United States Atmospheric Sciences
Environmental Protection Research Laboratory
Agency Research Triangle Park NC 27711
Research and Development EPA/600/3-88/OQ9 Feb. 1988"
Evaluation and Assessment of
UNAMAP
-------
EVALUATION AND ASSESSMENT OF UNAMAP
by
R. Ernest Baumann
and
Rita K. Dehart
Battelle
Columbus Division
Information Systems Section
2030 M Street, NW
Washington, DC 20036-3391
Contract Number
68-02-4189
Project Officer
D. Bruce Turner
Meteorology and Assessment Division
Atmospheric Sciences Research Laboratory
Research Triangle Park
North Carolina 27711
ATMOSPHERIC SCIENCES RESEARCH LABORATORY
OFFICE OF RESEARCH AND DEVELOPMENT
U.S. ENVIRONMENTAL PROTECTION AGENCY
RESEARCH TRIANGLE PARK, NORTH CAROLINA 27711
-------
NOTICE
The information in this document has been funded by the United States
Environmental Protection Agency under Contract No. 68-02-4189 to Battelle -
Columbus Division. It has been subject to the Agency's peer and administrative
review, and it has been approved for publication as an EPA document. Mention
of trade names or commercial products does not constitute endorsement or
recommendation for use.
n
-------
ABSTRACT
The Evaluation and Assessment of UNAMAP is a study to determine how
best to improve the usefulness and availability of the UNAMAP air
pollution dispersion models. This report describes a plan for
implementing a series of recommended improvements to the UNAMAP
program. It also describes the earlier parts of the study upon which
the plan is based.
The study consisted of a technology assessment followed by data
collection and analysis which were used to develop the strategy and the
specific improvements contained in the plan. The overview includes a
summary of the data collected during the study, and the conclusions
drawn from analysis of that data. The analysis indicates that
improvements to UNAMAP are needed in the areas of: 1) model accuracy,
2) model documentation, 3) user support, 4) data collection, 5) data
input, and 6) computer compatibility.
The report describes a plan which is based on a strategy that is
consistent with the long-term objectives for UNAMAP- The report
contains a strategic framework for improving UNAMAP. This strategic
framework guides the implementation and strikes a balance between the
goals of advancing dispersion modeling research and transferring
technology to the public. Included in the plan is a set of ten
specific recommended improvements which are described and broken into
specific tasks. Estimates of time and cost required to implement each
improvement are given at the task level.
Finally, the recommended improvements are grouped into three phases. A
schedule is presented at the task level for the five recommended
improvements in Phase 1.
iii
-------
EVALUATION AND ASSESSMENT OF UNAMAP
TABLE OF CONTENTS
PART I; PROJECT SUMMARY
Page
PROJECT SUMMARY 1-1
PART II: TECHNOLOGICAL ASSESSMENT
1.0 INTRODUCTION 11-4
1.1 Purpose II-4
1.2 Organization II-4
2.0 UNAMAP VERSION 6 TECHNOLOGY II-6
2.1 Contents of Version 6 II-6
2.2 Modeling Technology in Version 6 .. II-7
2.2.1 Air Pollution Meteorology II-7
2.2.2 Software 11-15
2.3 Data Technology in Version 6... 11-16
2.3.1 Meterological Data 11-16
2.3.2 Source and Emissions Data ... 11-19
2.4 Computer Technology in Version 6....... 11-20
2.4.1 Processing........................ ......... 11-20
2.4.2 Data and File Transfer.............. ....... 11-21
2.4.3 Graphics. 11-22
2.5 Summary: Technology Utilized by UNAMAP Version 6..... 11-22
3.0 DEVELOPMENTS AND ADAPTATIONS OF UNAMAP MODELS
TO CURRENT TECHNOLOGY n_23
3.1 Model ing Technology.. 11-23
3.1 = 1 Air Pollution Meteorology. 11-23
3.1.2 Software... ................................ H-24
-------
Page
3.2 Data Technology 11-25
3.2.1 Meteorological Data 11-25
3.2.2 Source and Emissions Data 11-26
3.3 Computer Technology 11-26
3.3.1 Processing 11-26
3.3.2 Data and File Transfer 11-27
3.3.3 Graphics 11-27
3.4 Summary: Developments and Adaptations of Air
Quality Models to Current Technology 11-28
4.0 TECHNOLOGY TO SOLVE CURRENT PROBLEMS 11-29
4.1 Modeling Technology 11-29
4.1.1 Air Pollution Meteorology 11-29
4.1.2 Software 11-30
4.2 Data Technology 11-31
4.2.1 Meteorological Data 11-31
4.2.2 Source and Emissions Data 11-32
4.3 Computer Technology 11-32
4.3.1 Processing 11-32
4.3.2 Data and File Transfer 11-33
5.0 SUMMARY 11-35
PART III: INTERIM REPORT
10 INTRODUCTION 111 -7
1.1 Background III-7
1.2 Data Collection and Analysis III-7
1.3 UNAMAP User Profile III-9
1.4 Analysis of Problems/Suggestions 111-13
1.5 Overview of Problem Areas 111-15
-------
Page
2.0 MODEL ACCURACY .. 111-17
2.1 Importance of Accuracy > 111-17
2.2 User Suggestions. 111-18
2.3 User Concerns and Alternatives...... 111-18
2.3.1 Concerns 111-18
2.3.2 Alternatives 111-19
2.4 Evaluating Accuracy Alternatives 111-19
2.5 Constraints 111 -20
3.0 DOCUMENTATION .. 111-21
3.1 Importance of Documentation 111-21
3.2 User Suggestions.... 111-23
3.3 User Concerns and Alternatives... 111-23
3.3.1 Concerns 111-23
3.3.2 Alternatives 111-24
3.4 Evaluating Documentation Alternatives 111-25
3.5 Constraints 111-25
4.0 USER SUPPORT 111-26
4,1 Importance of Support 111-26
4.2 User Suggestions 111-26
4.3 User Concerns and Alternatives... ...... 111-28
4.3.1 Concerns. 111-28
4.3.2 Alternatives 111-31
4.4 Evaluating Support Alternatives 111-32
4.5 Constraints . 111-32
5.0 DATA COLLECTION IH-33
5.1 Importance of Data Collection 111-33
5.2 User Suggestions 111.35
-------
Page
5.3 User Concerns and Alternatives 111-35
5.3.1 Concerns 111-35
5.3.2 Alternatives 111-36
5.4 Constraints 111 -37
6.0 DATA INPUT 111-38
6.1 Importance of Data Input 111-38
6.2 User Suggestions 111-39
6.3 User Concerns and Alternatives 111-39
6.3.1 Concerns 111-39
6.3.2 Alternatives 111-39
6.4 Evaluating Data Input Alternatives 111-40
6.5 Constraints III-40
7.0 HARDWARE COMPATIBILITY 111-41
7.1 Importance of Hardware Compatibility 111-44
7.2 User Suggestions 111-44
7.3 User Concerns and Alternatives 111-44
7.3.1 Concerns 111-44
7.3.2 Alternatives 111-44
7.4 Evaluating Hardware Compatibility Alternatives 111-48
7.5 Constrai nts 111 -48
8.0 OTHER AREAS OF CONCERN 111-49
8.1 Response Time.... 111-49
8.2 Output 111 -50
8.3 Specific Models 111-51
9.0 GENERAL CONSTRAINTS 111-56
9.1 Technological Constraints: Computer Resources 111-56
9.2 Economic Constraints: Budget and Staff.... 111-57
vii
-------
Page
9.3 Regulatory Constraints 111-57
9.4 Future Outlook for Air Quality Modeling and UNAMAP... 111-57
10.0 RECOMMENDATIONS 111-60
10.1 Recommendation 1: Establish an Electronic Bulletin
Board III-60
10.1.1 Description 111-60
10.1.2 Benefits 111-60
10.1.3 Time and Cost Estimates, 111-61
10.1.4 Alternatives Not Chosen ....... 111-62
10.2 Recommendation 2: Produce, Distribute, and Support
a Series of End-user Documentation 111-63
10.2.1 Description .................. 111-63
10.2.2 Benefits of End-User Documentation 111-63
10.2.3 Time and Cost Estimates 111-63
10.3 Recommendation 3: Provide UNAMAP Code, Data, and
Documentation for Multiple Computers. . 111-64
10.3.1 Description 111-64
10.3.2 Benefits .. 111-65
10.3.3 Disadvantages 111-67
10.3.4 Time and Cost Estimates 111-67
10.4 Recommendation 4: Improve the Accuracy and Technology
of Models Included in UNAMAP 111-68
10.4.1 Description 111-68
10.4.2 Benefits.o 111-69
10.4.3 Time and Cost Estimates 111-69
10.5 Recommendation 5: Develop a Consistent Set of User
Interfaces __ m_69
10.5.1 Description _ 111-69
10.5.2 Benefits.... 111-70
10.5.3 Disadvantages.................... ......... III-7Q
10.5.4 Time and Cost Estimates. ....... ...... III-7Q
VXll
-------
Page
10.6 Recommendation 6: Consolidate all Support for UNAMAP
Model s 111 -71
10.6.1 Description 111-71
10.6.2 Benefits 111-71
10.6.3 Time and Cost Estimates 111-72
10.7 Recommendation 7: Establish a Meteorology Data
Clearinghouse 111-72
10.7.1 Description 111-72
10.7.2 Benefits 111-73
10.7.3 Time and Cost Estimates 111-73
10.8 Recommendation 8: Develop or Acquire Specialized Models
For Inclusion in UNAMAP 111-74
10.8.1 Description 111-74
10.8.2 Benefits 111-75
10.8.3 Time and Cost Estimates 111-75
10.9 Recommendation 9: Support the Collection of Additional
and More Accurate Meteorological Data 111-76
10.9.1 Description 111-76
10.9.2 Benefits 111-76
10.9.3 Estimated Cost and Time 111-77
10.10 Recommendation 10: Support the Electronic Transfer of
UNAMAP Documentation 111-77
10.10.1 Description 111-77
10.10.2 Benefits 111-77
10.10.3 Time and Cost Estimates. 111-77
11.0 ALTERNATIVES NOT RECOMMENDED 111-79
11.1 Graphics Output for Models 111-79
11.2 Improving Response Time 111-79
12.0 SUMMARY 111 -80
ix
-------
Page
APPENDIX A: INTERVIEW GUIDE A-l
APPENDIX B: QUESTIONNAIRE B-l
PART IV: IMPLEMENTATION PLAN
1.0 INTRODUCTION IV-5
2.0 PROJECT BACKGROUND IV-8
2.1 Technology Assessment Overview .. IV-8
2.2 Data Collection and Analysis Techniques Overview IV-10
2.3 Problem Area Overview IV-11
2.3.1 Accuracy.... .... IV-13
2.3.2 Documentation IV-14
2.3.3 Support IV-14
2.3.4 Data Collection... IV-15
2.3.5 Data Input IV-16
2.3.6 Hardware Compatibility... IV-16
3.0 RECOMMENDED IMPLEMENTATION APPROACH IV-18
3.1 Changes in the UNAMAP Program..... IV-18
3.2 Research vs. Technology Transfer .. IV-19
3.3 Strategic Objectives IV-20
4.0 IMPLEMENTATION OF RECOMMENDATIONS.... IV-23
4.1 Recommendation 1: Establish an Electronic
Bulletin Board.... ... ^ IV-23
4.1.1 Implementation Tasks ^ IV-24
4.1.2 Estimated Costs and Time ft IV-25
-------
Page
4.2 Recommendation 2: Produce, Distribute, and Support
a Series of End-User Documentation IV-26
4.2.1 Implementation Tasks IV-28
4.2.2 Estimated Costs IV-29
4.3 Recommendation 3: Provide UNAMAP Code, Data,
and Documentation for Multiple Computers IV-31
4.3.1 Implementation Tasks IV-32
4.3.2 Estimated Costs and Time IV-33
4.4 Recommendation 4: Improve the Accuracy and Technology
of Models Included in UNAMAP IV-34
4.4.1 Implementation Tasks IV-34
4.4.2 Estimated Costs and Time IV-35
4.5 Recommendation 5: Develop a Consistent Set of
User Interfaces IV-36
4.5.1 Implementation Tasks IV-38
4.5.2 Estimated Costs and Time IV-39
4.6 Recommendation 6: Provide Centralized Support
for All Models IV-41
4.6.1 Implementation Tasks IV-42
4.6.2 Estimated Costs and Time IV-43
4.7 Recommendation 7: Establish a Data Clearinghouse IV-44
4.7.1 Implementation Tasks IV-45
4.7.2 Estimated Costs and Time IV-46
4.8 Recommendation 8: Develop Specialized Models IV-46
4.9 Recommendation 9: Support the Collection of
Additional and More Accurate Data.. IV-47
4.9.1 Implementation Tasks IV-48
4.9.2 Estimated Costs and Time IV-49
4.10 Recommendation 10: Support the Electronic Transfer
of UNAMAP Documentation IV-49
4.10.1 Implementation Tasks IV-49
4.10.2 Estimated Costs and Time IV-50
xi
-------
Page
5.0 SCHEDULE IV-52
5.1 Detailed Schedule for Phase I Tasks.. . IV-53
5.2 Budget and Schedule Summary for Phase I... IV-55
6.0 SUMMARY IV-58
APPENDIX...... ..... IV-60
xii
-------
Table 1.
Table 2.
Table 3.
EVALUATION AND ASSESSMENT OF UNAMAP
LIST OF TABLES AND FIGURES
PART II: TECHNOLOGICAL ASSESSMENT
Guideline Models - Appendix A.
Page
II-8
Guideline Models - Appendix B II-9
Non-Guideline UNAMAP Models 11-10
Table 4. Guideline Non-UNAMAP Models 11-12
Figure 1. Meteorological Data Transfer Process 11-17
PART III: INTERIM REPORT
Table 1. Questionnaire Respondents by Industry Group 111-10
Table 2. Years of Modeling Experience by Industry Group 111-11
Table 3. Nature of Respondents' Work with UANMAP 111-12
Table 4. Severity of Problems by Industry Group....... 111-14
Table 5. Documentation Importance Vs. Ease of Use 111-22
Table 6. Type of Support Requested By UNAMAP Users 111-27
Table 7. Names of Organizations Used for UNAMAP Support 111-29
Table 8. Types of Computers Used for UNAMAP Modeling 111-42
xiii
-------
Table 9. Version Number and Source of Models Used.
Page
111-45
Table 10. More UNAMAP Models Listed Than Exist................. 111-52
Table 11. Most Frequently Used UNAMAP Models............... 111-54
Table 12. Users Prefer to Buy UNAMAP 111-66
Figure 1. Types of Data Used by Air Quality Models... 111-34
PART IV: IMPLEMENTATION PLAN
Figure 1. Problem Areas Ratings of Seriousness.................. IV-12
Figure 2, Source Characteristics Data Screen............., ... IV-37
Figure 3= Schedule for Phase I Improvements............ ... .<. .. IV-56
Figure 4. Estimated Cost and Start Times for
Each Phase I Recommendation. ...
IV-57
XIV
-------
EVALUATION AND ASSESSMENT OF UNAMAP
PART I: PROJECT SUMMARY
February 1988
-------
EVALUATION AND ASSESSMENT OF UNAMAP
PART I: PROJECT SUMMARY
INTRODUCTION
The User's Network for Applied Modeling of Air Pollution (UNAMAP) is a software
library of air quality simulation models provided by the Environmental
Operations Branch (EOB) of EPA's Atmospheric Science Research Laboratory
(ASRL). The Evaluation and Assessment of UNAMAP project was designed to
facilitate EOB's ongoing efforts to improve the utility and availability of
UNAMAP to the public.
The results of the study describe a plan for implementing a series of
recommended improvements to the UNAMAP program. The plan consists of a
definition of strategy consistent with EPA's objectives for UNAMAP, and the
schedule and budget to implement the specific recommendations.
The improvements were derived during the course of the investigation, and were
based on two major sets of requirements. Technology requirements determine
what computer, data, and modeling technology are available to UNAMAP users now
and in the near future. User requirements determine what areas of UNAMAP
utilization are most difficult and determine where improvement would be most
beneficial. Each recommendation has been formulated both to take advantage of
the current technological environment and to help meet expressed and implied
user requirements.
The conclusions and recommendations in this study are based on research and
analysis which were performed in three phases. The first was a technology
assessment to determine technology available to UNAMAP. The second was data
collection (by interview and questionnaire) and analysis. The third was the
derivation of recommendations.
1-2
-------
TECHNOLOGY ASSESSMENT
All products follow a pattern of growth which involves changes to the product,
the users, and the market. UNAMAP has followed such as growth pattern. The
Technology Assessment Report evaluated the technological environment in which
UNAMAP exists today and that in which it will function most effectively in the
future. The information was used in subsequent stages of the project to
provide evaluation criteria for system alternatives and final recommendations.
The technology assessment found that all components of modeling technology have
changed dramatically since the development of the first computerized air
quality models in the 1960's. The technology is continuing to change at a fast
pace. The report covered three distinct phases in air quality model
development: (1) the technology as utilized by UNAMAP Version 6; (2) the
technologies used by models currently under development, as well as adaptations
made to UNAMAP to utilize current technology; (3) the technological
requirements for future models to solve current problems.
In Version 6, the technology applied to the models has progressed, while that
of the computer systems used to run them at EPA has not. Although the original
Gaussian dispersion mathematics are still used, the UNAMAP models have become
more sophisticated and consistent. Adaptations and new processors have
Increased the models' usability. The UNIVAC 1100 utilized for support,
however, represents no advances since UNAMAP was begun in 1973.
New modeling developments have grown from the user community's attempts to deal
with today's air quality modeling applications. Model developers are creating
models for more challenging environmental problems. UNAMAP users have adapted
the models to respond to situations that are different or more complex than
those for which the software was designed. Consultants and third-party vendors
have contributed more sophisticated data collection and input methods and
facilitated the use of new computer technology.
1-3
-------
Modeling software must advance to meet both application and user needs.
Modeling has become a successful and important tool in protecting the
environment. Because of this success, models are needed for even more complex,
real-world situations which need representation. Complex regional models are
needed by county, regional, and state agencies to study dispersion and
transport of pollutants over larger distances. More realistic complex terrain
models are needed to represent geographical areas where terrain is a factor.
The typical model user has changed since the advent of UNAMAP. Today, a user
may not be a "modeling expert." He may use the models only a few times a year
or lack the computer expertise to use mainframe versions easily. Services such
as modeling consultants and developers of menu-based microcomputer versions
have grown to meet the needs of today's user.
Users expect software which not only fits the application, but is also easier
to use. In judging ease of use, the user-friendly, microcomputer products
available for other applications will be used as criteria against which the
UNAMAP software will be judged.
The data used by the models will continue to be provided by the National
Weather Service (NWS) or collected at the site in question. Therefore, more
consistency of data input and output among the models will be needed to
accommodate the novice or infrequent user.
Computer technology will continue its trend toward distributed processing with
microcomputers used for an increasing amount of modeling activity, including
graphics output. User expectations will also require that data gathering and
file transfer technologies in UNAMAP match those available in other application
areas.
DATA COLLECTION AND ANALYSIS
A major portion of the Evaluation and Assessment of UNAMAP project was the
collection of data from UNAMAP users and other members of the dispersion
1-4
-------
modeling community. The data helped define who uses UNAMAP and how the models
are used. The data also identified the areas of the modeling process which
users felt should be improved.
