. He offered the following observations:
• The simple model most likely will be a derivative of some existing model-
ing approach. That is, because of time and money constraints, the model
will be some form of a statistical-based approach or a process-based approach.
Empirical-based approaches most likely are not acceptable because of the lim-
ited data on which to base such an approach. Statistical-based approaches
might include frequency distributions based on a large number of RADM runs.
Process based approaches might include simple Eulerian-type models, such as
a two-dimensional approach operated from a three-dimensional database (e.g.,
STEM II run in x-y mode or in z-x mode). Or they might include a Lagrangian-
type model (single cell or vertically resolved trajectory model) operated from
a three-dimensional database.
• Statistical approaches applied to engineering model results are probably not
adequate as options for simple models because of inadequate treatment of local
chemistry and details of state emissions in the model.
• In process-based models, chemistry is responsible for > 80% of the costs.
• The simple model will need a general, large-number-of-species, chemistry mod-
ule to be able to correctly represent the trade-offs among VOC, NOX, and SO2
emissions.
• A process-based simple model must have fewer dimensions than the engineering
model because the need for more complex chemistry and because the restriction
on size and cost of computer time limit how complex the total representation
can be. Even two cells of chemistry will cause significant computation time.
From these considerations, Jeffries proposed that the approach should consist
of some type of a process-based, two- or three-dimensional model with most of the
meteorological inputs supplied from a complex regional-scale model such as the
engineering model or RADM. The model will have to have the ability to deal with
fairly complex chemistry.
Participants and working committee members presented various other approaches,
including the following:
22
-------
Approach to the Simple Model Model Approach
• because of source-receptor relationship problems, a low resolution engineering
model with a reduced domain;
• a Lagrangian model, but not tied to engineering model (because it does not
exist), instead the model would be tied to observations, that is, calibrated to
observations;
• a trajectory-based Eulerian model, 3 cells wide by 3 cells deep by 10 cells long;
• nomographs derived from hundreds of engineering model simulations;
• a complex chemistry model because acid deposition is really a full oxidants
issue and its appropriate treatment would need more than just S transport and
deposition.
These proposals will be examined further in the next chapter.
23
-------
Refinement of Approach
Discussion of Problems
Niki felt that parameterizing the chemistry—which has never been done—was very
difficult for the purpose of addressing sensitivity to reduction in VOC, NOX, and SOj.
Others in the workshop felt that most of the "nonlinearity" was in the gas- and
aqueous-phase chemistry and that this would have to be as complex as possible to
capture these nonlinear elements. The simple model would not be repeatedly tuned
against the RADM, as would the engineering models, and thus it would have to have
the ability to utilize the RADM chemistry or other chemistry directly.
It appeared from the resource and computer limitations, that only one cell of
chemistry could be accommodated in the simple model, because the nature of the
trade-off and nonlinearity issues seems to require a comprehensive, and therefore
expensive, chemistry treatment,.
Venkatram, Walcek, and Peters argued that boundary condition problems would
greatly compromise any realistic assessment performed with a Lagrangian cell model.
That is, the source region of interest influences the chemistry of adjacent sources to
modify the ultimate impact. The specifications of the one-cell boundary conditions,
that is, specifying the inflow of sulfur and oxidants into the cell over a three-day
period, would be a very difficult task.
After long discussion, the choice appeared to be between complex chemistry in
a single Lagrangian cell using highly parameterized transport or a multibox, but
"confined" Eulerian model that treats transport in a more realistic manner, but
must use highly simplified chemistry.
The basis of the choice was "What is it going to be used for?"
24
-------
Discussion of Problems Refinement of Approach
Venkatram said, "The only way to find the boundary conditions for the one cell
would be to run RADM for every source reduction to determine the effect on the
boundary condition." Others disagreed. Johnson, for example, cited puff model
results. He said, "It has not been shown that there are enough heterogeneities—
spatial non-uniformity—in the emissions field to create major problems. There are
isolated point sources, but how much non-uniformity do you have to have for it to
be a problem? This has not been looked at."
Venkatram seriously questioned that any such simple model would be useful in
"calculating control options," and many were concerned that it would be abused.