There were two parts to the data collection activity: personal interviews and
mailed questionnaires. In-depth interviews were conducted with 22
representatives of the following organizations:
o EPA (ASRL and OAQPS)
o EPA Regional Meteorologists
o State air quality agencies
o Local or county air quality agencies
o Private industries (as users of the models)
o Educational institutions
o Modeling consultants
o Modeling software marketers
The interviews provided an overview of the UNAMAP system as well as information
on the modeling process. The data from the interviews were used to design a
questionnaire which was mailed to a larger segment of the UNAMAP community (256
organizations).
Usable data were returned in 112 questionnaires. These questionnaires were
analyzed, and certain types of data were extracted including a user profile,
which models are used, and problems encountered in the modeling process. A
typical UNAMAP user:
o Belongs to one of four major industry groups: consultant, state
government, private industry, or local government;
o Classifies himself as a user of the models;
o Has an experience level which varies by industry group.
The largest portion of the questionnaire dealt with the problems perceived by
the users and their suggestions for improvement. Twelve potential problem
areas were identified by Battelle project team members based on conversations
1-5
-------
with EPA. Questionnaire respondents were asked to rank the problem areas as to
the severity of the problem. User responses were grouped into low, medium, and
high categories. Problem areas scored as medium or high on the severity scale
by all major user groups were further analyzed. User suggestions to correct
these problems were categorized, and a percentage of suggestions in each
category was computed.
Six out of the original twelve potential problem areas were rated as major
problems by the users. These are the problems which Battelle's recommendations
attempt to rectify.
The following table shows the areas specified as major problems as well as the
potential problems not considered significant by most users:
PROBLEM AREAS
MAJOR PROBLEM NOT A MAJOR PROBLEM
Accuracy of the models Choosing a model
Documentation of models Buying/accessing the model
Support for models Hardware access
Data collection Unreliable hardware
Data input Response time
Hardware compatibility Other (users could specify
problems not listed)
Alternatives were generated from the suggestions and from typical components of
technical support in the informations systems industry. Evaluation criteria
and constraints were identified, and used to determine the strengths and
weaknesses of the alternatives.
The suggestions and alternatives were evaluated using the technological context
described in the Technology Assessment and industry experiences with currently
implemented systems as described in the trade literature. Through the
evaluation process, Battelle developed a list of recommendations. The
1-6
-------
recommendations were formulated to address multiple problem areas wherever
practical, and to utilize technology which is available to the UNAMAP support
staff and model users.
RECOMMENDATIONS
The UNAMAP program has great visibility and impact on the public, including
industry, state air pollution control agencies, and community groups. To take
advantage of this visibility, UNAMAP needs to provide a centralized modeling
service of high quality. The recommended improvements, when taken as a whole,
will allow UNAMAP to achieve the following goals:
o To function as the public source of newly developed and refined
air quality models.
o To distribute models which are easily executed on a variety of
commonly-used computers of all sizes.
o To provide a wide set of models which are relatively easy to
execute, even for the novice or occasional user.
o To offer modelers a central source of technical information,
meteorological data, and user support.
To attain these goals, a long-range strategy and milestones are required.
Therefore, the first part of the implementation plan is a strategic framework.
This serves as a guideline for the assignment of priorities to the
recommendations. The priorities are consistent with the long-term objectives
of the UNAMAP program. The strategy is designed to strike a balance between
the two objectives of UNAMAP: advancement of research in dispersion modeling,
and the effective transfer of modeling technology to the public.
The approaches discussed will first expand the users' ability to use the
models, and then serve to advance the modeling technology being used. The
initial changes must allow UNAMAP to evolve to meet the current expectations of
the modeling community. When using the regulatory models is less difficult,
modelers will be able to turn more of their attention to the research aspects
of modeling. As in the early days of UNAMAP, the user community will become
1-7
-------
more involved in the process of testing and validating new models as part of
the research cycle.
The strategy can be executed through a set of ten specific recommendations for
improvement to UNAMAP. They are the following:
o Establish an electronic bulletin board on a multi-user computer
system.
o Provide a set of end-user documentation for all UNAMAP models.
o Provide models which execute on IBM mainframes, DEC VAX computers,
and IBM PC's.
o Improve the accuracy of models.
o Develop consistent user-friendly interfaces for all models.
o Consolidate all support for all UNAMAP models.
o Establish a meteorological data clearinghouse,,
o Include more special purpose models in UNAMAP.
o Support the collection and use of additional and more accurate
meteorological data.
o Support the electronic distribution of UNAMAP documentation and
updates.
The report summarizes each recommendation, and discusses implementation tasks
and estimated costs.
Based on its contribution to the strategic framework, each improvement can be
assigned a priority and a completion time estimate. These two factors are used
to generate an implementation schedule and a budget.
Battelle's implementation plan estimates that the five recommendations grouped
into the first phase of improvements can be implemented over a schedule of 121
weeks for an estimated cost of $769,000. These are preliminary figures based
on current understanding of the work required. The cost estimates are over and
and above the current EOB budget and are predicated on contracting for all
model enhancement and documentation work. EOB resources would be reallocated
to perform on-going management functions generated by the recommendations.
1-8
-------
EVALUATION AND ASSESSMENT OF UNAMAP
PART II: TECHNOLOGICAL ASSESSMENT
November 1987
-------
TABLE OF CONTENTS FOR PART II
Page
1.0 INTRODUCTION II-4
1.1 Purpose 11-4
1.2 Organization 11-4
2.0 UNAMAP VERSION 6 TECHNOLOGY II-6
2.1 Contents of Version 6 II-6
2.2 Modeling Technology in Version 6 II-7
2.2.1 Air Pollution Meteorology II-7
2.2.2 Software 11-15
2.3 Data Technology in Version 6 11-16
2.3.1 Meterological Data 11-16
2.3.2 Source and Emissions Data 11-19
2.4 Computer Technology in Version 6 11-20
2.4.1 Processing 11-20
2.4.2 Data and File Transfer 11-21
2.4.3 Graphics 11-22
2.5 Summary: Technology Utilized by UNAMAP Version 6 11-22
3.0 DEVELOPMENTS AND ADAPTATIONS OF UNAMAP MODELS
TO CURRENT TECHNOLOGY 11-23
3.1 Modeling Technology 11-23
3.1.1 Air Pollution Meteorology 11-23
3.1.2 Software 11-24
3.2 Data Technology 11-25
3.2.1 Meteorological Data 11-25
3.2.2 Source and Emissions Data 11-26
3.3 Computer Technology 11-26
3.3.1 Processing 11-26
3.3.2 Data and File Transfer 11-27
3.3.3 Graphics 11-27
II-l
-------
Page
3.4 Summary: Developments and Adaptations of Air
Quality Models to Current Technology.......... .... II-28
4.0 TECHNOLOGY TO SOLVE CURRENT PROBLEMS... 11-29
4.1 Modeling Technology... 11-29
4.1.1 Air Pollution Meteorology 11-29
4.1.2 Software 11-30
4.2 Data Technology ... 11-31
4.2.1 Meteorological Data.... 11-31
4.2.2 Source and Emissions Data. 11-32
4.3 Computer Technology 11-32
4.3.1 Processing 11-32
4.3.2 Data and File Transfer 11-33
5.0 SUMMARY 11-35
II-2
-------
LIST OF TABLES AND FIGURES FOR PART II
Page
TABLE 1: GUIDELINE MODELS - APPENDIX A II-8
TABLE 2: GUIDELINE MODELS - APPENDIX B 11-9
TABLE 3: NON-GUIDELINE UNAMAP MODELS 11-10
TABLE 4: GUIDELINE NON-UNAMAP MODELS 11-12
FIGURE 1: METEOROLOGICAL DATA TRANSFER PROCESS 11-17
II-3
-------
EVALUATION AND ASSESSMENT OF UNAMAP
PART II: TECHNOLOGY ASSESSMENT
1.0 INTRODUCTION
1.1 Purpose
In 1973, The Environmental Protection Agency began making air quality
simulation models available through the User's Network for Applied Modeling of
Air Pollution (UNAMAP). Since then, models have been added and deleted from
the UNAMAP set. Certain preprocessors and postprocessors have been added.
Originally available only on a UNIVAC 1100 computer, the models have been
converted by users to run on a variety of other computers. The purpose of this
report is to assess the technology utilized by UNAMAP and to project the
technological environment in which UNAMAP will function most effectively in the
future. The information will be used in the subsequent stages of the
Evaluation and Assessment of UNAMAP, and will provide evaluation criteria for
system alternatives and final recommendations.
1.2 Organization
This report is organized into three main sections, each representing a distinct
phase in air quality model development. The first section will discuss the
technology as it is utilized by UNAMAP Version 6. The second section will
describe the technologies being used by models currently under development, and
will address the adaptations being made to UNAMAP to utilize current
technology. The third section will project the technological requirements for
models which will be needed to solve the current problems facing air pollution
meteorology and pollution control professionals.
Each section will address three technological components of an air quality
modeling system: the modeling software, the meteorological and emission source
data, and the computer system. The modeling software is the computer programs
which incorporate physical, chemical, meteorological, and statistical
II-4
-------
principles to predict the dispersion of air pollutants under historical
meteorological conditions. Each section of the report (Version 6, new
developments and adaptations, and emerging technologies) will describe the
modeling technology to be supported.
A variety of meteorological and emission data are required for the UNAMAP
models. The meteorological data required by the modeling software typically
include hourly readings of weather conditions. The data may be collected
either at the site to be modeled, or at the nearest climatologically similar,
first-order National Weather Service weather station which is generally an
airport or a military airbase. The emission or source data may include an
inventory of other air pollution sources in the area, measurements of proposed
stack dimensions, and engineering calculations on the volume of emissions to be
released by the facility. Each section of the report will address the
technological trends and requirements for these types of data.
The computer system technology related to UNAMAP includes processing,
communications, and graphics. Computer processing is the ability to manipulate
the modeling programs and data. Particularly important to UNAMAP are the areas
of FORTRAN compilers, processor speed for performing large numbers of floating
point calculations to the required number of significant digits, and cost of
executing the model. Communications technology assessment will focus on the
ability to access remote computers or to transfer data and results from one
location to another. Graphics technology will be assessed to determine how
hardware and software can provide flexible display capabilities for the UNAMAP
models.
Since the development of the first computerized air quality models in the
1960's, all components of the technology have changed dramatically.
Furthermore, they are continuing to change at a fast pace. This report will
concentrate on the differences in technology from one phase of model
development to the next.
II-5
-------
2.0 UNAMAP VERSION 6 TECHNOLOGY
This section addresses the modeling and computer technology utilized by the
publicly available, EPA-supported version of UNAMAP. An overview of the
contents of Version 6 and the rationale for the inclusion of these models is
followed by a description of the modeling and computer technology employed.
2.1 Contents of Version 6
Models acceptable for use in federal regulatory actions are listed in the
Guidelines on Air Quality Models (Revised) (EPA$ July, 1986). The Guideline
models are divided into two categories:
o Appendix A models which are preferred for specific regulatory
applications.
o Appendix B models which are acceptable for regulatory applications under
special circumstances
UNAMAP Version 6 contains 24 different dispersion models, a variety of pre- and
post-processors for the models, and test input and output data for each model.
UNAMAP includes models from both Appendix A and Appendix B as well as some non-
Guideline models.
Originally, UNAMAP contained only the dispersion models, but Version 6 contains
several processor programs which enhance the utility or capabilities of the
models. Processors are available for formatting meteorological data to meet
the model requirements, calculating running averages, plotting concentrations
on a grid, and eliminating the influence of calm conditions on pollutant
di spersion.
Version 6 contains two types of models: 1) those for regulatory use and 2)
newer models provided for inspection and evaluation. The selection of models
and supplementary programs included in UNAMAP varies from version to version to
reflect the needs of modeling applications. A variety of screening and
II-6
-------
evaluation models are included to meet the requirements of many air pollutant
dispersion problems.
2.2 Modeling Technology in Version 6
2.2.1 Air Pollution Meteorology
UNAMAP Version 6 contains numerous models because there is no all-purpose model
which is practical to use. The models differ in their ability to handle the
various aspects of the modeling process such as:
o type of pollutant being emitted and dispersed,
o distance at which concentrations can be predicted,
o height of points where concentrations can be predicted (relative to
height of source),
o number and type of pollution sources,
o averaging time of estimated concentrations (short term or long term).
However, some of the models actually perform duplicate functions. Some
agencies or groups advocate the use of one of these duplicate models over
another.
As mentioned earlier, UNAMAP Version 6 includes both Guideline and non-
Guideline models. Seven of the eight Appendix A models are in Version 6. The
one model not included is the Urban Airshed Model (UAM) which, due to its large
size, was excluded from the UNAMAP tape, but is available separately from NTIS.
A comparison of UNAMAP to Guideline models can be made from the following
series of four tables showing properties and capabilities of the various
models. Appendix A UNAMAP Version 6 models are listed in Table 1. Table 2
represents Appendix B UNAMAP models, and Table 3 contains the UNAMAP models
which are not included in the Guidelines. Table 4 lists those Guideline models
not in UNAMAP. In these tables, the phrase "primary pollutants" refers to
those pollutants which, for the purposes of modeling, will not undergo chemical
transformations in the atmosphere.
II-7
-------
IABU 1. UNAMAP HOOEI. OPtRAJlON CHARACHRISMCS
Guideline Models - Appendix A
CD
r
m
CTJ
cr
i t
~
m
O
m
to
i
in
PENDIX
Node!
AcronyB
Guideline BLP
Hoite 1 s
Appendix A CAUNE3
COH
CRSIER
ISCLT
ISCSI
HPIER
RAH
Model
G.P.
G.P.
G.P.
G.P.
G.P.
G.P.
G.P.
6. P.
PolluUnt
lypes
PrlB.
Pria.
Prla.
Prla.
Prlii.
Prla.
Prla.
Prla.
Pluae Rise
Ireataent
SIS
No
Brlggs
Brtggs
Brlggs
Brlggs
Brlggs
Brlggs
Downwash
Treatment
Building
No
Stack
Stack
Stack I bldg.
Stack t bldg.
Stack
Stack
Legend
Cheatcal Physical
Transfornat ton Removal
Linear decay
No
Exp decay
Cxp decay
Exp decay
Exp decay
Exp decay
Exp decay
No
Deposit ion
t settling
No
No
Deposition
I settling
Deposition
t settling
No
No
Wind Speed
Profile Ireatnent
Exp.
No
Exp.
E.p.
Exp.
Exp.
Exp.
Exp.
Urban or
Rural?
Rural
Either
Urban
Either
Either
Either
Either
Either
Appendix A Appendix A of EPA Guideline on Air Quality Models (1966)
Briggs « Brlggs 0969, 1971. 1975) pluae rise equations
CBM-H Carbon-Bond !I
eap, exponential
exp decay * exponential decay
G.P. ' Gaussian Pluae
Prta. * Prtaary
StS Schulaan and Sclre (I960)
-------
DO
I ABU ?. UNAMAP HOOCL OPCRAIION CHARACURISIICS
Guideline Models - AppendI« 6
m
cn
i i
0
i
* i
z
CTi
m
i
i
>
-o
0
X
03
Model
Ai ronya
Guideline APRAC-1
Hcntels
Appendix B
HIUAV-2
LONG;
Ml SO
PDIF
PAL-?
SIIORU
PlUVUl
Model
lype
G.P.
G.P.
G.P.
Gaussian puff
II superposition
G.P.
G.P.
G.P.
Pollutant
lypes
PrlB.
PrlB.
PrlB.
SO;. S04.
NO,. HNO-3
Non- react.
PrlB.
ISP. NO,.
SO;. H;S04
°J
Plume Rise
treatment
No
No
BIB
Brlggs
Brlggs
BIB
Brlggs
*
Oownwash
Considered
No(|)
No
Stack
No
No
Stack
No
Legend
Chemical
transformation
No
No
Cxp. decay
Rate constants
No
t«p. decay
9 reactions
Physical
Removal
No
No
Oep. I
settling
Oep. t wet
removal
Settling t
dry dep.
Settling i
dep.
Dry dep.
Wind Speed
Profile treatment
No
No
txp.
Specified
Ixp.
Ixp.
No
Urban or
Rural?
Urban
Other
Rural
Urban
Other
Other
Append I B * Append in B of CPA Guideline on Air Quality Models (1986)
BIB BJorklund and Bowers (I9B?)
Brlggs Brlggs (1969. I9M. 1975) pluae rise equations
CBH II - Carbon-Bond II Mechanlsa
dep. > depositon
G.P. * Gaussian Pluae
e»p decay exponential decay
PrlB. Primary
non-react. non-reactive
exp. exponential
-------
IABLC 1. UHAMAP MODfL OPCRA1ION CHARACURISIICS
Non-Guideline UHAHAP Models
fr 1
1
1 »
o
1
CD
|
m
21
0
1
c:
C t
0
m
m
3:
I
"D
O
0
m
i
Model
Acronym
INPUff
PtM-2
IUPOS 2.0
PBH
PIPIU-2
HP JDS
ROADMA*-2.0
Model
Type
G.P.
G.P.
e.p.
Box
G.P.
G.P.
Finite .
difference
Pollutant
lypes
Prim.
Prim.
Prim.
Photo-
Prim.
Prim.
Photo-
Plume Rise Downwash
Treatment Considered
Brlggs
Calculated Ho
from wind t
temperature
profile
Homogeneous No
distribution
throughout iKffi
Srlggs Stack
Srlggs Stack
Chemical
Transformation
No
first order
decay
No
63-step
Hnetk
mechanise
No
Pollutant
decay
Ves
Physical
Removal
Deposition
t settling
Deposition
i, settling
No
No
No
Deposition
i settling
Wind Speed
Profile treatment
User furnished
No uariat Ion
with height
Exponential
Exponential
Similarity
theory
Urban or
Rural?
Urban
No dispersion
ioefi ic lent;
are used
Cither
Rural
Legend
-- BjorUund and Bowers 098?)
Brlggs (1969, 8971. 197%) pluM rise equations
CBM-I1 = Carbon-Bond Si Mechanise
dep. ° depositors
G.P * Gaussian Plune
e«p decay ' exponential decay
Prla. * Primary
non-react. * non-reactive
exp. « exponential
-------
CO
m
0
o
z
1
1 1
z
m
o
z
0
1
rr
TABLE 3.
Mode! Model Pollutant Pluae Rise
Acronya Type Types Treatment
VALLEY G.P. Prim. Brlggs
COMPLEX-! G.P. Prl«. Brlggs
UHAMAP MOOCL OPERATION CHARACTERISTICS
Non-Guideline UNAHAP Models
(continued)
Downwash Chealcal Physical Wind Speed Urban or
Considered Transformation Removal Profile Treatment Rural?
No Pollutant No None Rural
decay
No No No None Rural
legend
cr
>
atB < Bjorklund and Bowers {I98i)
Brlggs Brlggs (1969. 1971, 1975) plu
CBH-II Carbon-Bond II Mechanise
dep. deposlton
rise equations
G.P. Gaussian Pluae
exp decay * exponential decay
Prl«. PrlMry
non-react. « non-reactive
exp. exponential
-------
1ABU 4. HOOCl OPfRAIION CHARAl KRIS! ICS
Guideline Non-UNAHAP Hodels
1
3>
CD
1
m
-fc.
e o
C75
cr
t a
O
m
i
i t
°y
m
2^
O
z.
c.
2E
3»
^
T3
O
CD
m
co
Model
Acronya
Appendix A: DAM
Appendix 8: AQOM
ARRPA
COMPIER
ER1AQ
GMLINE
IMPACT
MPSDH
Model
Type
3-dtBenilonal
nuaerlcal, photo-
cheaical grid
G.P.
Gaussian Segaented
Pltme
G.P.
G.P.
G.P.
[ulertan finite
difference
G.P.
Pollutant
Types
03
Non-react.
SO; I S04
Prla.
Prill.
Prla.
SO;. S04,
NO,, MO;.
A.
U3
Prla.
Pluae Rise
Ireatnent
Brlggs
Briggs
Piece-
wise
continuous
Brlggs
Brlggs
Heated
exhaust
Brlggs
Briggs
Downwastt
Considered
No
Ho
No
No
No
No
No
Stack
Chealcal
Transportation
CBM-ll
Ho
Rate constant
No
Exp. decay
No
Species
aechanisas
No
Physical
Removal
Deposition
No
Dep.
Ho
Dep
No
Exp. decay
No
Ulml Speed
Profile Ireatnent
Specified
No
Model-
predlcted
E»P.
Exp.
No
Specified
User
designated
Urhan or
Rural?
UrLiin
Rural
Mixture
I llher
Either
Not
specif ic
Not
spec If ic
Hot
spet If Ic
Appendix A I 8 * Appendix A or 8 of EPA Guidelines on Air Qua lily Models (1966)
BiB * Bjorklund and Sowers (198?)
Brlggs Brlggs (1969. 1979, J9?5| plume rise equations
CBM-ll Carbon-Bond II
dep. depositon
G.P. - Gaussian Plume
exp decay * exponential decay
Prta. - Prlaary
non-read. ' non-reactive
esp. - exponent la?
-------
TABU 4. MODEL OPERATION CHARACTERISIICS
Guideline Non-UNAHAP Models
CD
r^
.E 4, CONTINUI
1 ' 1
0
« e
fD
cr
o
rn
i
i i
z
0
c:
-o
O
0
Model
Acronym
MTDD1S
MULTIMAX
PLMSTAR
Model
Type
Variable trajec.
Gaussian puff
G.P.
Lagranglan photo
chemical
Pollutant
Types
Prim.
Prim.
Oj, NO,.
SB,. CO i
reactive
Plume Rise
Treatment
Brlggs
Brtggs
Brlggs
\IUII
Oownwash
Considered
No
No
No
I IIHJCUJ
Chemical
Transformation
Exp. decay
No
Atkinson (1982)
photochealctl
mechanism
Physical
Removal
Oep. I wet
removal
No
Dry dep.
Hind Speed
Profile Treatment
Linear
Exp.
Measured
I grldded
Urban or
Rural?
Not
specific
Either
Either
hydrocarbons
PLUME S
PPSP
RAOM
Appendix A i B Appendix A
BIB - Bjorklund and Bowers
firlnnc . Rrtnnc / IQhQ IQ7I
G.P.
G.P.
Random Walk
Lagranglan
dispersion
Prim.
Prim.
Inert or
exp. decay
Brlggs
Brlggs
Brlggs
or B of EPA Guidelines on Air Quality Models
(1982)
tQ7*kt nlitmmm) ric* mi * t IAHC
No
No
No
Legend
(1986)
Exp. decay or
03 limiting
No
Exp. decay or
formation
Exp. decay
No
Dry dep.
Extrapolated
or calculated
Exp.
Based on
roughness t
stability
Either
Either
Either
G.P. Gaussian Plume
exp decay * exponential decay
Prla Prlajlru
CBM-ll - Carbon-Bond 11 Mechanism
dep. depositor!
non-react. - non-reactive
exp. exponential
-------
TABLE 4. MODEL OPERATION CHARACTERISTICS
Guideline Non-UNAHAP Models
(continued)
1
m
o
o
1
m
o
e o
CD
cr
» 3
O
m
i
m
o
Z
i
>_
-o
o
0
Model
Acronym
Model Pollutant Plume Rise Dowmash Chemical Physical Ulnd Speed Urban or
Type Types Treatment Considered Transformation Revival Profile Treatment Rural?
RPM-II Reactive plume Oi, NO,, BHggs No CBH-M No Ho
SO, and
reactive
hydrocarbons
RTM-II EulerUn grid & SO; and Brlggs No Linear SO; Dry and wet Observed winds
lagranglan puff sulfate oxidation deposition
SCSTER
TCM-Z
TEM-8
Models 3141
£ 4!4I
Visibility
Model
ERT Visibility
Appendix A 1 B * Appendix A or B
G.P. Prim. Brlggs No Exp. decay Exp. decay Exp.
G.P. Prim. Brlggs No Exp. decay Exp. decay Exp.
G.P. Non-react. Brlggs Stack Exp. decay Exp. decay Exp.
G.P. Non-react. Brlggs Ho Ho Ho Exp.
G.P. TSP. N0;g Brlggs No First order Dry dep. No
sulftte.
nitrate
Legend
9f EPA Guidelines on Air QualUy Models (1986) G.P. - Gaussian Plume
either
Either
Rural
Urban
Either
Either
Rural
BIB Sjorklund and Bowers {1962}
Brfggs Brlggs IJ%8, 1971, 1975$ plus* rise equations
C8M-i! * Carbon-Bond II Mechanise
dep. °> depositor!
e«p decay * exponential decay
Prla. - PrlMry
non-react. « non-reactive
exp. exponential
-------
Perhaps the most remarkable property shown in the tables is that almost all of
the dispersion models in UNAMAP use Gaussian diffusion where the concentration
of a pollutant in the cross-section of a plume resembles the normal (bell)
curve.
A glance through Table 4, however, shows that some non-UNAMAP models use
different diffusion technologies. The IMPACT model simulates dispersion by
using finite-difference methodology to solve a conservation of species
equation. The RADM model incorporates a random-walk dispersion. The Guideline
model which departs most radically from the Gaussian method is the Urban
Airshed Model (UAM). UAM is an urban-scale, three-dimensional model for
computing ozone concentrations during one or two day episodes. The model
employs numerical methods over a three-dimensional grid.
However, as shown in the tables, the UNAMAP Version 6 models have incorporated
advances in other facets of air quality modeling technology. Some Version 6
models handle any number and mix of point, area, and line sources of pollution.
Some models incorporate methods to deal with hills or valleys. Some of the
more sophisticated models are designed to simulate physical removal or chemical
transformation of pollutants as well as dispersion.
2.2.2 Software
The Version 6 models are all written in ANSI Fortran 77 as implemented on the
Sperry 1100 Series. There are approximately 90,000 lines of code in all, with
individual models containing between 388 and 13,827 lines. Although the
accuracy and mathematical soundness of the algorithms has been verified through
extensive testing, there is no programming or documentation standard for the
models. Many have been modified several times. All models do include sample
input and output data sets for testing. The cost of running the Version 6
models against their sample data on the EPA UNIVAC ranges from $.58 to $263.32.
All but 8 of the models run for less than $5.00. These dollar amounts reflect
the 1986 charge algorithms.
11-15
-------
2.3 Data Technology in Version 6
2.3.1 Meteorological Data
Most Version 6 models use meteorological data provided by the National Weather
Service (NWS). Surface data are collected hourly while mixing height data are
produced twice daily. These data are collected at each of the approximately
300 primary weather stations located at airports plus several hundred other
sites designated as secondary, military or cooperative weather stations. Data
have been gathered at sites for various periods of time beginning in the late
1940s.
The data are compiled and made available in computerized form through the
National Climatic Data Center (NCDC), as illustrated in Figure 1 on the
following page. Data are usually aggregated annually, i.e., 8760 hourly
readings per year. Several Version 6 models can also use NCDC's STAR
(Stability Rose) program which provides a joint frequency distribution of the
occurrence of different categories of wind speed, wind direction, and
atmospheric stability.
The technology exists to collect meteorological data at the site under study.
For several reasons, onsite data are not widely used. First is the expense of
collecting the data, and, since onsite data are rarely reused, costs are not
reduced by providing the data to others. Also, time becomes a factor because
one year must be spent to collect one year's worth of data. Another problem
with onsite data is instrument reliability which results in too many missing
observations.
However, some of the Version 6 models require onsite data to portray the most
accurate conditions. MPDA is a data processor available in Version 6 which
accepts both NWS and onsite data. Currently, the output of MPOA may only be
used with the TUPOS model. Future versions of MPDA will provide output usable
by other UNAMAP models.
11-16
-------
National
Heather
Service (NHS)
collects data
NWS sends data
to NCDC
NCDC sends data in
response to requests
Requestors receive and use data:
f
States/Regions
EPA
Computer
Processed
Data
1
/Z/4 \^\
o
FIGURE 1: METEOROLOGICAL DATA TRANSFER PROCESS
11-17
-------
The NWS data could be greatly improved from an air quality modeling
perspective. Models cannot handle periods where data are missing - the period
of record must be complete. Users report that data from some NWS sources can
have missing values which limit its usefulness. Also, the raw data from NWS
are not immediately usable in the models but must be processed before use.
NWS data need to be processed in various ways. One type of processing is
needed to produce certain input variables needed for modeling that are not
directly measured. For example, the data contain no direct measure of
atmospheric stability which is an important characteristic for dispersion
models. Also, the data are collected at only one altitude (the standard is ten
meters above the ground) which does not allow consideration of variations in
conditions at different heights. Winds at other heights must be estimated by
extrapolation techniques. In comparison, onsite data collection can be
designed to meet the specific data requirements for the application.
The second type of processing is required due to a change in NCDC data format.
NCDC data was previously digitized in a card image format where one line
contained one hour of data. The new format, called compact format, has the
data organized by data element. NCDC will provide the user with a tape
containing all the data or they will process it to pull out selected data
elements.
NWS data for a specific time period present another problem to modelers. Until
1964, observations were made each hour by NWS, and the data were entered for
each interval. From 1964 until the middle of 1981, although observations were
still made each hour, only every third hour was digitized for most stations.
In 1981, the digitized interval returned to one hour. Users needing data in
this time period must pay an additional amount to get the other two hours
digitized by NWS.
11-18
-------
2.3.2 Source and Emissions Data
Two kinds of data about the potential generation of pollutants in the area
under study are required for most of the models in Version 6: an emissions
inventory and the emissions from the actual source. An emission inventory of
all sources in the area is often necessary to establish background
concentrations. These data are typically gathered by a voluntary, unverified
response to a questionnaire administered by the state pollution monitoring
agency. Although the data are usually available in automated form, they tend
to be incomplete, obsolete, and not in a standardized format from one state to
another.
There are three types of emission sources: point, area, and line. To provide
the emissions data needed to model a point source, the expected volume of
pollutants to be emitted by the pollution source under consideration must be
calculated. Emission data will include the emission rate in mass per unit time
and information about the source (e.g., stack height, stack diameter, exit
velocity, and stack temperature for a point source). The mass per unit of time
is generally calculated for the peak capacity of the pollution generating
facility. The emissions data can be derived from source testing, engineering
mass balance calculations, or generic emission factors applied to process
information. If the modeling is being done for a proposed source, the
emissions are based upon engineering design information.
For an area source, it would be difficult to obtain the emissions from each
piece of the source individually. However, the collective impact of the area
is important. For example, a coal pile is a true area source with the
particulates arising from the entire pile constituting the emissions.
Emissions from fuel combustion by residences, commercial establishments, small
industries, motor vehicles, railroads, and aircraft constitute the majority of
area source emissions although dust may also be a factor in some cases.
Several types of data are used to calculate the emissions for these sources
such as fuel sales, number of aircraft arrivals and departures, records of
construction sites, and agricultural activities. These data can be obtained
from various federal, state, and county records.
11-19
-------
Line sources are linear strips such as highways, streets, and airport runways.
For highways and streets, the emission calculations require data on the vehicle
miles traveled, the width of the street or highway, the number of lanes, and
the distribution of traffic among vehicle types. For airports, data is needed
on the number of airplane arrivals and departures, the distribution of traffic
among different types of aircraft, and the number of engines for each aircraft
type. The information is used with the appropriate emission factors to
calculate emission rates for unit length of the highway or runway.
2A Computer Technology in Version 6
2.4.1 Processing
The resident computer for all UNAMAP Version 6 models is EPA's UNIVAC which
resides at the National Computer Center (NCC) in Research Triangle Park, North
Carolina. The Sperry 1100 series architecture is approximately 20 years old,
and the first EPA UNIVAC machine was purchased in the early 1970s. The
operating system is OS 1100, Version 39 R3. There are two CPUs with 14Mb of
main memory, and 10 gigabytes of online disk storage* There are 106
synchronous communications ports which can be configured as dial-up or
dedicated and 96 asynchronous dial-up lines. The TYMNET communications network
is available.
To facilitate scientific computing of the type required for UNAMAP, the UNIVAC
has Release 11 of the ANSI FORTRAN compiler which was installed approximately
seven years ago. Floating point (decimal) calculations are handled by machine
language instructions which perform one floating point double precision add in
120 nanoseconds.
NCC charges only $1050 for an hour of CPU time. While this rate sounds high,
it compares quite favorably with some other government charges where a lower
CPU rate is offset by a memory charge rate.
11=20
-------
2.4.2 Data and File Transfer
Data communications technology is not used to transfer the modeling software or
the meteorological data. The National Technical Information Center (NTIS)
makes the UNAMAP software available to potential purchasers. NTIS staff
members copy the programs onto 9-track magnetic tape, take incoming orders,
ship the tapes, and announce the new versions of UNAMAP. Users purchase the
magnetic tape which contains all UNAMAP models. There is no provision for
buying only some of the models. A prospective UNAMAP purchaser can get the
NTIS order number or other information by using the online NTIS database which
is available on BRS, DIALOG, Mead, and SDC. There is no electronic
distribution and announcement procedure used by NTIS for UNAMAP-
Nationwide meteorological data are made available by NCDC to UNAMAP users on
standard 9-track computer tapes in a variety of formats. A fixed length
format, card image format and variable length element format are available in
ASCII and EBCDIC. For most sites, the entire period of record will fit on one
tape. Frequency distributions in paper copy are available.
The cost for NCDC data are dependent on the input/output requirements needed to
make the tape. For one year of data requiring one input tape and one output
tape, the cost is approximately $100.00. However, the average cost of an order
from NCDC ranges from $200 to $500. A written request is not required,
although the order must be pre-paid by check or credit card and data are
generally received within one to two months after the request is mailed.
Requests for current data may take longer.
The EPA UNIVAC at the National Computer Center (NCC) maintains a large amount
of meteorological data online which have been processed by the Office of Air
Quality Planning and Standards (OAQPS). Regional offices can order data from
NCDC, and ask OAQPS to process it. The processed data are then readily
accessible to all modelers who use the NCC UNIVAC.
11-21
-------
2.4.3 Graphics
Very little graphics technology is utilized in UNAMAP Version 6. Some of the
models can produce a data file which is used to plot lines showing coordinates
of equal concentrations. These lines, called isopleths, are only produced in
dot-matrix or character form; they are not plotted in a continuous line.
There are no scale adjustments or variable information which can be added or
deleted from the graph. No terminal display of graphics is supported in EPA's
Version 6.
2.5 Summary: Technology Utilized by UNAMAP Version 6
In Version 6, the technology applied to modeling has progressed, while that of
the computer systems used to run the models has not. Although largely based on
the original Gaussian dispersion mathematics, the UNAMAP models have advanced
by becoming more sophisticated and consistent and by increasing their usability
through adaptation and new processors. On the other hand, the UNIVAC 1100
computer technology utilized to support UNAMAP represents no advances since
UNAMAP was begun in 1973.
11-22
-------
3.0 DEVELOPMENTS AND ADAPTATIONS OF UNAMAP MODELS TO CURRENT TECHNOLOGY
The field of air quality modeling is still quite dynamic. New, sophisticated
models have been developed to model more specialized or more complex problems.
Existing models have been adapted to a variety of computers. Many ease of use
features have been added to them. The adaptations and processors have been
written by users, consultants, and third-party marketers. Together, these new
models and features of existing models are important indicators of the
technological direction of UNAMAP.
This section will describe the technologies used by models currently under
development and adaptations being made to UNAMAP models.
3.1 Modeling Technology
3.1.1 Air Pollution Meteorology
New models are being created both within and outside EPA. One example of EPA
design is the Complex Terrain Model Development project which is scheduled for
completion in 1987. It is expected that information from this study will help
to remedy UNAMAP's lack of an adequate, refined Guideline model for complex
terrain. Two UNAMAP models, VALLEY and COMPLEX I, are commonly used for
screening purposes in complex terrain. Most states recommend VALLEY over
COMPLEX I because COMPLEX I requires onsite meteorological data. Moreover,
VALLEY gives conservative predictions which helps to protect air quality.
Modeling efforts outside EPA include New York's development of a subroutine for
ISCST used to predict pollutant concentrations at receptors that lie above the
top of the shortest stack being modeled.
Gaussian models are not applicable to variable conditions such as short-period
wind fluctuations and emergency response situations. One new type of model,
the "puff" type, overcomes some of these limitations. It concentrates on the
trajectory taken by the emissions and predicts the pollutant concentration
which occurs within an isolated portion --a puff of the plume as it is
11-23
-------
transported and diffused. The INPUFF model, which is part of UNAMAP, is an
example of this type of model.
To fill another need, the Urban Airshed Model has incorporated an atmospheric
chemical transformation subroutine which allows the modeler to choose a
chemical kinetic mechanism to depict the photochemistry of hydrocarbon-nitrogen
oxide-ozone interactions.
3.1.2 Software
Since the UNAMAP source code is readily available to the modeling community,
modelers are free to adapt them for problems specific to a locality.
Modifications include the addition of new or revised algorithms for calculating
plume rise, stack downwash, and pollutant decay rate. Some modeling groups
have made extensive modifications to the models to adapt them to their specific
requirements.
Pre- and post-processors also adapt the models to the needs of a state or an
industry as well as promote ease of use. Post-processors aid in calculation of
human exposure values for industries where risk assessment is necessary. Many
pre-processors are used to prepare data to meet the varying requirements of the
models.
Data output by the models can be cumbersome and voluminous. Many post-
processors are used to summarize model results, reformat the output to
highlight important points, and to facilitate analysis of the results. They
are also used to produce graphic output for presentation to non-modelers.
Processors to ease the task of data input have been developed by individuals;
the best known user interfaces are those provided by such third-party marketers
as Trinity Consultants and Bowman Environmental Engineering. Their interfaces
prompt the user for the necessary data. The user does not have to know the
various data input formats required by the different models.
11-24
-------
3.2 Data Technology
Evolution of software is not the only area of progress in modeling. Pre-
processed and site-specific data improve modeling accuracy by increasing the
quality and representativeness of the data used as input to the models.
3.2.1 Meteorological Data
Users may avoid processing raw NWS meteorological data by purchasing pre-
processed data. NCDC will pull selected data elements from the compact format,
but processing by NCDC is expensive. Some consultants, such as Trinity and
Bowman, will process the data into the required format for modeling and provide
it to the user on diskettes. The Office of Air Quality Planning and Standards
(OAQPS) will provide some data on 9-track tape at the request of the Regional
Offices. Some states (e.g., Ohio, New Jersey, and Wisconsin) make the pre-
processed data available on floppy diskettes at a nominal charge. A year of
processed data can fit on one diskette.
In some areas, airport data are not representative of the site. Despite the
drawbacks of cost and difficulty of input, onsite data are used when either the
state or the industry decides that site-specific data are necessary for a true
picture of meteorological conditions. Some major sources have their own
meteorological stations, but the costs for set up and maintenance are too high
for most sites. However, the technology needed for data collection is becoming
cheaper as microcomputer costs drop, and such stations may be more widely used.
Onsite data are not as easily used with most UNAMAP models as are NWS data, but
the improvement in representativeness is frequently worth the cost. When the
MPDA pre-processor produces output usable by models other than TUPOS, it will
be easier to input onsite data into the UNAMAP models.