Lamb suggested that the present NCAR engineering models might be the simplest
viable model that could be developed.
Jeffries asked, "Do we believe that? That there is no possibility to put together
a model simpler than the engineering models that still have utility?"
Kleinman said, "There are two problems: getting the transport right and fol-
lowing the fate of the emitted pollutants. In simple Lagrangian models, you pick
a trajectory and in doing so you ignore certain components in the transport, for
example, vertical wind shear and vertical motion. But you may indeed have a 'rep-
resentative' trajectory, although not a trajectory that applies to any one situation.
On any particular realization, you are going to be in error. So granted, there are
difficulties with the transport. As far as following the chemistry, if you can aver-
age the chemical conditions, then your calculations will be satisfactory, except that
they will not apply to any particular realization. This is because, if averaging over
a domain of influence of emissions is an acceptable procedure, then you will have
results which are of interest, although not pertaining to a particular situation. The
issue is to demonstrate whether or not you can do that averaging."
Venkatram replied, "The simple approach may be satisfactory for educational
purposes, for determining qualitative effects, but as soon as you use it for assess-
ment, these uncertainties become very important."
Johnson added, "If the model suggests big controls, then people want the 'best
we can.'"
Jeffries concluded, "In that case, we will compromise. We will take, as a first
approximation, the boundary conditions from the baseline run of the RADM and
assume that those are the correct numbers. This is better than having to make
up the numbers. If we are willing to give up the actual calculation of any control
25
-------
Refinement of Approach Summary and Conclusions
values using the simple model's predictions, then what difference does it make that
these boundary conditions will change slightly when I do my calculation?"
Venkatram commented, "As an educational tool that demonstrates qualitative
effects, the approach is satisfactory."
Jeffries added, "It is important for us to decide that if this is a limitation to
the approach then we say so. That is, we say 'The approach is satisfactory for this
purpose and is not satisfactory for that purpose.'"
Walcek stated, "The purpose is primarily educational and to study and learn
about potential trade-offs. This suggests a qualitative type of model that might
be used to generate, for example, isopleths of deposition. It is not possible to
combine, in one model, both the rigorous, numerically accurate calculation of trade-
off effects—which implies high complexity—and an educational and learning tool
based on a simple model. We can all learn a great deal from a simple framework."
Kleinman said, "One way to make a model simpler is to ask simpler questions
of it. For example, you may get the overall geography wrong, but you might get
the overall transformation, mass balance, and nonlinearity correct. You need to be
careful of the questions being asked of the model."
Trout added, "The user's guide must put limitations on the simple model ap-
plications so that its abuse is decreased."
Summary and Conclusions
The working committee had two potential approaches to consider:
• complex chemistry in a single Lagrangian cell using highly parameterized trans-
port; or
• multibox, but "confined" Eulerian model that treats transport in a more real-
istic manner, but must use highly simplified chemistry.
In addition, there was a general consensus that it would be highly questionable
to use such a simple tool to perform quantitative control calculations. In an absolute
sense, the tool will not likely agree with the RADM or observations, but it might be
very good in a qualitative sense.
The simple model, however, has a good potential to be an excellent educational
tool, not only for the policy maker, but also for the scientist.
26
-------
Working Committee Meeting
The five-person working committee met for three days immediately after the work-
shop. This committee debated and discussed the approaches, selected a general
approach, and wrote a statement of work for its implementation. The statement of
work appears in the next chapter and provides detailed specifications for the model.
This chapter provides general insight into the creation of the specifications.
Evolution of the Simple Model Description
The committee chose to stress that the primary use of this tool was educational and
not to perform assessments in competition with the engineering models and RADM.
Initial discussion was directed at deciding on the basic approach. The committee
discussed, and attempted to draw, isopleths of deposition as functions of various
variables. They concluded that this was a difficult task and perhaps did not really
illustrate the transformation factors and issues adequately. The committee decided
that it was unreasonable to require the contractor to produce such specific diagrams,
but that the contractor should spend effort developing adequate display techniques
that were suitable for the role of the model as an educational tool. The details of
these requirements are given in the statement of work in the next chapter.