Site-specific data are not frequently reused due to the lack of a mechanism
which would make the data accessible and promote their use. The potential for
re-use of onsite data is limited by the requirement for climatological
11-25
-------
similarity. The decision on site similarity is based on professional analysis
of basic topographical conditions such as size, type, and orientation of
geographical features. No quantitative method is available to make this
decision.
3.2.2 Source and Emissions Data
Emissions data are now more complete since states are continually improving
their emissions inventories. The National Emissions database is available at
Research Triangle Park, and is updated quarterly by the states. However, there
are some problems with the accuracy of the emissions data. There are errors in
such data as estimated emissions and stack heights, and there is little
consistency across the country in how to report fugitive emissions rates.
3.3 Computer Technology
3.3.1 Processing
The UNAMAP models have been converted to run on a variety of computers. The
conversions have been mainly done, however, by modeling groups using a variety
of FORTRAN compilers. It is not difficult to convert the models to run on
Digital Equipment VAX computers because of the similarities between the VAX and
the UNIVAC FORTRAN compilers. It is also easy to convert the models for the
IBM microcomputer. Conversion to the IBM mainframe is the most difficult,
while there is an intermediate level of difficulty with the Burroughs.
Some states and regions make their converted version of the models available to
groups within their area, and there are two conversions of the UNAMAP models
which are commercially available. HMM, in Waltham, MA, sells an IBM mainframe
version. Trinity Consultants, Bowman Environmental Engineering, and Clary
Associates sell IBM microcomputer versions of the models. The microcomputer
versions also use a variety of FORTRAN compilers such as Microsoft and Ryan-
McFarland.
11-26
-------
3.3.2 Data and File Transfer
Currently, communications technology is mainly used to access either the
EPA NCC computer or Regional Office computers. There is very little electronic
transfer of data or model code.
3.3.3 Graphics
Graphics technology, however, is widely used. Graphics are particularly
important in communicating the results of the model to non-modelers. Both
microcomputer and mainframe users of UNAMAP produce graphic output. For
mainframe users, post-processors are used to provide the model output in a file
for use with graphics software packages.
The microcomputer versions of the models have graphics as part of the package.
The most common graphics output is production of isopleths which are connected
points of equal concentrations scaled to a USGS 7.5 minute topographic map.
Isopleths in different scales and 3-dimensional representation of
concentrations are available. Both dot-matrix printers and graphics plotters
are used for output, but color displays are less common.
11-27
-------
3.4 Summary: Developments and Adaptations of Air Quality Models
to Current Technology
Innovations by UNAMAP users have resulted in adaptations to the modeling
software, more sophisticated data collection methods, and the increased use of
new computer technology. This activity reflects the changes in the typical
modeling situation since the UNAMAP models were developed. Modelers have
responded to situations that are different or more complex than those for which
the software was designed.
Also, the typical model user has changed since the advent of UNAMAP. A user is
not necessarily a "modeling expert". Some people use the models only a few
times a year and have to relearn them each time. Others, while expert in the
meteorological aspects of the models, do not have the computer expertise
necessary to use the mainframe versions easily. Individual modelers, state
agencies, and third-party marketers are helping to meet today's modeling needs.
-------
4.0 TECHNOLOGY TO SOLVE CURRENT PROBLEMS
The adaptations and new developments discussed in the previous section have
arisen out of the user community's need to deal with operational problems
encountered in typical air quality modeling applications. Model developers are
now starting to turn their attention to more challenging environmental
problems. This section will project a picture of the technological milieu in
which today's and tomorrow's models will function.
4.1 Modeling Technology
4.1.1 Air Pollution Meteorology
Since its beginning in the late 1960s, air quality modeling has been a useful
and important tool in protecting the environment. Because of the acceptance of
air quality modeling, models are now being asked to meet the needs of even more
complex, real-world situations which need representation.
Regional models are needed by county, regional, and state agencies to study
dispersion and transport of pollutants over larger distances. Regional models
must handle data from several areas along the path of the plume. Multiple wind
speeds, wind directions, and atmospheric stability situations must be
represented. Trajectories are plotted to discover the destination of the
pollutant and its concentration.
Because of the complexity of the situations being modeled, regional models will
be larger and more complex. Although interest in acid precipitation will
increase the application of regional models, these models will not have as many
users as those for geographically localized sources. Their use will be limited
by the smaller number of situations needing such a model, the model's large
size, and the increased difficulty of use. To meet user requirements within
EPA for the execution of these models, the Meteorology Division is establishing
a program for advanced model operation and analysis. The program will provide
the models, computer resources, and the personnel to run them.
11-29
-------
Complex terrain models will more realistically represent geographical areas
where terrain is a factor. These models will need to solve problems such as
valley stagnation, where the wind speed is zero, and treatment of the leeside
of terrain obstacles. The early UNAMAP models would not handle an obstacle
higher than the lowest stack. Later models, such as COMPLEX and VALLEY,
allowed the plume to rise over the obstruction.
Modeling has great potential for use in such areas as toxic gases, radiological
pollutants, emergency release situations, and offshore sites. For toxic gases,
models are being used to predict the dispersion of fugitive emissions as well
as emissions from known point sources. Some toxic gases, however, are heavier
than air, and current Guideline models cannot be used for them.
Modeling of instantaneous releases of gaseous pollutants has increased. The
releases may be from a spill or other accidental occurrence where the emission
is not continuous. These models may include the release rate plus vaporization
activity so the model depicts physical changes as well as dispersion.
Offshore sites and those on shorelines have special dispersion characteristics
due to the differences in stability over land and water. More realistic models
are being developed to represent dispersion of emissions from offshore sites.
4.1.2 Software
The modeling software needed to handle current and future modeling situations
must meet certain requirements. The continued use of a standard language such
as ANSI FORTRAN 77 is necessary for conversion to the various hardware systems
available now and in the future.
Changes in programming methods can affect other modeling needs. Structured
programming techniques will promote better internal program documentation and
allow easier modification of the models. Run-time efficiency is being improved
to facilitate the use of larger models and microcomputer processing*
11-30
-------
Changes will also be seen within the modeling software itself. The software
will include more intricate equations to adequately represent the complex
situations being modeled. New functions such as statistics and health-risk
assessment factors will be added.
Consistency of input and output among the models will need to be increased as
the number of non-expert modelers increases. Standard data input formats will
ease the data-entry bottleneck which currently exists. The format of the model
output will become more standardized, and the models will produce reports which
are easily interpreted and quickly generated.
4.2 Data Technology
4.2.1 Meteorological Data
The data technology must improve to meet the demands of air quality modeling in
the future. This improvement will result in better data for current models and
will meet the data requirements of new models. Improvements in the quality and
quantity of both onsite and NWS data collected for current models will increase
the accuracy of the models' predictions. The current trend of collecting more
onsite data will continue, and the technology for the collection and processing
of this data will improve.
NWS data will continue to be important in modeling. The data collected,
however, will change. Improvement will come through the use of automated data
gathering instrumentation. More data will be collected per site, and at more
sites, than is possible with present NWS staffing. Modeling requires better
representation of wind conditions, such as an hourly wind value which is an
average of the winds during the hour instead of a single reading. Hourly
measurement of the mixing height, as is possible with new monitoring equipment,
would be a considerable improvement over interpolations from measurements taken
every 12 hours. Measurement of the standard deviation of horizontal wind
direction fluctuations can be used directly to calculate horizontal dispersion.
11-31
-------
NWS has various plans for continuing to improve their data. The plans include
updating the following activities: observation procedures, reporting
processes, and archiving functions.
4.2.2 Source and Emissions Data
Emissions data will also need improvement to increase model accuracy.
Emissions inventories will expand to include more substances, and the accuracy
of reported data must increase.
4.3 Computer Technology
4.3.1 Processing
Processing for air quality modeling will continue on the current path toward
more distributed processing. Most of the day-to-day modeling activities will
be done on microcomputers while the large regional and complex terrain models
will be run on mainframes and supercomputers. Other modeling activities such
as software development and time-sharing will be done on more user-friendly,
scientific machines such as the VAX.
Microcomputers have already begun to replace mainframes for running the
standard models. Improvements in microcomputer technology are producing faster
equipment with increased disk storage at affordable prices. The technology is
available to process the most common modeling runs on microcomputers using
little more total elapsed time than that needed to submit the job and receive
the results via telecommunications facilities to a large system. Some of the
Version 6 models are available on microcomputers, and others will be converted.
Mainframe computers will continue to be a necessary processing environment for
some types of modeling situations. While IBM mainframe computers are not the
specialized, scientific machine best-suited to modeling activities, their
general popularity makes them an attractive target for larger models such as
the regional models.
11=32
-------
4.3.2 Data and File Transfer
The activities necessary to acquire data require no modeling expertise but can
require a great deal of time. Future utilization of electronic communications
will allow modelers to share data and modeling software more easily.
Electronic bulletin board software allows multiple users to dial in to a system
which will provide news, "mailbox" communications with other users, and the
ability to download data from the system. If such a system were operated by
EPA, users would be able to look at a list of data files available from OAQPS
and from other users.
Data transfer could be facilitated in two ways. For data only available from
other users, the owner plus a description of the data would be listed, and the
requestor would contact the owner directly or through the electronic mail
facility. After any payment or other issues were worked out between them, the
owner could upload the data onto the system for downloading or data diskettes
could be mailed. For onsite data, the costs for collection are high, both in
time and money. Sharing this data would allow some cost recovery for onsite
collection as well as provide others access to more representative data in a
timely manner. The data would remain the property of the collector, but it
could be shared with others willing to pay a price set by the owner. Onsite
data has been used by modelers other than the collectors of the data, but there
is no method currently in place to streamline the process. The most cost-
effective method would be for EPA to provide a mechanism by which the modeling
community could share information on what data are available. Permitting
agencies may have a need to access onsite data files used in the permitting
process in order to verify their accuracy. The bulletin board, however, is not
seen as the best mechanism for providing that type of access.
Certain NCDC or OAQPS data could be available for downloading by users through
the bulletin board. One or more telephone lines into the bulletin board could
be reserved for downloading so that other, less time-consuming, functions would
not be adversely affected. Electronic transfer of NWS data would ease several
data problems. The first problem is the time modelers spend waiting for the
11-33
-------
arrival of data. By downloading data through the bulletin board, the data
would be available within a few hours or less depending on the amount of data
transferred. Even if data diskettes were sent through the mail, the time spent
waiting would be less because the request process would take less time.
Second, sharing of data by users (both NCDC and onsite data) would be easier.
Shared NCDC data would lower the demand on NCDC staff.
A third problem addressed by electronic data transfer is the time and knowledge
required to use a preprocessor on a data file. Sharing of preprocessed NWS and
onsite data would make running the models quicker and easier, especially for
the novice or infrequent users.
In addition to electronic transfer of data, file transfer will also address
problems confronted by modelers. Infrequently used or new models could be
shared allowing all areas to serve their end-users most economically. File
transfer could provide corrections to models and documentation in a manner
which is more timely and economical than is currently available. The bulletin
board could also allow users to share modifications they have made to the
models or processors. Such sharing would encourage consistency and allow all
users to enjoy the benefits provided by the modifications.
11-34
-------
5.0 SUMMARY
All products follow a pattern of growth, the life cycle of the product, which
involves changes to the product, the users, and the market. UNAMAP has
followed the evolutionary stages common to such a cycle. The first stage,
typical of a new product, is a testing period where changes are made to bring
the program into line with the original specifications.
When the UNAMAP program began in 1973, the field of air quality simulation
modeling was still new. The testing period was made easier because the models
were a great improvement over previous methods, the modeling community was
small, and modeling needs were well-defined. The computer technology chosen
was the best available product which met the needs of the time. At this stage,
most users were "experts" in the product because there were few support
services available. During this period, the UNAMAP users and developers (often
the same people) worked together to improve the program.
In the product life cycle, the testing period is followed by one of adaptation.
During this period, memory of life without the product fades, and new
requirements emerge to be met with a wave of modifications. UNAMAP is
currently in this phase of the cycle.
As the field of modeling and the UNAMAP product matured, modeling began to be
applied to new situations. People are working around problems in UNAMAP by
developing processors and adapting the models to specific conditions. The user
community has shifted toward the non-expert user who is not willing or lacks
the skills necessary to provide his own support. A peripheral market of
services and products has grown around UNAMAP with consultants to advise or
perform the modeling activities, software conversions to other hardware, and
proprietary models.
By Version 6, the models had increased in sophistication and consistency while
processors and adaptations improved the usability of the software. However,
the computer technology for which UNAMAP models are designed is no longer the
11-35
-------
most appropriate. Many users want the models to run on their own
microcomputers or on local or regional hardware. Early in the period of
adaptation, it is difficult for the developers to identify the needs which will
be important to the community as a whole. Users, therefore, institute the
changes themselves or find an outside party to do it for them.
The third stage of the cycle is the emergence of a mature product usable by
those with a variety of skill levels and with an increased flexibility and
range of application. The objective of this project is to provide the
information necessary for UNAMAP to move into the third stage. The modeling
software must continue to change to meet current needs, and the range of
applications must increase. Although some regional and complex terrain models
currently exist, refined, fully-developed models are required. Models for
toxic materials, emergency release situations, and offshore sites must be
included.
The UNAMAP user will continue to be non-expert. The proliferation of user-
friendly, microcomputer products will lead to increased expectations which the
UNAMAP software must meet. There must, also, be more consistency of data input
and output among the models to facilitate the novice or infrequent user.
Computer technology will continue its trend toward distributed processing with
microcomputers used for an increasing amount of modeling activity including
more graphics. User expectations will also require that data gathering and
file transfer technologies in UNAMAP match those available in other areas.
The UNAMAP models have been and will continue to be important tools for
protecting air quality. The final report of this project will provide the
information necessary to make them responsive to the needs of the user community.
11-36
-------
EVALUATION AND ASSESSMENT OF UNAMAP
PART III: INTERIM REPORT
November 1987
-------
TABLE OF CONTENTS FOR PART III
Page
1.0 INTRODUCTION 111-7
1.1 Background 111-7
1.2 Data Collection and Analysis III-7
1.3 UNAMAP User Profile III-9
1.4 Analysis of Problems/Suggestions 111-13
1.5 Overview of Problem Areas 111-15
2.0 MODEL ACCURACY II1-17
2.1 Importance of Accuracy 111-17
2.2 User Suggestions 111-18
2.3 User Concerns and Alternatives 111-18
2.3.1 Concerns 111-18
2.3.2 Alternatives 111-19
2.4 Evaluating Accuracy Alternatives 111-19
2.5 Constra 1 nts 111-20
3.0 DOCUMENTATION 111-21
3.1 Importance of Documentation 111-21
3.2 User Suggestions 111-23
3.3 User Concerns and Alternatives 111-23
3.3.1 Concerns 111-23
3.3.2 Alternatives 111-24
3.4 Evaluating Documentation Alternatives 111-25
3 = 5 Constraints 111-25
4.0 USER SUPPORT 111-26
4.1 Importance of Support 111-26
4.2 User Suggestions 111-26
4.3 User Concerns and Alternatives 111-28
4.3.1 Concerns 111-28
4.3.2 Alternatives 111-31
III-l
-------
Page
4.4 Evaluating Support Alternatives....... 111-32
4.5 Constraints I!J-32
5.0 DATA COLLECTION .. IH-33
5.1 Importance of Data Collection.. 111-33
5.2 User Suggestions 111-35
5.3 User Concerns and Alternatives 111-35
5.3.1 Concerns - 111-35
5.3.2 Alternatives 111-36
5.4 Constraints 111-37
6.0 DATA INPUT. 111-38
6.1 Importance of Data Input. 111-38
6.2 User Suggestions... .......... 111=39
6.3 User Concerns and Alternatives. 111-39
6.3.1 Concerns 111-39
6.3.2 Alternatives............ .... 111-39
6.4 Evaluating Data Input Alternatives...... .......... 111-40
6.5 Constraints II1-40
7.0 HARDWARE COMPATIBILITY.. 111-41
7.1 Importance of Hardware Compatibility.................. 111-44
7.2 User Suggestions. 111-44
7.3 User Concerns and Alternatives. ......... 111-44
7.3.1 Concerns........... . ................... 111-44
7.3.2 Alternatives.. 111-44
7.4 Evaluating Hardware Compatibility Alternatives. 111-48
7.5 Constraints II1-48
III-2
-------
Page
8.0 OTHER AREAS OF CONCERN II1-49
8.1 Response Time II1-49
8.2 Output II1-50
8.3 Specific Models 111-51
9.0 GENERAL CONSTRAINTS 111-56
9.1 Technological Constraints: Computer Resources 111-56
9.2 Economic Constraints: Budget and Staff 111-57
9.3 Regulatory Constraints 111-57
9.4 Future Outlook for Air Quality Modeling and UNAMAP 111-57
10.0 RECOMMENDATIONS II1-60
10.1 Recommendation 1: Establish an Electronic Bulletin
Board II1-60
10.1.1 Description II1-60
10.1.2 Benefits II1-60
10.1.3 Time and Cost Estimates 111-61
10.1.4 Alternatives Not Chosen 111-62
10.2 Recommendation 2: Produce, Distribute, and Support
a Series of End-user Documentation 111-63
10.2.1 Description 111-63
10.2.2 Benefits of End-User Documentation II1-63
10.2.3 Time and Cost Estimates II1-63
10.3 Recommendation 3: Provide UNAMAP Code, Data, and
Documentation for Multiple Computers 111-64
10.3.1 Description 111-64
10.3.2 Benefits 111-65
10.3.3 Disadvantages 111-67
10.3.4 Time and Cost Estimates 111-67
III-3
-------
Page
10.4 Recommendation 4: Improve the Accuracy and Technology
of Models Included in UNAMAP .. II1-68
10.4.1 Description II1-68
10.4.2 Benefits. 111-69
10.4.3 Time and Cost Estimates II1-69
10.5 Recommendation 5: Develop a Consistent Set of User
Interfaces II1-69
10.5.1 Description II1-69
10.5.2 Benefits 111-70
10.5.3 Disadvantages 111-70
10.5.4 Time and Cost Estimates 111-70
10.6 Recommendation 6: Consolidate all Support for UNAMAP
Models II1-71
10.6.1 Description 111-71
10.6.2 Benefits 111-71
10.6.3 Time and Cost Estimates II1-72
10.7 Recommendation 7: Establish a Meteorology Data
Clearinghouse ... . ........................... 111-72
10.7.1 Description 111-72
10.7.2 Benefits 111-73
10.7.3 Time and Cost Estimates 111-73
10.8 Recommendation 8: Develop or Acquire Specialized Models
For Inclusion in UNAMAP.. 111-74
10.8.1 Description ..... HI-74
10.8.2 Benefits III-7S
10.8.3 Time and Cost Estimates...... 111-75
10.9 Recommendation 9: Support the Collection of Additional
and More Accurate Meteorological Data 111-76
10.9.1 Description 111-76
10.9.2 Benefits.... 111-76
10.9.3 Estimated Cost and Time.... ............... 111-77
III-4
-------
Page
10.10 Recommendation 10: Support the Electronic Transfer of
UNAMAP Documentation II1-77
10.10.1 Description 111-77
10.10.2 Benefits 111-77
10.10.3 Time and Cost Estimates 111-77
11.0 ALTERNATIVES NOT RECOMMENDED 111-79
11.1 Graphics Output for Models II1-79
11.2 Improving Response Time II1-79
12.0 SUMMARY 111 -80
APPENDIX A: INTERVIEW GUIDE A-l
APPENDIX B: QUESTIONNAIRE B-l
III-5
-------
LIST OF FIGURES AND TABLES FOR PART III
Page
Table 1. Questionnaire Respondents by Industry Group............ 111-10
Table 2. Years of Modeling Experience by Industry Group . III-ll
Table 3. Nature of Respondents' Work with UANMAP... ..... 111-12
Table 4. Severity of Problems by Industry Group 111-14
Table 5. Documentation Importance Vs. Ease of Use ... 111-22
Table 6. Type of Support Requested By UNAMAP Users 111-27
Table 7. Names of Organizations Used for UNAMAP Support......... II1-29
Table 8. Types of Computers Used for UNAMAP Model ing,,........... II1-42
Table 9. Version Number and Source of Models Used.. ......... 111-45
Table 10. More UNAMAP Models Listed Than Exist.................. 111-52
Table 11. Most Frequently Used UNAMAP Models.................... 111-54
Table 12. Users Prefer to Buy UNAMAP............................ II1-66
Figure 1. Types of Data Used by Air Quality Models.............. 111-34
III-6
-------
EVALUATION AND ASSESSMENT OF UNAMAP
PART III: INTERIM REPORT
1.0 INTRODUCTION
1.1 Background
The User's Network for Applied Modeling of Air Pollution (UNAMAP) models are
provided by the Environmental Protection Agency (EPA). This software library
is made up of the models whose development was funded by EPA or whose use is
recommended for environmental impact analysis by EPA's Office of Air Quality
Planning and Standards (OAQPS). The Environmental Operations Branch (EOB) is
part of the Meteorology Division within EPA's Office of Research and
Development (ORD). EOB staff have two major areas of responsibility: 1)
continuing research in dispersion modeling; 2) technology transfer of air
quality dispersion modeling to the public. Distribution and support for UNAMAP
are funded by EOB as part of their technology transfer function.
As part of its ongoing effort to improve the utility and availability of UNAMAP
to the public, EOB commissioned the evaluation study of which this report is a
part. A major portion of the study was to collect data from UNAMAP users and
other segments of the dispersion modeling community. The data collection was
to serve two objectives. One objective was to determine who the UNAMAP users
were and how models were used. The second was to find those areas where the
modeling community felt that improvement was required. The purposes of this
report are: (1) to present the data collected from the users, (2) to describe
areas where improvement of UNAMAP would be most beneficial, and (3) to make a
set of recommendations for those improvements.
1.2 Data Collection and Analysis
Data was collected in two phases. The first phase was a set of 22 in-depth
interviews with key members of the UNAMAP community as identified by Battelle
III-7
-------
and EOB. Interviews were conducted with representatives of the following
organizations:
EPA (including ORD & OAQPS) 6
State agencies 6
Regional Meteorologists 2
Consultants 2
Private industry 2
Local or county agencies 1
Software marketers 2
Educational institutions 1
TOTAL 22
The interviews yielded great insight into the complexity of UNAMAP and the
nuances of modeling. They also formed the basis for the construction of a
questionnaire to be mailed to a larger segment of the UNAMAP community.
The questionnaire was mailed to 275 organizations selected from the directory
of the Air Pollution Control Association and from the list of UNAMAP
purchasers. The questionnaire includes the following sections?
o User profile
o Models used
o Training, documentation, and support experience
o Hardware available
o Problem areas
o Suggestions for improvement
Responses were submitted by 106 organizations. A copy of the survey guide is
in Appendix A; the questionnaire is in Appendix B.
The remainder of Section 1 presents a summary of the findings including a user
profile, discussion of how the problem area data was analyzed, and an overview
of the most frequently mentioned problem areas* Sections 2 through 7 discuss
in detail each major area of improvement, and Section 8 summarizes other user
concerns. A description of overall constraints or factors limiting improvement
III-8
-------
is given in Section 9, and Battelle's recommendations are presented in Section
10. Section 11 describes why certain alternatives or suggestions were not
recommended, and Section 12 is a brief summary.
1.3 UNAMAP User Profile
Response from a cross-section of the modeling community was necessary to
evaluate needs and make recommendations applicable to all. The first section
of the questionnaire elicited a profile of the UNAMAP users. Users were asked
about their industry affiliation, their years of modeling experience, and the
nature of their work with air quality simulation models (AQSM).
Respondents to the questionnaire are members of distinct segments of the in-
dustry. Consultants and state government agencies returned the largest number
of questionnaires (35 each), followed by private industry (23), local
government users (9), and other users (4). Table 1 shows the questionnaire
respondents by industry group. Due to the structure of the sample group, it is
not possible to say that consultants and state users actually represent the
majority of the total UNAMAP users.
The experience levels varied by industry groups. Consultants had the highest
average with more than 10 years experience, and state governments had the most
persons with 5 to 10 years experience. Table 2 gives the number of respondents
per user group and their years of experience.
Respondents were also asked to categorize the nature of their involvement with
the models. Because the majority of questionnaire respondents indicated they
were users of the models, the two terms, users and respondents, will be used
interchangeably throughout this report. Other categories of involvement are
data provider, developer, remarketer, software support, and validator. See
Table 3 for more specific statistical data.
III-9
-------
Other
3.7%
State Gov't.
33%
Consultants
33%
Private Industry
21.7%
Table 1. Questionnaire Respondents by Industry Group
111-10
-------
YRS OF MODELING EXPERIENCE BY INDUSTRY
UNAMAP Questionnaire Respondents
State Gov't
Local Gov't Private Industry Consultants
Other
1-4 Yrs.
IV\l 5-10 Yrs.
10+ Yrs.
Table 2. Years of Modeling Experience by Industry Group
III-ll
-------
NATURE OF WORK WITH UHAMAP OR OTHER MODELS
CO
z
c
-1
fD
0
-*>
T3
O
a.
a>
o
*:
_Jo
<-(
3-
cr
z
State
Government
Local
Government
Private
Industry
Consultants
Other
TOTALS
Total
Respondents
35
9
23
35
4
106
Data
User Provider
34 3
9 2
21 3
35 2
2
101 10
Developer Remarketer
1
2
1
3 3
2
9 3
Software
Support Validator Other
1 2
2
4 1
4
1
12 3 0
-------
1.4 Analysis of Problems/Suggestions
The largest portion of the questionnaire dealt with the problems perceived by
the users and their suggestions for improvement. Twelve problem areas were
identified by Battelle project team members based on conversations with EPA.
The areas included in the questionnaire and interview guide are:
1) Choosing a model
2) Accuracy of the models
3) Buying/accessing the models
4) Documentation for the models
5) Support for the models
6) Data collection
7) Data input
8) Hardware compatibility
9) Hardware access
10) Unreliable hardware
11) Response time
12) Other
Questionnaire respondents were asked to rank the list of twelve possible
problem areas on a scale of 0 to 10 where 0 indicated the area is not a problem
and 10 that the area is a severe problem. User responses were grouped with 0
to 2 identified as a low ranking, 3 to 6 a medium, and 7 to 10 a high ranking.
Data from the interviews was summarized into the same categories as those of
the questionnaire. Unless otherwise indicated, all discussion of user opinions
and requirements will represent a combination of questionnaire responses and
interview transcriptions.
To qualify for further analysis, a problem area was required to be scored
"medium" or "high" on the severity scale by all of the major user categories:
state agencies, private industry, consultants, and local agencies. Table 4
gives the complete data.
111-13
-------
PROBLEM AREAS (Q.I5)
fa
CT
CO
o>
n>
-$
c-t-
O
-5
o
cr
(T)
3
to
O.
C
CD
O
C
T3
User
Group
Private Industry
Consultants
Educational Insts.
federal Government
State Government
Local Government
OVERALL
Total
Respondents
23
35
1
3
35
9
106*
Hen:
I
L
L-H
I
t
L
H
L
2
H
H
M
H
H
H
H
3
L
L
H
L
L-H
L
4
H
M
H
M
M
M
H
5
H
H-H
H
H
H
H
6
EVEN
H
H
L
H
H
M
(EVEN)
7
EVEN
L
H
L
L
EVEN
L
e
EVEN
H
L
L
EVEN
H
L
(EVEN)
9
L
L
L
L
L
L
L
10
L
L
L
L
L
L
L
11
L
L
L
L
M
L
12
Hens:
1. Choosing a nodel
2. Accuracy of node)
3. Buying/accessing the model
4. Oocunentatton for the uodel
5. Support for the model
6. Data collection
7. Data Input
6. Hardware compatibility
9. Hardware access
10. Unreliable hardware
H. Response tine
12. Other
Explanation: Respondents were asked to rani: each He« on a seal* fro* 0 to 10. where 0 want that the I tea was 'not a proble*1.
and 10 oeant that the Hem was « 'severe problea.' in the above table, 'L' corresponds to rank values 0-2,
'H' to 3 6, and 'H' to 7 - 10, Indicating the low, aedtua, or high significance of the itea as a problea for the
respondents. The 'I' under Hen i for Private industry, for example. Indicates that acre Industry respondents
gave Itec J a rank value in the range 0-2 than gave it any other rank. 'EVEN' indicates that the responses were
evenly distributed snong ihe r«nk values.
° 118 questionnaires were returned, but 12 were blank, for e total of 10§ respondents.
-------
Each qualified problem was then analyzed by the project team to determine the
following:
o Importance to the users
o User suggestions
o Alternatives
o Evaluation criteria
o Constraints
Determination of importance was based on the proportion of questionnaire
respondents who felt that there was a problem. Interview transcriptions were
examined for further verification. The number of suggestions by users was also
a factor.
The suggestions themselves were analyzed by putting them into categories.
Percentages were calculated for each category.
Alternatives were generated from interview and questionnaire suggestions and
from the experience of the Battelle staff in solving similar problems. Evalua-
tion criteria and constraints were identified from the same sources, and used
to determine the strengths and weaknesses of the alternatives.
By evaluating the suggestions and alternatives, Battelle developed a list of
recommendations based on the data and previous professional experience. The
recommendations were formulated to address multiple problem areas wherever
practical, and to utilize technology which is available to the UNAMAP support
staff and model users. Estimates are given for the time and cost necessary to
implement the recommendations (see Section 10).
1.5 Overview of Problem Areas
As indicated in Table 4, user responses show that the following problems were
of greatest concern to questionnaire respondents:
o Accuracy of models
o Documentation of models
ni-i5
-------
o Support for models
o Data collection
o Data input
o Hardware compatibility
Respondents consistently ranked accuracy of the models high on the scale, in-
dicating that it is an area of significant concern among UNAMAP users.
Documentation and support for the models are also items of concern for all
users. State government users (including one respondent from an educational
institution) see data collection as a significant problem. In addition, local
government users, as well as others, view hardware compatibility as a major
problem.
The concerns about data input and output format were expressed primarily in the
interviews. Those interviewed generally indicated data input seemed inor-
dinately complex and time-consuming, given the actual processing that occurs.
This is made even more obvious when comparing to the microcomputer versions of
models produced by Bowman Engineers and Trinity Consultants. The output of the
models was regarded by the users as too voluminous, and the format was seen as
not useful. Interviewees suggested that more flexibility in output format
would be an improvement in this area. In particular, the ability to suppress
the insignificant portions of the concentration matrix was requested.
The following sections will discuss the major problem areas.
111-16
-------
2.0 MODEL ACCURACY
While no model represents reality with absolute accuracy, the degree of ac-
curacy and its limits are important to model users. In regulatory usage,
UNAMAP model results are the primary basis for decisions such as State
Implementation Plan (SIP) revisions, prevention of significant deterioration
(PSD) permitting, and permits for proposed sources of air pollution. Users of
UNAMAP and management decision-makers must know how to evaluate the accuracy of
the model results.
The accuracy of a model is determined by comparing the concentration estimates
derived during the modeling process with actual measurements of the air
quality. There are two components to model accuracy inherent uncertainty
and reducible uncertainty. The inherent uncertainty of the models results from
variations in unknown conditions such as the exact dispersion characteristics
of the pollutants. Reducible uncertainties are caused by uncertainties in
emission characteristics, errors in meteorological data, and inadequacies in
model physics and formulation.
The most commonly stated accuracy measurement for air quality simulation models
is a factor-of-two accuracy. The actual measurements may be as low as the
model results divided by two or as high as the results multiplied by two.
Actually, however, the accuracy of model estimates vary depending on the model
used, the type of application, and site-specific characteristics. Also, many
users believe that the models tend to over-predict in order to protect the
environment. Users report that models are quite sensitive to minor deviations.
For example, the choice of one switch over another or changing one number in
wind direction can drastically change the accuracy of the results for the
specified actual meteorology and emissions.
2.1 Importance of Accuracy
A majority of users consider model accuracy a problem. Several users expressed
serious concern that decision-makers did not know how to interpret information
on model accuracy. Accuracy was given a high ranking (7 to 10 on the scale) by
48% of questionnaire respondents. Another 26% gave it a medium ranking (3 to 6
111-17
-------
on the scale). For those interviewed, 55% cited it as a problem with one
declaring decisions on the accuracy of a model in complex terrain as his
biggest problem.
2.2 User Suggestions
A total of 38 suggestions for improving the accuracy of the models were made by
UNAMAP users. Most suggestions fell into four major categories:
o Of the suggestions, 50% dealt with model validation and evaluation,
including peer review and real-world testing, as well as improving the
availability of validation/evaluation results.
o Improvement of the scientific basis of dispersion models and correcting
perceived errors in current models were suggested by 30% of the users.
o Eighteen percent of the users suggested the inclusion of a statement of
the error bounds or uncertainty with the models, and/or stating the
error in terms of confidence intervals.
o Improved data and more models which can handle site-specific data were
suggested by 13% of the respondents.
2.3 User Concerns and Alternatives
2.3,1 Concerns
Users identified three types of issues within the area of model accuracy.
First, users do not know the accuracy of a model in some specific situations
such as complex terrain. Second, there is no easily understood method of
communicating the accuracy of a model to users and decision-makers. Third,
proving and improving the accuracy of current and future models is needed to
increase industry confidence.
To some extent, model accuracy is limited by data precision. The relationship
between the factors of data and accuracy should be determined. In this
111-18
-------
section, we will address non-data related alternatives. Data problems are dis-
cussed in Sections 5 and 6.
2.3.2 Alternatives
There are several alternatives available which address one or more of the
accuracy issues. The ultimate solution is to develop and utilize a model which
is more defensible from a technical standpoint. This alternative would include
more model validation and evaluation activities, within specific situations,
for a variety of models. A broadening of the peer review process would give
more segments of the modeling community a voice in identifying and developing
solutions to help identify accuracy problems in the current models. Funding
would also be used for research to improve the scientific basis of dispersion
modeling.
A second alternative is to establish an improved dissemination process for
those evaluations and validations which have been conducted. This could be
combined with non-technical discussions of accuracy. For example, general
statements regarding model accuracy and comparison of results could be included
in the user's manual for each model.
The third alternative is to state the model results in terms of confidence
intervals. Techniques to incorporate confidence intervals into the models or
into a post-processor would be developed.
2.4 Evaluating Accuracy Alternatives
Improvements to the accuracy of UNAMAP models are not currently cheap or easy
to come by. The obvious model improvements have generally been made. The same
criteria are important for each of the three concerns situation-specific
accuracy, easily communicated statements of accuracy, and improvements to
accuracy. The first criterion is the cost of the solution. The estimated
dollars and staff time to pursue any alternative must be considered. Second is
the solution's effectiveness: To what degree will the alternative improve
accuracy or improve users understanding of accuracy? Third is the
111-19
-------
alternative's applicability. The breadth of impact must be evaluated in terms
of number of models, segments of the UNAMAP community, and number of specific
situations currently and in the anticipated future.
2.5 Constraints
A variety of constraints make knowing and improving AQSM accuracy challenging.
First, there are not many thorough model evaluations in the literature.
Therefore, the amount of information available, especially for specific
situations, is quite limited. Second, resources for new model development or
improving existing models are quite limited. This fact not only slows research
and development, it increases the gap between the best modeling technology and
regulatory model technology by limiting the work which is required for
regu1atory approva1.
111-20
-------
3.0 DOCUMENTATION
Most UNAMAP Version 6 models (all of them except two) have some type of manual
-- most frequently a user's guide. Guideline models have the bibliographic and
ordering information for the documentation listed with the description of each
model.
Because UNAMAP models have been developed over a long period of time and by
many organizations, the documentation for a model is generally written by the
developer of the model. Therefore, the type of manuals available as well as
their content, format, and quality varies from model to model. Current UNAMAP
documentation must serve three types of users: decision-makers, dispersion
meteorologists, and data processing personnel. Each of these user categories
has distinct information needs. To help improve the quality and consistency of
the user's manuals, EPA produced a manual, Handbook for Preparing User's Guides
for Air Quality Models (EPA, May 1983).
3.1 Importance of Documentation
Documentation is very important to the users. Many users stated that the
manual was their main means of learning how to use the models. Also, the
proliferation of user-friendly, microcomputer products for other applications
has led to a generally increased level of expectation in terms of documentation
clarity and ease of use. The UNAMAP system must meet these expectations. When
asked to rank the importance of the documentation to them on a scale of 0 to 10
the responses have a mean value of 8.6 with a standard deviation of 1.45.
Users were also asked to rank UNAMAP documentation in terms of ease of use.
The mean response was 5.4 on a 0 to 10 scale (where 0 was very difficult and 10
was very easy). The 3 point difference indicates that documentation useability
may not be commensurate with its importance. One interviewer said that using
the documentation was his biggest problem. Table 5 shows the responses by user
categories as well as the total number of responses.
111-21
-------
DOCUMENTATION IMPORTANCE VS. EASE OF USE:
Mean and Standard Deviation of Both Questions by Industry Group
State
Government
Local
Government
Private
Industry
Consultants
Other
TOTALS
IMPORTANCE
N Mean
34 8.4
9 8.8
21 8.6
35 8.7
2 9,0
101 8.6
OF DOCUMENTATION1
Standard Deviation
1.48
1.48
1.40
1.53
1.41
1.45
EASE OF USE
N Mean
33 5.4
9 5.4
21 5.2
35 5.5
2 6.5
100 5.4
OF DOCUMENTATION2
Standard Deviation
1.78
2.30
2.18
1.29
0.71
1.97
Ranked on a scale from 0 to 10, where 0 meant 'not at all' and 10 meant
'totally'.
o
Ranked on a scale from 0 to 10, where 0 meant 'extremely hard' and 10 meant 'very
easy'.
Table 5. Documentation Importance Vs. Ease of Use
111-22
-------
3.2 User Suggestions
Most respondents said that, at a minimum, documentation needed some
improvement. A total of 72 suggestions were made by users, primarily in the
following categories:
o The most common suggestions (over 26%) concerned rewriting, reorganiz-
ing, or otherwise making the documentation easier to read and under-
stand.
o Another 18% of the suggestions were that all models needed documenta-
tion. Complex I and II were mentioned in nearly every case. Some also
suggested documenting the preprocessors such as RAMMET.
o Many suggestions (15%) were to include more examples of model usage
especially more variety of input files, area sources, and elevated
terrain in the models that deal with those items.
o Receiving updates and corrections as models and documentation are
revised was mentioned in 14% of the suggestions.
o More timely preparation and distribution of documentation was the topic
of over 12% of the suggestions.
o Writing various parts of the documentation for different users
(evaluators, computer analysts, and end-users) was suggested over 8% of
the time.
3.3 User Concerns and Alternatives
3.3.1 Concerns
The above suggestions point to the conclusion that UNAMAP model documentation
is not oriented to those segments of the community that need it most: the
novice or infrequent user. Instead the documentation tends to be directed to
the AQSM researcher or evaluator. The novice or infrequent user has difficulty
111-23
-------
locating the information he needs due to inconsistencies in format and level of
detail across models.
Also, the lack of a working, established mechanism for the distribution of
changes or updates to the documentation compounds the problem for experienced
and novice users alike. Experienced users work from memory of previous
versions, thus making errors and not taking advantage of new capabilities.