The committee felt that, from an educational point of view, there are three areas
that must be well represented for the model to be acceptable to scientists:
1) gas-phase chemistry,
2) aqueous-phase chemistry, and
3) cloud physics.
Committee members drew upon their experience with models to estimate that
27
-------
Working Committee Meeting Inclusion of Important Factors
• cutting the Eulerian STEM model from 10,000 cells to 100 cells would still
require 30 minutes to 1 hour for calculations with a small chemical module;
and
• one cell in the RADM, with its gas- and aqueous-phase chemistry, requires 30
minutes of VAX time.
These factors, plus points made in the workshop discussion, led the committee to
decide that the approach should be based on a simple meteorological, comprehensive
chemistry framework. The most likely form of such an approach would be a one-
cell Lagrangian model, although other approaches that had similar computational
requirements could be proposed in response to the statement of work.
In view of the emphasis, the model was named the Comprehensive Chemistry
Acid Deposition Model (CCADM).
Inclusion of Important Factors
Figure 5 is a pictorial representation of the important components of the model.
This is a conceptual figure and is not meant to portray the actual elements in the
final model. As described in the statement of work,
".. .the model will most likely treat atmospheric transport in a highly sim-
plified, yet reasonable manner. It is also likely that the source-receptor
relationship would be established external to the model. Similarly, model
inputs along the pathway would most likely be established external to the
model."
In addition, the transport processes used in the model must account for lateral and
vertical fluxes of material.
In the box, gas-phase chemistry, emissions, and dry deposition occur. To incor-
porate the cloud processes, we envisioned that the box top would sometimes be in
contact with clouds that could extract material from the box, process it (by aqueous
chemistry), and return it as rain, or the cloud could vent the material above the
box, causing permanent loss of the material. Clouds could also entrain material
from above the box as well. In addition to these external clouds, sub-box-scale
clouds could also occur within the box. These clouds would process material and
return it to the box.
Although described as a-"simple model," this model is actually very complex. In
addition to creating and exercising the model, the exploration of display methods
28
-------
Inclusion of Important Factors
Working Committee Meeting
Entrainment
Out-of-box
Cloud
Venting
Background
Concentrations
Subsidence
In-box
Cloud
Variable
Height
Extraction]./^ Aqueous
' Tj^- Chemistry
I \r?r<\
•)
Release!
Dry-
Deposition
Clear Air
Chemistry
Figure 5. Major Components of the CCASM.
29
-------
Working Committee Meeting Inclusion of Important Factors
and derivation of long-term deposition estimates from an inherently short-term
model were added to the requirements. The total project was seen as being at the
upper limit of the time and funds available for the project.
30
-------
Statement of Work
Title
Development of a Comprehensive Chemistry Acid Deposition Model (CCADM).
Period of Performance
Approximately 12 months.
Background
EPA is sponsoring the development of Eulerian, regional-scale acid deposition mod-
els. The Regional Acid Deposition Model (RADM), under development at the Na-
tional Center for Atmospheric Research (NCAR), is expected to provide scientifi-
cally credible answers to many issues related to source-receptor relationships. The
completeness and complexity of RADM, however, results in a very large requirement
for computer resources, and therefore a series of Eulerian "engineering" models are
also under development at NCAR. These will be derived from RADMand designed
to use approximations of the complex processes in RADM, especially the chemistry,
to produce a simpler, less computationally demanding model for more routine as-
sessment purposes. These models will be the primary tools for assessment of acid
deposition issues.
EPA/ORD staff requested that an even less computationally demanding model
be designed. This model would be suitable for educational use in understanding
the linear or nonlinear nature of the acid deposition process and could assist in
conducting a "what if" analysis for a particular situation. It might also be used
to design control scenarios for testing in the more complex RADM and engineering
models. A workshop, attended by EPA and non-EPA scientists, discussed the de-
sign issues for two days. A five-person steering committee of non-EPA workshop
31
-------
Statement of Work Purpose
attendees wrote this statement of work for producing a model to meet the majority
of the needs originally expressed by EPA staff.