Novice users call for more support.
3,3.2 Alternatives
To meet the needs of UNAMAP users, there is no question that the documentation
should be improved in the following areas: media, priorities, and format.
One alternative, which would address the problems of documentation and update
distribution, is to provide documentation via electronic media. The distribu-
tion and update process would be simplified, improved, and would allow for more
timely provision of information.
This solution, however, is not without problems of its own. Decisions would be
needed on the precise format of the medium, the distribution mechanism, and how
graphics would be handled. Various media formats are available such as word
processing diskette, tape, or a file transfer process. Distribution mechanism
alternatives include NTIS, EPA Regional Office, EOB, or an electronic bulletin
board. Graphics would be the biggest problem since the format is specific to
an output device. It is possible that graphics page updates could be provided
on paper to give a more uniform look to the manual.
An alternative is needed to attack the ease-of-use issue. Producing a separate
set of user manuals for the novice or infrequent user would allow information
to be presented with more "cookbook" instructions and more examples.
A different alternative is to update the existing manuals in line with user
suggestions. Creating new indexes, clarifying certain sections, and preparing
a manual for Complex I are examples of improvements under this alternative.
111-24
-------
3.4 Evaluating Documentation Alternatives
Several factors must be weighed before a documentation improvement approach is
chosen. First, the feasibility of electronic documentation depends on the
existence, cost, and efficiency of an electronic distribution network, and on
the availability of hardware for output (especially graphics). Second, any
alternative must be evaluated on the method used to incorporate updates and
changes. Third, a pilot test of any improvement must be performed by the
users.
3.5 Constraints
Preparing documentation is time-consuming and requires personnel with certain
skills. EPA currently has limited staff, and the staff that is available does
not have the specialized knowledge necessary to organize an extensive
documentation project. Another constraint is the existing distribution chan-
nels. The formal distribution mechanism of ordering documentation from NTIS is
not conducive to frequent updates. While Regional Offices provide informal
documentation distribution services, EOB does not have the personnel resources
available to assume formal responsibility.
111-25
-------
4.0 USER SUPPORT
One of EOB's mandates is to facilitate the technology transfer process in air
quality modeling. In this process, the technology provider supplies the exper-
tise necessary for the recipients to apply the technology successfully. For
UNAMAP, technology transfer means that support must be provided to model users,
and that support must be consistent and accessible.
User support consists of information and assistance provided to answer
questions and give guidance as to the proper application and successful
execution of the UNAMAP models. Generally, support can be described as either
policy guidance or technical support. Policy guidance refers to the
specification of models, options, and data necessary to satisfy regulatory
requirements. Technical support is largely a troubleshooting exercise as
errors occur in such processes as compilation, data preprocessing, or
execution. Troubleshooting is also necessary after execution if improbable
results are obtained. The distinction between policy guidance and technical
support can blur, however, when questions of accuracy or meteorological
appropriateness arise. Table 6 indicates the steps in the modeling process
where respondents seek support.
4,1 Importance of Support
Overall, support was rated as at least a medium level problem by all user
groups. Consultants rated this area as a medium to high level problem.
4.2 User Suggestions
Thirty-four suggestions for improving UNAMAP support were offered in the ques-
tionnaires and interviews. In general, more experienced and sophisticated
users reported fewer technical problems. Some of these users, however, had the
most severe policy guidance problems. Suggestions for improving support tended
to fall into the following categories'
111-26
-------
SUPPORT REQUESTS BY UNAMAP QUESTIONNAIRE RESPONDENTS:
Type of Support Requested by Industry Group
ro
cr
i
n>
T3
n>
O
n>
.a
a>
fD
CL
00
cz
^
-o
a>
Number Choosing
of a Model Supplies Software Interpreting
Entries* to Use Data Problems Results
State
Government 34 16 7 12 18
Local
Government 14 10 3 7 7
Private
Industry 17 9 5 8 7
Consultants 47 15 14 26 10
Other 3003 2
TOTALS 115 50 29 56 44
Programming
Questions
9
8
6
20
2
45
* Greater than number of respondents in an industry group because multiple support organizations
used.
-------
o The most common suggestion (44%) was to identify and publicize a dedi-
cated UNAMAP support group to deal with all issues.
o Nearly 18% of the suggestions were to establish either a newsletter or
a computer bulletin board to distribute modeling information, dec-
isions, updates, and corrections.
o Almost 12% thought support could be improved by converting all models
to a specified, common version of Fortran.
o Nine percent suggested that uniformity and standards within the models
would solve their support problems.
o Another 9% felt that support could be improved through a clarification
of ORD vs. OAQPS roles regarding UNAMAP.
4.3 User Concerns and Alternatives
4.3.1 Concerns
Although the quality of support was generally given a high rating, the real
problem was determining the source of the required support* This was reflected
in the number of organizations listed by users as providing support to them.
For example, within EPA, the Meteorology and Assessment Division helps users
with the modeling software, while the Office of Air Quality Policy and
Standards (OAQPS) answers questions on regulatory issues. A composite list by
user group, shown in Table 7, lists the names of support organizations verbatim
from the questionnaires.
In the majority of the suggestions above, users are indicating that UNAMAP
support is too fragmented and decentralized. This fragmentation, however, is
explainable given the history of UNAMAP- Each of the 24 models is a complex
piece of software incorporating sophisticated meteorological mathematics. The
models were developed by different sources over a period of nearly 15 years.
111-28
-------
LIST OF ORGANIZATIONS USED FOR UNAMAP SUPPORT
BY QUESTIONNAIRE RESPONDENTS
NAME OF ORGANIZATION PROVIDING SUPPORT
# RESPONDENTS
State
Government
EPA Regional Office
EPA
EPA/RTP
OAQPS/RTP
U.S. EPA - Atmospheric Science
Research Lab
Miss. Dept. of Natural Resources
Radian Corp.
Trinity Consultants
22
3
3
2
1
1
1
1
Local
Government
EPA Regional Office
U.S. EPA
EPA/RTP
EPA Ohio
California Air Resources Board
Lawrence Livermore Laboratories
PEI
Philadelphia AMS
Trinity Consultants
5
2
1
1
1
1
1
1
1
Private
Industry
Texas Air Control Board
U.S. EPA
EPA Regional
U.S. EPA/OAQPS
EPA - Meteorology & Assessment Div,
NCC
B. Turner
NTIS
Coast Guard
NOAA/NCDC
Oregon State Air Quality Agency
Trinity Consultants
3
3
2
1
1
1
1
1
1
1
1
1
Table 7. Names of Organizations Used for UNAMAP Support
111-29
-------
NAME OF ORGANIZATION PROVIDING SUPPORT # RESPONDENTS
Consultants EPA 10
EPA Regional Office 7
EPA/RTP 6
U.S. EPA - Environmental Opers. Branch 3
EPA - Meteorological Lab 1
EPA - Model Applications Section 1
EPA - Model Clearinghouse 1
OAQPS 1
National Climatic Center 4
Cramer (H.E. Cramer Co.) 2
Texas Air Control Board 2
Bowman Environmental Engineering 1
California Dept. of Transportation 1
Colorado State Agency Modelers 1
Indiana Dept. of Environmental Mgto 1
MMS 1
Radian 1
South Coast AQMD 1
Trinity Consultants 1
Table 79 continued. Names of Organizations Used for UNAMAP Support
111-30
-------
As a result, the developer is frequently the one called upon to answer
technical questions. Regulatory suitability issues have been consistently
referred to persons (in OAQPS) who are familiar with the regulations.
Another facet of the problem is that technical support is not the primary
activity for EOB personnel. EOB research meteorologists are evaluated on their
research efforts in meteorology rather than on their ability to help a novice
state agency user through the set-up for a routine ISCLT run.
The remainder of the suggestions reiterate the user's perception of UNAMAP as
disjointed and non-standard. Just as there is no single point of contact for
support, users see little uniformity in the model components and no attempt to
distribute a universally executable model. These two impressions increase the
amount of technical support that users want available to them.
4.3.2 Alternatives
There are two approaches to consolidating support activities. One is a decen-
tralized approach with all support activities integrated into the Regional
Offices. Support would be closer to the users, and more tailored to local
circumstances. In this alternative, Regional Offices have access to EOB and
the model clearinghouse in OAQPS when they cannot resolve the question.
One disadvantage to this alternative is that users may see it as further
fragmentation of UNAMAP activities. This would particularly affect consultants
and industry users who deal with multiple regions.
The second alternative is for centralized support based on the establishment of
a Modeling Support Branch within EPA. This staff would have expertise on all
the UNAMAP models, a variety of hardware, and would be authorized to convey
regulatory requirements when necessary. The support group could communicate
with users through a "hot-line" facility or via an electronic bulletin board.
A third alternative is to provide standardized user interfaces, algorithms, and
other model components. This approach would help correct the user's perception
111-31
-------
of UNAMAP as fragmented. A "help" facility included in the code could also
reduce support requirements.
4.4 Evaluating Support Alternatives
The critical factor for support activities is the level of service given to
users. Goals and standards for the necessary level of service must be deter-
mined. Standards could include time for initial response to user, time to
solve the problem, frequency of referring user elsewhere, and other related
measures. Alternatives must be judged against the level of service provided.
Good user support obviously is worth some price. However, the cost effective-
ness must be considered. For each alternative, the value of the difference in
service must be compared with the difference in cost.
Support alternatives should also be evaluated in the context of other planned
improvements. For example, documentation upgrades would reduce the amount of
support required, as would menu and user interfaces. Adding new models to the
Guidelines, however, would increase calls for support.
4.5 Constraints
The primary constraint is staff resources. Current staffing levels are inade-
quate to assign dedicated support personnel. Also, the experts required for
technical and policy support are not readily available. A support staff train-
ing program and/or staff rotation would be necessary.
111-32
-------
5.0 DATA COLLECTION
AQSMs use meteorological data usually consisting of hourly meteorological
readings to characterize the transport and dispersion conditions at the site.
Models also require data about the emission source and about the terrain sur-
rounding the source. Figure 1 shows the types of data required by AQSMs.
Because of the variety of data and the importance to the modeling effort, data
collection is a significant part of air quality modeling. Source emission
rates, heights, and locations must be derived from engineering calculations for
the facility. Background emission inventory data and terrain data are also
required in many situations. The data collection effort includes the
activities necessary to determine the appropriate data and generate or acquire
the data.
There are two sources of meteorological data: data from a National Weather
Service (NWS) installation near the site or data collected at the site itself.
This section addresses collection problems of both NWS and onsite data.
5.1 Importance of Data Collection
State agency users of UNAMAP rated data collection as a severe problem; consul-
tants and local government users ranked it as a medium-level problem. Industry
user rankings were evenly spread through high, medium, and low. One respondent
said that data collection and input together took 80-90% of his modeling time
(he felt that it should be 5%). Many anecdotes about the problems of data
collection were related in interviews.
111-33
-------
C
(TB
o
-h
O
P>
P>
CO
CL
cr
C
pi
Emissions
Inventory
Onsite
Meteorological
Data
^
Air Quality
NWS
Meteorological
Data
Terrain
Data
-------
5.2 User Suggestions
Users offered a wide range of suggestions for solving their data collection
woes. The total of 56 suggestions covered all data collection activities for
UNAMAP models. Neither cost nor political difficulty inhibited the
suggestions. The major suggestion categories are described below:
o Improvements in the availability of data were suggested in over 21% of
the responses. Data availability includes speed of access to the data
as well as knowledge of existing data. Most of the suggestions for
improved data availability were concerned with NWS and preprocessed
data held by EPA.
o Another 21% wanted meterological data collected for more sites or more
representative sites. Typically, the reference was to "on-site" data.
o Nearly 20% suggested improving the quality control for existing NWS
data. The usual problem mentioned was missing data.
o Nine percent of the suggestions were to collect more, or better, NWS
data such as hourly averages or upper air data.
o An additional 9% said that the data should be less expensive.
o Improving source/emission data was in 5% of the suggestions.
5.3 User Concerns and Alternatives
5.3.1 Concerns
The main complaint of users is that there is no comprehensive source of
information on what data are available, and how to get it. This complaint
encompasses NWS data, preprocessed data at EPA, and on-site data. Acquiring
NWS data from National Climatic Data Center (NCDC) is seen by the user
community as a lengthy, cumbersome process. A great deal of time and effort is
111-35
-------
required to collect on-site data. Since these data most accurately represent
the site, use of them should be facilitated in any way possible.
5.3.2 Alternatives
There are a variety of data categorizing and distribution services which a data
clearinghouse could offer to the modeling community.
A quick and inexpensive alternative is to use a newsletter format to provide,
for example, a list of preprocessed data available through OAQPS and a descrip-
tion of NCDC ordering procedures. Users could contribute their knowledge and
references of data to subsequent issues. In addition to NWS meteorological
data, a clearinghouse could provide references to other publicly funded data
such as the National Emissions Data Bank, state monitoring station data, and
USGS terrain data. References to on-site meteorological data collected at
private expense could be maintained to facilitate the re-use of on-site data
whenever possible.
A more elaborate option is to establish the data clearinghouse as part of an
electronic bulletin board system. As in the paper alternative, existing data
would be listed, including items submitted by users*, An electronic system,
however, would allow any data available to the bulletin board host computer to
be downloaded by users.
The services to be provided and the medium of communication depend on the
amount of initial effort required. An inventory of preprocessed NWS data at
OAQPS should be easily obtained, but a list of privately funded, on-site data
is much more difficult. The utility of monitoring station data may be too low
to justify including it. '
There is one issue regarding the inclusion of private data. EPA must decide
whether it is appropriate for a government agency to publicize the availability
of private data when the owner of the data may charge for its use. The cost of
ensuring equal treatment to all owners and potential users must be included in
the evaluation.
111-36
-------
5.4 Constraints
The amount of data available to the clearinghouse may constrain its services or
medium. Too little data will make the entire effort unproductive; too much
data may overwhelm a bulletin board system or its host computer.
111-37
-------
6.0 DATA INPUT
After data collection is complete, the data are not yet ready for use. The
data input process includes activities to resolve various meteorological or
source data problems, putting the meteorological data into the proper format,
and entering receptor and possibly other terrain data. Data input is often a
manual function which requires a great deal of user time and effort, but little
modeling expertise. Preprocessors (programs used to distill the data) can
correct several types of data problems. For example, all parameters needed by
the models are not directly measured. Preprocessors are used to produce these
parameters as well as take care of other data problems such as calms or missing
data.
Simulation models, like most other software, expect data to be presented in a
specific fashion. Since models were developed independently, data input
methods vary from model to model. There are two general formats for AQSM's:
fixed and free format. Those models with a fixed format require that the data
be input in a strictly defined way which is specific to that model and
extremely inflexible. The input requirements for free format, while less
restrictive, still are not consistent among models. To solve the inconsistency
problem, microcomputer models have interactive, user-friendly programs which
lead users through the data entry process.
6.1 Importance of Data Input
Data input is a problem for novice or infrequent users. They must expend a
great deal of effort in determining how to enter data for each specific model.
Of the questionnaire respondents, 46% ranked data input as a medium or high
level problem, and 60% of those interviewed mentioned some problem with this
area. As is shown in Table 4 (see p. II1-14), government users ranked data
input as more of a problem than did private industry or consultants.
111-38
-------
6.2 User Suggestions
There were 49 suggestions made to improve or ease the input of meteorological
data into UNAMAP models. User suggestions covered the following categories:
o Forty percent of the suggestions were for a user-friendly front-end
program to make data input easier. One user also suggested a data
checker to catch errors before running the model.
o Twenty-five percent of the suggestions requested better handling of
certain data problems (e.g., missing data and overstrikes in NWS data).
Better handling of on-site and complex terrain data were also
requested. The increased amount of elevation and coordinate data in
complex terrain situations is cumbersome in many UNAMAP models.
o Seventeen percent of the suggestions were for a consistent data input
format from model to model and another 6% wanted free format input into
all models.
o Eight percent concerned better documentation to aid users when
inputting data.
6.3 User Concerns and Alternatives
6.3.1 Concerns
From the suggestions, we conclude that UNAMAP users think data input can and
should be made easier. The majority of suggestions were for ease of use im-
provement; only 25% were concerned with better handling of specific types of
data.
6.3.2 Alternatives
As discussed in the Technology Assessment, the UNAMAP user is increasingly
likely to be a novice or inexperienced user. Therefore, data input must be
111-39
-------
made easier. There are three alternative approaches: change the data, change
the models, or provide an interface between data and models.
Changing the data would require providing data in the multiplicity of formats
required by the various models. The second approach would change varying data
input requirements for all models into one, standardized format. The third
alternative would require a program which first prompts users for input data,
then structures that data into the format required by the model being used.
6.4 Evaluating Data Input Alternatives
The major criterion is benefit to the users. Using AQSM's is a process
requiring many steps, while experienced modelers have less difficulty with
data input, the trend in modeling is toward more novice and infrequent users.
These users are more likely to become frustrated when too much effort is
required in an activity preceived to be relatively simplistic.
6.5 Constraints
The main constraint for data input alternatives is the level of effort required
to implement the various approaches. In the first alternative, processing and
storing data for different models would increase processing time for data
providers as well as adding to data storage and distribution costs. The second
alternative for data-compatible versions of each model would require extensive
programming and testing. The third alternative, a user-interface program,
would also require a great deal of programming time including a distinct
version for each type of hardware supported.
For an interface program, hardware compatibility must be evaluated. Just as
each type of hardware must have its own version of the model, any screen-
oriented user-interface programs will require distinct hardware versions.
111-40
-------
7.0 HARDWARE COMPATIBILITY
Of the twelve possible problem areas on the questionnaire, three concerned
hardware: hardware access, hardware reliability, and hardware compatibility.
The majority of respondents ranked hardware access and reliability as minor
problems. Hardware access was ranked as a low level problem by 62% of the
respondents, and reliability was given a low ranking by 71%. Hardware
compatibility, however, caused difficulty for many users.
There are many types of hardware used for modeling by the UNAMAP community. Of
the 106 questionnaire respondents using mainframe or minicomputers, 43% use IBM
computers, 26% use VAX, 19% use UN IVAC, and 44% use one of several other
systems. IBM-compatible microcomputers are used by 44% of the respondents, and
only 7% have microcomputers which are non-IBM-compatible. See Table 8 for
further data on hardware use and availability.
Several hardware versions of UNAMAP are available. Currently. UNAMAP models
may be used via timesharing on the National Computer Center (NCC) UNIVAC com-
puter. The UNIVAC version is the standard for the UNAMAP models, and it may be
purchased through the National Technical Information System (NTIS) on tape. A
variety of UNAMAP models have been converted from the UNIVAC to other hardware.
For example, some regions and state agencies convert models for their own use
and make these available to others within their area. Also, several private
companies have versions of UNAMAP models for sale. For example, HMM provides
an IBM version of UNAMAP models. Companies such as Bowman Engineering and
Trinity Consultants have developed microcomputer versions.
The standard version of UNAMAP will change. NCC is converting all UNIVAC
operations to IBM, and future models will be supported on hardware other than
the UNIVAC. However, whatever the standard version, users of other hardware
will be required to convert the models.