Purpose
The purpose of this work is to produce a computer model that estimates changes
in acid deposition downwind at a receptor site as a function of changes in emis-
sion in an upwind source region. The model would estimate these changes along a
fixed trajectory for a fixed set of meteorology-related variables. The model should
use state-of-the-art gas-phase and aqueous-phase chemistry kinetics modules and
an appropriate parameterization of cloud physics. It also should include auxiliary
programs to convert information from external sources, such as output from more
complex regional-scale Eulerian models, into input needed for this model. The
model shall be exercised in a series of representative scenarios. Appropriate graphi-
cal displays shall be designed to illustrate and explain acid deposition sensitivity to
changes of emissions. Some attention should be given to techniques for demonstrat-
ing the role of short-term or episodic deposition in producing monthly, seasonally,
or annually averaged deposition. A comprehensive user's guide, including test cases
that illustrate the processes of producing inputs, running the model, and creating
graphical displays shall be produced. In addition, a second document that describes
the development of the model and the procedures that should be followed to test
and evaluate the model shall be produced.
32
-------
Conceptual Framework Statement of Work
Specifications of Model
Conceptual Framework
The objective of this effort is to develop a computationally feasible tool for edu-
cational use and "what if" analysis related to linearity or nonlinearity of the acid
deposition process. To achieve this goal, the model will most likely treat atmo-
spheric transport in a highly simplified, yet reasonable manner. It is also likely that
the source-receptor relationship would be established external to the model. Simi-
larly, model inputs along the pathway would most likely be established external to
the model. Because nonlinearities of the transformation process are of great inter-
est in the application of this tool, comprehensive chemistry is needed. One concept
thought to be adequate for the task is based upon a one-cell Lagrangian framework;
other alternative approaches are acceptable if they have similar computer resource
requirements. The basic elements of the concept include the following:
> Lagrangian box or column model
> one cell, no vertical shear
o instantaneously mixed, vertical extent determined by boundary layer height
o boundary layer should be modeled realistically including nighttime behav-
ior
> box or column area should not exceed 50 km x 50 km
> box top will sometimes be in contact with clouds
o clouds extract material from box, process it, return it as rain
o clouds extract material from box, vent it, causing permanent loss of ma-
terial
> sub-box-scale clouds can occur within box cell
o clouds incorporate material from box, process it, eventually return mate-
rial to the box
> gas-phase chemical reactions take place in the box
> aqueous-phase chemical reactions take place in the clouds
> mass transfer occurs between phases
> emissions are added to box as a function of time
> wet deposition occurs from rain in clouds, dry deposition results from uptake
of gases and aerosols by the surface
> material is entrained into box from aloft and from sides
33
-------
Statement of Work Processes To Be Represented in Model
Processes To Be Represented in Model
The minimum principal processes to be included are described below.
Chemistry
Both gas- and aqueous-phase processes must be treated.
o Gas-phase photochemical transformation of VOC, NOX, and SOz
The mechanism selected for use should
o be comprehensive and state-of-the-art
o be fully tested and documented
o include temperature-dependent reactions
o include important nighttime processes
o treat biogenic hydrocarbons
Suitable candidate chemical kinetics mechanisms include
a) SAI expanded Carbon Bond Mechanism
b) ERT Acid Deposition and Oxidant Model mechanism.
o The aqueous phase chemical transformation of SOj and NOX
o should be comprehensive and state-of-the-art,
o should include mass transfer between gaseous and aqueous phases,
o should treat free radical chemistry,
o should include oxidation/reduction reactions,
o should be updated as new information from laboratory kinetic studies
becomes available, and
o must predict cloudwater and rainwater composition.
Suitable candidate aqueous phase kinetics mechanisms are
a) ERT detailed aqueous-phase mechanism
b) NCAR RADMmechanism
c) SAI cloud chemistry mechanism
d) STEM II cloud chemistry
34
-------
Processes To Be Represented in Model Statement of Work
Deposition
o Dry Deposition
The approach should use a resistance model framework. That is,
a) surface resistance values should be derived from the values available in the
literature; and
b) atmospheric resistance, which must be expressed as a function of height,
should be derived from micrometeorology.
o Wet Deposition
Wet deposition can be derived from rainfall rate and predicted rain water
composition.
Emissions
Emission inputs are expected to come from existing emissions inventories supplied
externally, for example, from the EPRI inventory (1982) or the EPA inventory
(1983). In addition,
o the model should have capacity to treat
o NO, NO2,
o VOC as up to 10 species and CO,
o S02,
o NH3,
o Soil dust (Fe, Mn, CO3),
o Cl,
o biogenic emissions; and
o emissions should be represented as two types:
o those associated with the source area, and
o those external to the source area along the pathway to the receptor.
Cloud Processes
The model should
o treat at least stratus and cumulus clouds,
o allow precipitating and non-precipitating clouds,
o account for venting by large clouds, and
o include a method for treating the effects of patchiness of clouds.
35
-------
Statement of Work Entrainment and Dilution Processes
Entrainment and Dilution Processes
The model should allow:
o diurnally varying boundary layer thickness (box height) including
o entrainment of material from aloft due to rise of mixed layer,
o entrainment of material from the sides of the box, and
o dilution of material in box due to rise of mixed layer;
o for time variable side entrainment; and
o for teh inclusion of entrainment of background air into clouds.
Boundary Processes
The model should include time-variable boundary values aloft and laterally.
Model Inputs
Chemistry Related Inputs
To the extent possible, the chemical mechanism input module should allow the
gas-phase and aqueous-phase mechanisms to be changed or replaced in a relatively
simple operation, as re-linking modules. Coupling to the emissions input and to the
meteorological inputs should be done in as general a manner as is possible.
Meteorological Related Inputs
The minimum meteorological inputs should be
o locations and time of year,
o photolysis rates for the photolyzing species in the chemical mechanism,
o time varying mixing height,
o air temperature,
o time varying boundary conditions aloft and lateral to the box, and
o cloud parameters (inputs to aqueous phase reactor).
For these cloud parameters, the following parameters must be self consistent
and some of the parameters cannot be varied independently:
o time of occurrence
o lifetime of cloud
o lifetime of cloud droplet
36
-------
Model Outputs Statement of Work
o liquid water content of cloud
o temperature of cloud
o freezing level
o venting or updraft velocity
o cloudiness, patchiness
o entrainment
o rainfall rate
o vertical and horizontal extent
Emissions Related Inputs
Emissions inputs should be time varying.
Control Inputs
The mimimum control inputs should include beginning and ending simulation times
and the range of variation for VOC, NOX, and SO2 emissions.
Model Outputs
The normal model output should include displays of deposition of selected species
over the pathway and displays of concentrations of selected species over the path-
way. In addition, the output should include displays illustrating the importance of
processes at selected times.
The model must also include diagnostic output, such as data files for subsequent
processing and details of intermediate calculations for QA purposes.
37
-------
Statement of Work Statement of Work
Statement of Work
Task 1-Develop and Submit Detailed Work Plan
Create complete project plan and submit within two weeks after award. Identify
algorithms, strategies, and approaches to be used and work needed to implement
them. Outline or flowchart major code components. Estimate computer resource
requirements both to produce the model and, separately, to run the model in a
representative scenario. Identify and plan quality assurance steps. Review and ap-
proval by project officer required before proceeding to Tasks 2 and 3. The approved
project plan will be incorporated into the work plan.
Task 2—Develop and Test Major Model Components
> Create model component subroutines
> Demonstrate correct functioning of model subcomponents
Each major subcomponent of the model (e.g. the cloud chemistry submodel)
must be numerically tested and the contractor must demonstrate correct func-
tioning by showing test cases in progress reports. The results of major in-
termediate calculations must also be demonstrated, e.g., dilution rate profile,
N(>2 photolysis rate. Review and approval by project officer required before
proceeding to Task 3 and Task 4.
Task 3—Integrate Model Components and Develop Input Processor
> Create main model computer program
> Must produce auxiliary programs to assist in calculating inputs
Create a "user friendly" input processor to assist users in selecting from a
collection of "worst case" situations and in modifying the input parameters
in a consistent manner. Inputs to this program may come from analysis of
Eulerian regional models output (e.g., RADM output), or from observations, or
they may be created from climatological/statistical data.
o Must use a method to assure cloud parameters are consistent by basing
the treatment of clouds on one or more of the following methods:
o look-up tables consistent with observations and/or
o cloud models
> Project officer's approval needed before proceeding to Task 4.