111-41
-------
i
-&,
rss
cr
n>
oo
T3
fD
o
o
o
c
rt-
fD
-J
l/J
I/)
fD
CL
O
MAIMFRAMS/MIMlCOHPtmRS USED FOR MODELING BY UNAMAP QUESTIONNAIRE RESPONDENTS:
Availability and Utilization
IBM
VAX
UNIVAC
Other*
State
Government
Local
Government
Private
Industry
Consultants
Other
On-site Off-site Used
5 12 16
1 1
9 6 11
5 14 17
2 1
On-site Off-site Used
525
8 7
7 8 13
2 2
On-site Off-site Used
1 10 11
1
3 25
1 3 4
On-site Off-site Usec
12 4 12
7 4 6
12 1 8
8 12 16
3 3
o
CL
fD
'Other brands mentioned Include:
AMDAHL
APOLLO
BURROUGHS
CDC CYBER
CRAY
DEC-10,-20. PDP-11
DG
FOUR PHASE
HARRIS PRIME
HEWLETT-PACKARD SUN
HONEYWELL UNISYS
PERKIN ELMER HANG
-------
p>
cr
CO
o
o
MICROCOMPUTERS. PLOTTERS. AND TELECOMMUNICATIONS EQUIPMENT
USED FOR MODELING BY UNAHAP QUESTIONNAIRE RESPONDENTS:
Availability and Utilization
CO
tt>
Q.
T3
CO
O
O
O
tt>
-1
I/I
(D
Q.
PC
(IBM-compatible)
PC
(Not IBM-compatible)
Plotter
(Any brand)
Telecomm. Equipment
' (Any brand)
State
Government
.ocal
Government
Vlvate
Industry
Consultants
Other
On-slte Off-site Used
27 24
6 5
14 12
35 1 5
3 1
On-site Off-site Used
S 4
3 1
3 2
3 1
On-slte Off-site Used
12 4 11
4 1 2
14 11
18 4 IB
6 2
On-slte Off-site Used
21 14
4 1 2
11 1 7
22 IB
3
-------
7.1 Importance of Hardware Compatibility
Of the questionnaire respondents, 52% ranked hardware compatibility as either a
medium or high level problem. Although UNAMAP models are available for certain
hardware other than UNIVAC, many users go through the conversion process. This
is time-consuming, labor intensive, and prone to error. It is, also, the main
reason many users cited for not changing to newer versions of UNAMAP. Table 9
shows data on version number and source of UNAMAP models as reported by users.
7.2 User Suggestions
There were a total of 47 user suggestions on improving hardware compatibility.
Of these suggestions, 30% wanted an IBM mainframe version, 30% suggested a
microcomputer version, and 21% requested other hardware versionSo One user
suggested that EPA finance the conversion of models by states. Another user
suggested that hardware to run models be available in Regional Offices which
would cut down on the variety of computer versions needed,,
7.3 User Concerns and Alternatives
7.3.1 Concerns
When UNAMAP models were first developed, many people used them on the computer
at NCC. Now, however, due to increased computer availability to users and long
turnaround time at NCC, modelers want the software available on their local
computers. Although the switch to IBM will be compatible with many users,
other hardware versions of certain models will still be needed.
7.3.2 Alternatives
Alternative approaches to increase hardware compatibility fall into two cate-
gories: support multiple hardware or make the conversion process easier.
Supporting multiple hardware would require that EOB make versions of UNAMAP
available on the most commonly used machines.
111-44
-------
VERSION NUMBER AND SOURCE OF MODELS BY INDUSTRY GROUP:
For Most Frequently Used UNAMAP Models
-u
en
m
n>
o
c
cr
n>
Q.
CO
o
o
n>
o
CL
n>
c:
1/1
n>
Q.
MODEL
ISC
INDUSTRY GROUP
State Government
Local Government
Private Industry
Consultants
Other
VERSION
Version Number Frequency*
6 11
5 8
4 6
5 1
5 5
6 13
5 14
4 6
3 1
2 1
1 1
HOW OBTAINED
Source Frequency*
EPA 6
EPA Regional 5
NCC 3
NTIS 2
HMM 2
Trinity 2
Other 2
EPA 1
NTIS 4
EPA 3
HMM 1
NTIS 15
EPA 5
NCC 1
Bowman 1
EPA 1
FREQUENCY OF USE
Avg. f times Range
used per year (1 times per year]
173 2-1000
1
31 1-140
283 1-5200
4
* Number of respondents
-------
cr
n>
ID
O
o
rt-
I
-Si
01
in
cr
fD
-J
Q.
O
-J
O
n>
o
o.
VERSION NUMBER AND SOURCE OF MODELS BY INDUS1RY GROUP:
For Most Frequently Used UNAMAP Models
MODEL
PTPLU/PTMAX
VALLEY
INDUSTRY GROUP
State Government
Local Government
Private Industry
Consultants
State Government
Consultants
VERSION
Version Number Frequency*
6 I
S 1
4 2
5 1
2 I
5 1
4 I
5 1
6 3
§ 3
4 3
3 2
5 i
HOW OBTAINED
Source Frequency*
EPA 2
EPA Regional 1
NTIS 1
NTIS 1
Trinity 1
EPA 1
Other 1
EPA 2
EPA Regional 1
Trinity 1
FREQUENCY OF USE
Avg. 1 times Range
used per year (H times per year)
110 24-260
18
9 5-12
52
30 3-104
1
* Number of respondents
-------
I
-U
pi
cr
vo
o
o
3
3
n>
Q_
-J
o'
cr
ro
-5
Q-
O
c
n
n>
o
Q.
n>
VERSION NUMBER AND SOURCE OF MODELS BY INDUSTRY GROUP:
For Most Frequently Used UNAMAP Models
MODEL
COMPLEX I&II
CRSTER
MPTER
INDUSTRY GROUP
Private Industry
Consultants
Local Government
Private Industry
Consultants
State Government
Private Industry
Consultants
VERSION
Version Number Frequency*
5 1
5 1
5 1
6 1
5 2
5 1
5 1
HOW OBTAINED
Source Frequency*
EPA 1
NTIS 1
NTIS 2
EPA 1
NTIS 1
EPA 1
NTIS 1
FREQUENCY OF USE
Avg. 1 times Range
used per year (i times per year)
260
24
10
17 1-48
1
52
24
104
* Number of respondents
in
n>
CL
-------
There are two ways in which the conversion process can be made easier. First,
conversion instructions could be written for all types of hardware. Another
alternative would remove all machine-specific code from the models, thus
reducing the conversion process to a very simple, mechanical exercise.
7.4 Evaluating Hardware Compatibility Alternatives
Choice among hardware compatibility alternatives must be based on cost
effectiveness and usefulness to the modeling community. Support of multiple
types of hardware would benefit the users, but would be extremely costly in
terms of resources.
Providing conversion instructions for various computers would take roughly the
same amount of effort as that necessary to produce different versions. The
instructions would ease the conversion process, but time and effort would still
be required by each site. Taking machine-specific code out of the models may
not solve all conversion problems encountered by users, and would require a
major programming effort.
7.5 Constraints
Resources are major constraints on hardware compatibility alternatives. The
ideal hardware compatibility solution would provide a version of the UNAMAP
models for every type of hardware used by the modeling community. This ideal
cannot be met with current staffing levels. Only a very few versions can be
written and maintained.
111-48
-------
8.0 OTHER AREAS OF CONCERN
Other aspects of the modeling process, while not a problem for the majority of
users, were of concern to certain respondents. Three problem areas were
mentioned most frequently: response time, output format, and addition of
specific models to UNAMAP. Interestingly, output format and addition of
specific models were not listed on the questionnaire in advance. These were
identified by users when discussing problems other than those listed. All
three problem areas are discussed briefly in this section.
8.1 Response Time
Response time is the time needed for the modeling software to produce results
after model setup and data input have been accomplished. The response time
reported by users varied depending on the size of the computer, the number of
simultaneous jobs, and the specific model used.
Although many questionnaire respondents (46%) said response time was only a
minor problem, 29% ranked it as a medium to severe problem. This range of
responses is due to the variety of computer configurations used. Some users
reported that the size of some models caused overnight execution for models run
on NCC and other shared mainframe computers. Running some models on a
microcomputer also takes several hours.
Solutions to improving response time are faster (i.e., bigger) computers or
faster software. Although running models on microcomputers does not decrease
execution time, the user has control over when the job is processed. While
running the model overnight on a microcomputer will give optimum use of
hardware and possibly the same response time as a mainframe job the user
may choose to start the job immediately. Even though it will tie up his
microcomputer for several hours, the output will be available that day for
study. As a result, another model run may then be performed immediately.
The software can be modified so that it will execute faster to improve response
time. Some models would run faster if the code were more efficient. This
would require re-writing the models using new programming techniques and look-
111-49
-------
up tables to decrease the processing time. Another approach would be to make
certain calculations optional so that only the processing needed for that
specific application is performed.
Evaluation of response time solutions must be based on the degree of im-
provement achieved, cost effectiveness of the approach, and general
applicability of the improvement. Modification of software is the most costly
and time-consuming approach. If models are re-written for other purposes, they
could be made more efficient at the same time. The improvement of response
time alone, however, may not be cost effective. Response time for
microcomputer versions of models can be improved by increasing the power of the
computer (e.g., 80386 based). As microcomputer costs decline, this alternative
promises to give a large benefit for the cost involved.
8.2 Output
Users report that printed results from UNAMAP are confusing, voluminous, and
not consistent from model to model. Users suggested that the output can be
improved by changing the format of output, giving summaries of data, and by
providing more graphics. Specific suggestions on format changes included:
o Make output consistent from model to model
o Print in columns (not blocks) so that multiple runs can be compared
easily
o Print up to 50 sources per page
o Print numbers in scientific notation
o Show peak concentration for each meteorological scenario
o Provide running averages rather than block averages
Data summaries would cut down on the amount of output, and allow users to spot
problem areas quickly. Output time and costs would be reduced if users were
111-50
-------
then permitted to request more detailed data on individual items or specific
types of printouts such as the data required for regulatory use. Graphics
would also reduce the amount of output required and increase its usefulness,
particularly to decision makers.
Reformatting the output to be consistent across all models would require pro-
gramming time. Using a post-processor to re-format output from each model
would require less programming time than re-writing each model. Executing the
post-processor, however, would increase response time.
8.3 Specific Models
Modelers want changes to current models, and they want models for new purposes.
To gather a picture of current model usage, both the questionnaire and
interview guide have questions regarding the UNAMAP and non-UNAMAP models used.
Table 10 shows that more UNAMAP models were listed by questionnaire respondents
than exist in the current version. Some of these models are in the Guidelines
but are not part of UNAMAP such as the Texas models and Urban Airshed Model,
while others were possibly part of past versions of UNAMAP. Some models listed
by respondents as non-UNAMAP are actually UNAMAP, but are listed in the Table
exactly as users perceived them. Table 11 lists most frequently used UNAMAP
models.
Users suggested that current models be modified to handle the following
situations:
o plume meandering
o multiple small sources
o sources closer than 100 meters
o plume rise of almost zero
111-51
-------
MODELS MENTIONED IN RESPONSE TO QUESTION ON UNAMAP MODEL USAGE
BLP
CALINE 3
CALMPRO
COM
COMPLEX I & II
CRSTER
EKMA
HEM
HIWAY2
INPUFF
ISC (ISCLT & ISCST)
MESOPUFF
MOBILES
MPTER
PAL
PLUVUE
PTDIS
PTMAX
PTMTP
PTPLU
RAM
SHORTZ/LONGZ
TCM-2
TEM-8
TERRAIN
TEXAS CLIMATOLOGICAL MODEL
TEXAS EPISODIC MODEL
UAM
UTMCON
VALLEY
Table 10. More UNAMAP Models Listed Than Exist
111-52
-------
MODELS MENTIONED IN RESPONSE TO QUESTION
ON NON-UNAMAP MODEL USAGE
AIRDOS
ADISP
AIRSHED
ARAC
AVACTA II
AQDM
BUGGR
CALINE
CALINE 3Q
CALINE 4
CALQ3
CANNY
CAVITY
CDMW
CHARM
COBUGGR
COMPLEX- IM
COOLMOD
DEGADIS
DM03 & OM04
EKMA 10ZIPP
ESEERCO
GAMMA DOSE
GAUSS 80
GEMS
HARM
HASTE
HI WAY
HUMAN EXPOSURE MODEL
HWYEM ICOVOL9
Hybrid Plume Dispersion
IMM
IMPACT
LIRAQ
MARYLAND PPSP
MESOPUFF A
MOBILE 3
MPTER-Urban
MSPUFF
MULTIMAX
NYSDEC
OCD
OZIPM-Z
PATHRAE
PAVAN
PHLCO
PLM STAR
PTFUM
PTPLU
PUFF SA
Reactive Plume Model
RPM-II
RTDM
SHORTZ
SPILLS
STAR
STRAJ
TCM
TEM
TEM 8
TEXAS EPISODIC
TEXIN
TRACE
UAM
URBAN AIRSHED/PARIS
VIS LEL1
WCAPCD
XOQDOQ
Table 10, Continued. More UNAMAP Models Listed Than Exist
111-53
-------
MOST FREQUENTLY USED UNAMAP MODELS:
Models Used by Ten or More Respondents
# of Respondents
Model Who Use This Model
ISC (ISCLT & ISCST) 81
PTPLU & PTMAX 55
CRSTER 35
VALLEY 33
COMPLEX I & II 31
MPTER 22
Table 11. Most Frequently Used UNAMAP Models
111-54
-------
Users also requested new models to fill the following needs:
o More screening models (less complicated, easier to use, and faster to
run)
o Refined complex terrain model which addresses lee side concentrations
o Generic lake/sea breeze model
o More ozone models
o Heavier-than-air models
o Toxic models
o Non-steady state models
The Environmental Operations Branch (BOB) has limited responsibility for the
heavier-than-air and toxic models. However, users see UNAMAP as the vehicle
for all their modeling needs and expect EOB to be as responsive to these needs
as they have been to past needs. For non-steady state applications, a UNAMAP
model, INPUFF, is already available. It is not, however, a Guideline model.
111-55
-------
9.0 GENERAL CONSTRAINTS
Any alternative for improving UNAMAP is subject to some external constraints.
These constraints make up the technological, economic, and regulatory environ-
ment in which UNAMAP presently exists and the environment in which it must
exist in the future.
9.1 Technological Constraints: Computer Resources
Three computer environments will be available for the future development and
support of UNAMAP. Traditionally, computer resources were provided by EPA's
National Computing Center (NCC) on a UNIVAC 1100 series mainframe. However,
the NCC is migrating all applications from the UNIVAC to an IBM 3090 system.
The conversion is scheduled for completion by October, 1988.
The Meteorology Division recently acquired a DEC MicroVax II multi-user
microcomputer, terminals, and local area networking (LAN) system. Many UNAMAP
models have been converted to the MicroVax, and most internal model development
is currently done on that system.
As shown in Table 8 (see p. 111-42), microcomputers, specifically IBM PC/XTs
and compatibles, are becoming widely available to UNAMAP developers and users.
One UNAMAP model (INPUFF) was originally distributed in a microcomputer
version. This model is now available on the UNAMAP version 6 tape.
These three computer environments (IBM Mainframe, DEC VAX, and IBM PC)
represent the most widely used systems in the world today. They offer a
variety of capacities and price ranges, and are compatible with the majority of
user mainframes (IBM), EPA regional systems (VAX/MicroVax), and office
microcomputer systems (IBM PC). Any alternative which involves new computer
hardware or software for UNAMAP must consider and utilize some combination of
these.
111-56
-------
9.2 Economic Constraints: Budget and Staff
Considering its importance and regulatory mandate, UNAMAP operates with very
limited resources. There is currently no EPA staff member dedicated full-time
to development, enhancement, and/or support of UNAMAP. NOAA research
meteorologists assume UNAMAP duties as a collateral assignment. Total funding
was approximately $550K in FY 1986. This amount included $425K in salary, and
$125K in onsite contractor funding. Although funding increases in future years
are possible, immediate improvements to UNAMAP (FY 1988) must be implemented
within current budget figures regardless of potential benefits.
9.3 Regulatory Constraints
The authority of EOB is limited to the study of airborne, non-nuclear, criteria
pollutants. In addition, models which are to be included in the Guidelines are
specified according to a regulatory policy which is somewhat independent of
computer or modeling technology or efficiency. To be seriously considered for
implementation, any alternative for UNAMAP improvement which involves adding or
deleting models must fall under the purview of OAQPS and be consistent with
policy as embodied in the Guidelines.
9.4 Future Outlook for Air Quality Modeling and UNAMAP
The success of air quality modeling as an important tool in protecting the
environment is now taken for granted by researchers, industry, and regulators.
Future development lies in three directions:
1) Solving increasingly complex or large-scale air quality problems, such
as regional and complex terrain models;
2) Developing AQSM's to deal with more specialized situations, thereby
broadening applicability;
3) Supporting a larger community of regulatory users.
111-57
-------
The models used to solve large, complex problems will themselves be large and
complex. Large regional models will be used to study dispersion and transport
of pollutants, such as acid precipitation ingredients, over large distances.
The models must handle multiple wind speeds, wind directions, and atmospheric
stability situations. These functions entail a significant increase in model
complexity. There will be a smaller number of situations needing such a model.
The Urban Airshed Model is a complex urban photochemical model, whose large
size kept it from being included on the UNAMAP tape.
AQSM's were originally designed for general use in a variety of situations.
As users and developers have become more sophisticated, models have been
tailored to meet specific needs. Sites where the terrain is not flat for
example, is one area of major concern. Effective air quality modeling in
situations require accurate complex terrain models. Some current models can be
used in complex terrain, but they are not as sophisticated as the problems they
must model. The development of CTDM -- currently underway -- is a reaction to
these needs.
New application areas for future models are toxic gases, radiological
pollutants, emergency release situations, and offshore sites. For some
applications, the addition of new capabilities to current models, rather than a
new model, will fill the need. New analytical functions, such as statistics
and health-risk assessment factors, will be added to some models. While EOB is
responsible for current models, they have no research responsibility for many
of these new areas.
Regulatory usage will remain high, and novice or infrequent users will continue
to account for a large portion of regulatory use. Because of their lack of
experience, user support will be required at a higher level than is currently
available. These users will tend to run models on microcomputers or on local
or regional hardware with which they are familiar, not on NCC's computer,,
To aid these users, data input formats should be standardized to ease the data-
entry bottleneck which currently exists. The format of model output should
also become more standardized, and models should produce reports which are
easily interpreted and quickly generated.
111-58
-------
One point which applies to all three directions (model complexity, model spe-
cialization, and novice regulatory users) is that the quality of modeling
results is inextricably tied to the quality of data available. The lack of
automated data gathering instrumentation inhibits the collection of more data
per site and at more sites. The manual transfer of data has increased the time
spent waiting for data and minimizes the sharing of on-site and NCDC data.
The concept of UNAMAP should change in the near future. To match the trend
toward a split between novice users and more sophisticated developer-users,
UNAMAP Version 7 (or 8) may do best to exist in two parts. One part would
consist of the Appendix A models and documentation directed at the novice who
will be using UNAMAP for regulatory purposes. The other UNAMAP would be the
remaining, less frequently used, models. This version would include more
technical documentation and theoretical background information.
If such a division is made, a corresponding set of changes may be necessary in
the distribution mechanism and computer system compatibility of the novice user
portion.
111-59
-------
10.0 RECOMMENDATIONS
In this section we describe recommendations for the improvement of UNAMAP. The
recommendations are not individually intended to address the specific problems
described in Sections 2 through 8, but instead will facilitate the growth of
UNAMAP in the future and eliminate current problems. Implementation of these
recommendations will move toward a new concept of UNAMAP.
The system began as a centralized, timeshared system used by a relatively small
group of experienced modelers whose needs were, for the most part, being met.
The trend now is toward a more distributed system with modelers using a mix of
hardware to meet different modeling requirements. UNAMAP, therefore, must be
upgraded and streamlined to meet the needs of a larger, less-experienced user
group who are using models to meet increasingly complex problems. The new
UNAMAP will also serve the larger user community by providing improved access
to model changes, documentation updates and meteorological data. The new
"Strategic Vision" for UNAMAP is an acceptance of its evolution.