38
-------
Task 6-Develop Display Technique Statement of Work
Task 4-Exercise Model
> The chemical mechanisms selected for use in these tests must be approved by
the project officer
> Perform limited sensitivity studies:
o e.g., assess effects of including/excluding side entrainment of material, and
o e.g., assess effects of excluding the uptake of free radical species by the
aqueous phase droplet.
> Scenarios should include at least the following factors:
o travel time 1/2 day to 3 days
o summer and winter seasons
o path mostly rural and path mostly polluted
o wet path and dry path
o non-precipitating and precipitating clouds
o high background and low background
o initial starting times varyed.
> Based'on the results for the above scenarios, determine the situations and time
scales over which the acid deposition process appears to be linear or nonlinear
Task 5—Assess Feasibility of Deriving Long-term Deposition Estimates
Although the basic model will only provide short averaging time (1-3 hr) estimates
of outputs such as S deposition, the contractor should assess the feasibility of pro-
ducing estimates averaged over episodes (1 to 5 days), months, seasons, and years.
This may not be feasible given the nature of the model and its computational re-
quirements, but the contractor is encouraged to investigate the feasibility of deriving
such estimates.
Task 6-Develop Display Technique
A graphical display technique should be developed that will demonstrate, in a
straightforward manner, the deposition and concentration of sulfur, nitrate, and
oxidant at selected points (e.g., day one, day two, day three) in the pathway as a
function of VOC, NOX, and SOj emissions from a particular source area. Since these
deposition estimates are expected to be highly variable, it would be desirable to see
numerous "representative" trajectories analyzed for meteorological conditions asso-
ciated with dry and wet periods during several seasons. It would also be useful to
see estimates of deposition and concentrations along the path of a given trajectory.
39
-------
Statement of Work Task 7-Propose Procedures for Testing and Evaluating Model
Results should be presented in a fashion allowing the user to assess the contribu-
tion of the source area of interest to the total deposition in the receptor area and
allowing any "nonlinearity" in the process to be readily perceived.
Task 7-Propose Procedures for Testing and Evaluating Model
Propose procedures for testing and evaluating the model against observations and/or
regional Eulerian model output (e.g., RADM output) to uncover limitations in model.
Such a testing/evaluation would be conducted by an by independent contractor not
involved in original development. Testing and evaluation procedures should include
testing and examination of the computer code.
Task 8—Documentation of Model
> Technical Support Document contents
o Describe the model
o Describe all known assumptions and limitations
o Describe procedures and algorithms used to produce calculations; in-
clude references to sources of algorithms
o Describe the results of simulations run under Task 4 to exercise the model
o Describe the time and space scales over which the acid deposition process
appears to be linear or nonlinear
o Describe followup procedures that should be undertaken to test and eval-
uate the model
> User's Guide contents
o Provide instructions on how to run the model
o Describe procedures to extract the appropriate information from regional
models like RADM for use as input information in the CCADM
o Describe, by example, how a RADM run might be designed based on results
obtained with the CCADM
o Describe computer resources required to run the programs
o Provide instructions for installing programs on computers
o Provide block diagrams, flow of control diagrams
o Provide a definition of every variable in every common statement
o Provide fully commented code listing
40
-------
Computer Programming Requirements Statement of Work
Technical Proposal Instructions
Minimum Qualifications
Experience in acid deposition modeling, cloud chemistry modeling, gas- and aqueous-
phase chemistry modeling.
Key Personnel
The on-site personnel must include at least one chemist with experience in gas-
and aqueous-phase chemical model development and who will be available to the
project on a routine basis. The project team must include expertise in cloud physics,
meteorology, and Lagrangian acid deposition modeling.
Computer Programs
Contractor must have existing in-house code for chemical kinetics modeling of the
gas- and aqueous-phase chemistry.
Computer Programming Requirements
The model must be supplied to EPA in ANSI standard FORTRAN, preferably
FORTRAN??, and it must be designed and coded to run on different machines, at
a minimum, the machines are:
o EPA UNIVAC
o IBM 4300-type machines
o DEC VAX
41
-------
Appendix
A. Discussion Leaders and Topics
B. List of Participants
42
-------
Discussion Leaders and Topics
Session 1 : Introduction
• Open Workshop
> B. Dimitriades (20 min)
Describe why the workshop is being held, how it was organized. Name the main groups
of participants and what role they play.
Introduce the participants one by one. Each describes affiliation, background.
• Introductory Remarks by ADPS, Washington
> D. Trout (15 min)
Upper level decision maker has broader perspective. Describe, from a policy maker's
viewpoint, the type of model needed and why.
• Introductory Remarks by Working Committee
> H. Jeffries and Working Committee (20 min)
Describe the workshop purpose and approach.
Describe the expected products.
Describe organization and timing of discussion.
Individual WC member comments.
• Model Descriptions (35 min)
This material will provide general information about models that will be discussed frequently
in later sessions. For each model, describe
o Purpose
o Formulation basis
o Development process
o Application methodology
o Operational Status
o Costs
> NCAR Regional Models (10 min)
o C. Walcek, NCAR
> Canadian Regional Models (8 min)
o A. Venkatram, ERT
> Other Regional Models (8 min)
o L. Peters, UK
> EKMA-type models (8 min)
o G. Whitten, SAI
43
-------
Session 2 : Description of Model Objectives
• Concept of Hierarchical Application of Models (10 min)
> D. Trout, ADPS/EPA
• A Scenario For Assessment (20 min)
> H. Jeffries, UNC &: D. Trout, ADPS/EPA
• What Will Be Calculated by the Simple Model? (30 min)
t> H. Jeffries, UNC & J. Durham, ASRL/EPA
• Expected Use of Model Results (10 min)
> D. Trout, ADPS/EPA
• Resource and Time Constrains (2 min)
> H. Jeffries, UNC
• Summary of Background Information (8 min)
> H. Jeffries, UNC
44
-------
Session 3 : Development of Approach
• Assumptions about Approach to Simple Model (10 min)
• Approaches for Model Components (80 min)
> Chemistry (10 min)
o A. Lloyd, ERT
o H. Niki, Ford
t> Meteorology (40 min)
o source-receptor relationship (20 min)
o A. Venkatram, ERT
o mixing (10 min)
o J. Shannon, ANL
o clouds and storms (10 min)
o C. Walcek, NCAR
t> Emissions (10 min)
o L. Peters, UK
> Boundary Conditions (10 min)
o A. Venkatram, ERT
> Display/Analysis of Output (10 min)
o G. Whitten, SAI
45
-------
Session 4 : Refinement of Information Needed
• Information Needed in Each Component (90 min)
> Working Committee
Session 5 : More Refinement
• Linkage with Regional Model (90 min)
> Working Committee
Session 6 : Limitations on Approaches
• What can the model not calculate?
> C. Walcek, NCAR
• Are there situations in which results will be poor?
> H. Niki, Ford
• What is the range of appliabilty?
t> A. Venkatram, ERT
• What are estimates of uncertainty?
Session 7 : Performance Criterial and Testing
• What Should the Performance Criteria and Test Procedures Be?
> L. Peters, UK
Session 8 : Summary
• Working Committee
46
-------
Appendix B
EPA Workshop on Development of a Simple Model for Acid Deposition
List of Participants
Dr. Joseph Bufallni
Atmospheric Sciences Research Laboratory
U.S. Environmental Protection Agency (MD-84)
Research Triangle Park, NC 27711
919/541-2422
Dr. John Clarke
Atmospheric Sciences Research Laboratory
U.S. Environmental Protection Agency (MD-80)
Research Triangle Park, NC 27711
919/541-3660
Dr. Basil Dimitriades
Atmospheric Chemistry and Physics Division
Atmospheric Sciences Research Laboratory
U.S. Environmental Protection Agency (MD-59)
Research Triangle Park, NC 27711
919/541-2706
Dr. Marcia Dodge
Atmospheric Sciences Research Laboratory
U.S. Environmental Protection Agency (MD-84)
Research Triangle Park, NC 27711
919/541-2374
Dr. Jack Durham
Atmospheric Chemistry and Physics Division
Atmospheric Sciences Research Laboratory
U.S. Environmental Protection Agency (MD-57)
Research Triangle Park, NC 27711
919/541-2183
Mr. Jerry Gibson
Off ice of Air Quality, Planning, and Standards
U.S. Environmental Protection Agency (MD-14)
Research Triangle Park, NC 27711
919/541-5522
Mr. Mark Hodges
Off ice of Air Quality, Planning, and Standards
U.S. Environmental Protection Agency (MD-15)
Research Triangle Park, NC 27711
919/541-5665
Dr. Harvey Jeffries
Dept. of Environmental Sciences
and Engineering
Room 105, Rosenau Hall (201-H)
UNC-Chapel Hill
Chapel Hill.NC 27514
919/966-5451 or 942-3245
Mr. David Johnson
Office of Air Quality Planning and Standards
U.S. Environmental Protection Agency (MD-15)
Research Triangle Park, NC 27711
919/541-5516
Dr. Warren Johnson
Atmospheric Science Center
SRI International
333 Ravenswood Avenue
MenloPark, CA 94025
415/859-4755
Dr. Larry Kleinman
Brookhaven National Laboratory
Atmospheric Sciences Division
Building 51
Upton, NY 11973
516/282-3796
Dr. Robert Lamb
Meteorology Division
Atmospheric Sciences Research Laboratory
U.S. Environmental Protection Agency (MD-80)
Research Triangle Park, NC 27711
919/541-4181
Dr. Alan Lloyd
Environmental Research and Technology
975 Business Center Circle
NewburyPark,CA 91320
805/499-1922
-------
Mr. Brian McLean
Office of Policy Development (ANR-445)
401 M Street, SW
Washington, DC 20460
202/382-7748
Mr. Ned Meyer
Office of Air Quality Planning and Standards
U.S. Environmental Protection Agency (MD-14)
Research Triangle Park, NC 27711
919/541-5522
Dr. P. K. Misra
Air Resources Branch
Ontario Ministry of the Environment
880 Bay Street, Fourth Floor
Toronto, Ontario
CANADA M5S 1Z8
416/965-5068
Dr. Brand Niemann
Acid Rain Policy Office (OAR-445)
Office of Air and Radiation
U.S. Environmental Protection Agency
401 M Street, SW
Washington, DC 20460
202/382-4423
Mr. Brock Nicholson
Office of Air Quality, Planning, and Standards
U.S. Environmental Protection Agency (MD-15)
Research Triangle Park, NC 27711
919/541-5516
Dr. Hiromi Niki
Ford Motor Company
P.O. Box 2053
Dearborn, Ml 48121
313/337-8322
Dr. Leonard K. Peters
Dept. of Chemical Engineering
163 Anderson Hall
University of Kentucky
Lexington, KY 40506-0046
606/257-4959
Dr. Christian Seigneur
Systems Applications, Inc.
101 Lucas Valley Road
San Rafael, CA 94903
415/472-4011
Dr. Jack Shannon
Environmental Research Division
Building 181
Argonne National Laboratory
Argonne, IL 60439
312/972-5807
Dr.JackShreffler
Atmospheric Sciences Research Laboratory
U.S. Environmental Protection Agency (MD-59)
Research Triangle Park, NC 27711
919/541-2191
Mr. Joseph Tikvart
Office of Air Quality Planning and Standards
U.S. Environmental Protection Agency (MD-14)
Research Triangle Park, NC 27711
919/541-5561
Dr. Dennis Trout
Acid Deposition Planning Staff
Room 38-17 (RD-676)
U.S. Environmental Protection Agency
401 M Street, SW
Washington, DC 20460
202/382-5945
Dr. Akula Venkatram
Environmental Research and Technology
975 Business Center Circle
Newbury Park, CA 91320
805/499-1922
Dr. Chris Wakek
National Center for Atmospheric Research
P. O. Box 3000
Boulder, CO 80307
303/497-1686
Dr. Gary Whitten
Systems Applications, Inc.
101 Lucas Valley Road
SanRafael.CA 94903
415/472-4011
-------
------- |