10.1 Recommendation 1: Establish an Electronic Bulletin Board
10.1.1 Description
A multi-user bulletin board system should be implemented on the MicroVax
system. The initial uses of the bulletin board would be to provide UNAMAP
support (answers to questions via electronic mail) and to distribute informa-
tion of value to the UNAMAP community at large. Minor software patches and
documentation updates could also be provided over the bulletin board. The
services provided could expand as computer resources and communication lines
allow. The system could be expanded to allow distribution of new versions or
releases of the models, new manuals, and even data.
10.1.2 Benefits
There are significant benefits in an electronic bulletin board system:
111=60
-------
o UNAMAP support is removed from reliance on the telephone. The result
will be more complete and faster information for users and a more
efficient use of support staff time, which will benefit both the user
and the support personnel.
o Users can enter their questions into the system without regard to the
availability of support staff. Support providers will be freed from
the instantaneous demand of the telephone.
o A bulletin board allows for direct communication among all segments of
the UNAMAP community. Improved communication, a higher general level
of expertise, and a more closely knit user community can occur.
o The bulletin board is the foundation for electronic software distribu-
tion and data clearinghouse (see Recommendation 8). Once users and
developers are accustomed to electronic transfer of information, more
capabilities can be added.
o The time, effort, and cost to distribute new versions, corrections, and
updates to UNAMAP models and documentation could be dramatically re-
duced. Data requests to NCDC typically take two months from initiation
to receipt. Similar times are required to order UNAMAP from NTIS.
Many of these requests could be answered over the bulletin board within
24 hours.
10.1.3 Time and Cost Estimates
Several bulletin board systems are available. For comparison, any candidate
system should offer bulletins, private electronic mail, distribution lists, and
file transfer. One such system has been developed by Battelle for DOE and is
run on a MicroVax II. To tailor the basic software for EPA, install it on the
Meteorology Division MicroVax, and provide system operator training would
require:
111-61
-------
o Bulletin Board S/W:
tailored, installed,
system manager training
and documentation.
o Hardware:
MicroVax ports (3)
and modems (3).
o 3 telephone lines.
o System management time:
in the DOE application,
a small amount of time is
required for file transfer,
update of bulletin boards, and
message clean-up.
Labor Cost: $4,000 - $7,000
Time: 45 days after installation.
Ports Cost: $1,800 for 8.
Modem Cost: $600 each.
Time: 30 days after order.
Cost: GSA Cost.
Time: 45 days after order.
Staff Time: Approx. 12 hrs/mo.
10.1.4 Alternatives Not Chosen
A PC-based bulletin board system was considered. Such a system would have
certain potential advantages, especially within the PC environment. For
example, compiled (executable) versions of model code could be transmitted and
run on similar PCs without re-compiling. Also, if binary data existed in PC
format, it could be transmitted for immediate execution on a similar PC.
Battelle has previous experience with single-user (PC) as well as multi-user
bulletin board systems. Because of limitations of the PC-DOS operating system,
these systems are only available to a single user at a time on most office PCs.
Our experience has shown that as soon as a single-user system becomes well-
known, functionally beneficial, and used, it becomes overloaded. This is a
source of user frustration, and causes more harm and waste than benefit. Also,
a PC-based system would not be able to provide directly executable code or
binary data to mainframe computers.
111-62
-------
10.2 Recommendation 2: Produce, Distribute, and Support
a Series of End-user Documentation
10.2.1 Description
New documentation should be designed, developed, tested, and distributed to
end-users. The documentation design and organization would be based on a
functional or task orientation. The manuals would be written for the non-
technical reader, tested by novice users, and distributed through the EPA
Regional Meteorologists. Since so many users depend on the manuals to learn
the models, manuals should include some self-paced teaching exercises. Manuals
for each Appendix A and B model would be produced, including Complex I and II.
10.2.2 Benefits of End-User Documentation
Although documentation may not appear to be a vital component of AQSMs, it can
have a greater operational impact on UNAMAP than such internal model changes as
the handling of the receptor grid, for example. High quality, consistent end-
user documentation will increase the uniformity of model application. Good
documentation will reinforce correct interpretation of results with clear,
detailed examples, and reduce the guesswork in most standard modeling situa-
tions.
Clear, easy to use documentation will also reduce requirements for direct EPA
support to end-users. For example, by prominently featuring answers to the
most frequently asked questions or system error messages, improved
documentation would free support staff to deal with more difficult questions or
to perform other functions.
10.2.3 Time and Cost Estimates
Unless there is a supply of user-documentation specialists available to the
Meteorology Division, the initial design and development of end-user manuals
for UNAMAP should be done out-of-house. The following is required:
111-63
-------
o Design and develop
user manuals for 12 Appendix
A and B models including
task analysis, examples,
and self-paced exercises.
Labor Cost: $250,000 - $300,000
Time: 8 to 12 months plus
procurement lead time.
10.3 Recommendation 3: Provide UNAMAP Code, Data,
and Documentation for Multiple Computers
10.3.1 Description
It is inevitable that modeling will continue its move into a distributed en-
vironment with fewer activities done on timeshared mainframes, some moving to
local mainframes or minicomputers, and a great deal on microcomputers.
Battelle recommends that 'EPA actively support multiple computers by providing
software, data, documentation, and support services for a variety of hardware.
UNAMAP software should be available for the following types of computers: IBM
mainframes, DEC VAX, and IBM-compatible PCs. These are the computers most
frequently used for UNAMAP modeling, (See Section 7). This improvement will
directly serve the long-range goals of UNAMAP in many ways. It will make the
conversion process either vastly easier or totally unnecessary, which will
allow new versions of the models to be executed with less expenditure of time
and effort. This ease of upgrading and conversion will in turn encourage the
use of non-regulatory models.
The major effort may be providing input data formats that can be run on all
three computers. As part of the multiple computer support approach, Battelle
recommends that EPA provide, or allow for the use in all models, of
preprocessed data in a form that is directly transferable to an ASCII disk
file. Also, to simplify conversion and maintenance, a standard PC hardware and
software system configuration should be established and communicated to the
UNAMAP community.
111-64
-------
10.3.2 Benefits
o PCs provide a low-cost, readily available alternative when conversion,
compatibility, or access is a problem on mainframes.
o The majority of users surveyed have, and use, IBM mainframes for
modeling. Thus, the largest group of users would be spared the
conversion process. The number of modelers using obsolete versions of
UNAMAP would probably decrease if IBM translations of the current
version were available.
o Time-sharing at NCC would decrease. Users overwhelmingly prefer to buy
UNAMAP (See Table 12).
o The number of models actually installed and available on a user's
mainframe would increase.
o Models could be distributed by an electronic medium or by diskette.
o Some model development, revision, testing, and documentation activities
could be performed on PCs and translated to mainframes.
o The VAX line is the second most commonly available mainframe for UNAMAP
users and is currently used for modeling by 27% of the users.
o The MicroVax is already being used for EOB UNAMAP development. Thus,
little additional conversion of code would be required. This would
also facilitate the distribution of models, documentation, and data
through the MicroVax bulletin board system.
111-65
-------
1
a\
o>
CT
o>
-5
-5
a>
CD
1st Choice
QUESTIONNAIRE RESPONDENTS' PREFERRED SOURCE OF UNAHAP MODELS
2nd Choice 3rd Choice
4th Choice
State
Government
local
Government
Private
Industry
Consultants
Other
TOTALS
RTF NTIS NTIS
lainframe Tape Diskettes Other
4 14 12 6
2134
12 6 4
1 18 13 3
1 1
7 46 35 17
RTP NTIS NTIS
Mainframe Tape Diskettes Other
10 7 6 4
1 4 4
2 3 10
2 9 13 4
1 1
15 24 36 8
RTP NTIS NTIS
Mainframe Tape Diskettes Other
3795
322
5211
11 6 1 3
1
23 17 13 9
RTP NTIS NTIS
Mainframe Tape Diskettes Other
15 4 2 3
3 2
7 3 2
14 1 3 3
1
40 10 7 6
-------
10.3.3 Disadvantages
o Currently, mainframe-oriented UNAMAP preprocessors do not provide a
disk-based ASCII file or the capability to translate into the PC or
MicroVax binary storage format. A standard microcomputer configuration
would require either such a conversion program or an ASCII (not
binary), preprocessed, disk-based (not tape) input file. Transfer and
execution of input data on mainframes is commonly done in ASCII, disk-
based format as well.
o Moving to multiple development or distribution environments will in-
crease "version control" problems, even if a "universal" approach is
taken.
o Users will expect an increasing amount of machine-specific support.
o It may be necessary for the EPA support organization to maintain three
UNAMAP environments. Incidently, a PC system requires proportionally
as much system management as any large computer system.
o The IBM ANSI FORTRAN compiler may not be available to all sites.
Therefore, a compatibility problem may still remain.
10.3.4 Time and Cost Estimates
The following resources will be necessary for development of a "universal"
system:
o A minimum of 2 PC configurations Cost: $6,000 - $8,000
with the standard Fortran. Time: 30 days if purchased
commercially.
111-67
-------
o Translation of UNIVAC specific Cost: $50,000-$75,000
versions of UNAMAP models and Time: 6 to 9 months, plus
data preprocessors to IBM, VAX, procurement lead time.
and IBM PC including documentation.
o Quarter to half time PC Staff Time: 60 hrs/month.
systems manager.
10.4 Recommendation 4: Improve the Accuracy and Technology
of Models Included in UNAMAP
10.4.1 Description
Battelle recommends that EPA continue research to improve the predictive
capability of dispersion models used by the public. Improving the accuracy of
the regulatory models and advancing the modeling technology in UNAMAP are
necessary to continue the viability of the program. Although improvements to
the accuracy of UNAMAP models will not be cheap or easily accomplished,
Battelle recommends that the new UNAMAP include major efforts in this area.
Improving the predictive accuracy of UNAMAP models would be done in three
phases. The first phase will be to improve the peer review process used to
identify accuracy problems within the current models. Improvements could be
gained by involving a greater segment of the UNAMAP community. The second
phase is to conduct additional research to correct these problems and improve
the models. The research will include more model validation and evaluation
activities, within specific situations, for a variety of models. Research will
also be done to improve the scientific basis of dispersion modeling. The third
phase is to have any improvements incorporated into the regulatory process.
Technical discussions of model accuracy can be provided through the bulletin
board and included in the user's manual for each model. Non-technical discus-
sions of accuracy should also be provided through the bulletin board.
111-68
-------
10.4.2 Benefits
o The primary benefit of additional accuracy research is that advances in
modeling technology will occur through the research activities. This
will improve the state of air quality models.
o Increased user knowledge of accuracy and of current research will
increase the use and acceptance of new research models.
10.4.3 Time and Cost Estimates
The activities described for improving the accuracy and technology of the
UNAMAP models are not easily quantifiable either by cost or level of effort.
They represent a direction in which to focus research resources and have the
greatest possible impact on UNAMAP users. The UNAMAP modeling community
considers these activities to be of the utmost importance.
10.5 Recommendation 5: Develop a Consistent Set
of User Interfaces
10.5.1 Description
Battelle recommends that screen-oriented software be added to all UNAMAP
models. The goal is to make all models in UNAMAP appear as much alike as
possible. All conversions of units, formatting to the model's requirements,
decimal conversions, etc. would be performed by the interface. Variations
would be required in the software to allow for variations in source
characterization, such as end points for line sources, and diameters or corners
for area sources. The interface software would determine, based on the model
and the options selected, which data was required. The interface would ensure
that the data was put into the input file in the format needed by that model.
111-69
-------
10.5.2 Benefits
o These interfaces will ease the data input and model setup burden, thus
making UNAMAP models easier to use.
o By providing a similar interface across regulatory and research models,
the use of the non-regulatory models will be encouraged.
10.5.3 Disadvantages
o Because screen-oriented software is specific to a given type of
terminal, separate versions may be required for each type of hardware
(i.e. IBM, VAX, and PC) supported.
o The effort required to provide interfaces to all models and In all
hardware versions is significant.
o A screen-oriented interface residing on an IBM mainframe would require
that users have a terminal capable of interpreting IBM 3270 protocol.
10.5.4 Time and Cost Estimates
The costs below are in terms of estimated contract rates for systems analysts
and senior programmers. Hourly estimates are given so that EPA can estimate
the level of effort to perform the work9 regardless of whether or not it is
contracted. The estimates below are for one hardware version only. They
should be adequate for any of the three environments.
o Develop Specifications 360 hours analyst
for 12 highest priority Labor Cost: $21,600
UNAMAP Models. Time: 90 days
o Develop Interfaces 540 hours programmer
for 12 highest priority Labor Cost: $32S400
UNAMAP Models. Time: 180 days
111-70
-------
o Develop Specifications 300 hours analyst
for 12 remaining UNAMAP Labor Cost: $18,000
Models. Time: 75 days
o Develop Interfaces 450 hours programmer
for 12 remaining UNAMAP Total Cost: $27,000
Models. Time: 120 days
10.6 Recommendation 6: Consolidate All Support
for UNAMAP Models
10.6.1 Description
A single support group should be established to provide a unified voice on
model application, technical issues, data appropriateness, and regulatory
requirements to the users. This could be a telephone "hot line", an electronic
mailbox address, a newsletter, or any combination. The support group and
method of contact would be publicized and communicated to users. This group
would act as the first line of support in all UNAMAP modeling areas, and would
draw on advice from specific experts as required.
10.6.2 Benefits
Two primary benefits accrue from consolidating user support: better direct
service to users, and furthering the EPA goal of consistency. By providing a
single point of contact, the user effort and time needed for a response to a
question or problem would be reduced. The user would always know who to
contact and the support group would decide whether they could answer the
question or refer the user to the outside expert who was most qualified. By
being fulltime, the support staff would quickly gain expertise, thus improving
response time to the user. Consolidated support will provide consistent
responses to the users. A single source for support means that the same
question tends to get the same answer every time. Also, the support task will
become easier as the staff gains experience. Once a question is answered or an
issue resolved, that information becomes part of the support group's knowledge
111-71
-------
base (preferably in an automated troubleshooting log). The knowledge is in
place whenever a similar question or issue arises again.
10.6.3 Time and Cost Estimates
Current support time and costs are difficult to estimate for two reasons.
First, support is currently fragmented among the regions, OAQPS, and ORD thus
making it difficult to compile figures on the support effort expended. Second,
the demand for support in the future will fluctuate significantly as changes
are made. For example, conversion to IBM 3090 will increase support require-
ments; better documentation will reduce requirements. The following estimates
are based on present conditions:
o Plan and establish support Staff time: 12 person-months
system, including training.
o PC, Software, and terminal. Equipment Cost: $5,000
o Operate support branch. Staff time: 300 hrs/month
o Associated equipment: two Equipment Cost: $1,000
telephone lines with answering
machines (or data communications
equipment).
10.7 Recommendation 7; Establish a Meteorology
Data Clearinghouse
10.7.1 Description
A data clearinghouse should be established to provide a central source of
information about the availability of meteorological data. It should include
information about onsite data as well as NWS data. The availability of pre-
processed data should also be included. This clearinghouse could operate as
part of an electronic bulletin board system on the Meteorology Division Micro-
Vax. When available, data could be transmitted electronically.
111-72
-------
10.7.2 Benefits
o The time spent waiting to acquire data could be dramatically reduced in
some cases. For example, data currently managed by OAQPS could be
translated and copied to the MicroVax, then downloaded to the user's
mainframe or PC for inclusion in a modeling project, all within the
same working day.
o The existence of onsite data could be made known to potential, secon-
dary users. Even if the data are proprietary, a contact point for the
owner would be provided. In some cases, the owner might transmit the
data to the clearinghouse for forwarding to the secondary user.
o Emission inventories, topographic data, and other data could all be
made available from a single source in a standardized format.
10.7.3 Time and Cost Estimates
This recommendation is seen to be an internal EPA task; therefore, all
estimates are in terms of staff time. Elapsed time estimates make allowances
for response from other parties where necessary.
Collect information from
data providers, including
lists of data available,
formats, locations, costs.
Staff Time: 200 hours over 3
months.
o Set up bulletin board area,
including bulletin board
specifications for each
different data source, load
data (if applicable), and test.
Staff Time: 64 hours
for each data source.
Time: 30 days
111-73
-------
o Publicize the service to users. Staff Times 72 hours
including preparation of initial Time: 30 days
newsletter and user documentation
for bulletin board area.
10.8 Recommendation 8: Develop or Acquire Specialized
Models for Inclusion in UNAMAP
10.8.1 Description
Battelle recommends that EPA provide specialized models as part of the UNAMAP
library to help the modeling community fulfill all aspects of their obligation
to protect the environment. Specialized models will be designed for use in
situations which are handled poorly or not at all by the current UNAMAP models.
These models may need different modeling algorithms or new functions such as
human exposure or health risk assessment.
To stay abreast of current trends in AQSM, and to increase the applicability of
UNAMAP models, ORD should concentrate development resources on models in the
following areas:
o Shore Breeze
o Complex Terrain
o Spills with Evaporation
o Toxic Gases
o Heavy Gas Models
o Health and Vegetation Effects
Not all these areas are within ORD's current sphere of responsibility. The
first two are logical extensions of existing criteria pollutant dispersion
modeling. The CTDM model is a refined complex terrain model currently under
development by ORD.
111=74
-------
Toxic gas and spills with evaporation are not clearly within ORD's area.
Models dealing with these problems, however, are in demand by state government,
private industry, and consultant users. The modeling community looks to UNAMAP
to meet their expanded modeling needs. ORD has the centralized expertise to
develop these models as well as the respect and influence within the modeling
community and regulatory circles to carry out such a project.
10.8.2 Benefits
o Shore breeze and complex terrain models would improve the accuracy and
credibility of EPA's regulatory modeling efforts. The current "work
arounds" and adaptation of other models would be eliminated.
o Evaporation and toxic gas models have been developed outside of EPA.
By acquiring rights or licenses to such models, ORD can gain exposure
to new types of models and expand their availability to the public.
10.8.3 Time and Cost Estimates
Such models can either be developed or acquired. Model development cost
depends, of course, on the specifications which are to be met. The tasks
required to develop these models are familiar to EPA. One of the specialized
models, complex terrain, is already under development by EPA; thus, its cost
should be fairly well determined.
While the cost of acquiring an existing model is easier to estimate, other
costs must be included. Acquisition estimates should include the following:
o Purchase rights or license Total estimated cost per
model: $20,000
o Installation or conversion Availability to users:
of the model 6 to 9 months
o Learning to use and support
the model
o Preparation of documentation
111-75
-------
10.9 Recommendation 9: Support the Collection of
Additional and More Accurate Meteorological Data
10.9.1 Description
Battelle recommends that EPA work closely with NWS, other government agencies,
and industry to promote the collection and dissemination of meteorological data
which will facilitate the advancement of dispersion modeling technology.
10.9.2 Benefits
There are no specific quantifiable activities which will guarantee that
improved meteorological data is captured. Rather, this recommendation is for
EPA to press for data improvements through publicity, research, interagency
agreements and regulatory changes. By increasing awareness throughout the air
quality modeling community, EPA could increase public support for data
collection improvements. A proportion of ASRL research could focus on the
impact of data on model results. Increased cooperation with the government
agencies responsible for meteorological data collection may eventually result
in better data being available to UNAMAP modelers. In addition to NWS,
approaches to NRC, DOE, and DoD may result in improved modeling data, at least
for certain locations.
The Meteorology Division should continue to work closely with OAQPS to see that
the best possible data is applied to modeling in the regulatory environment.
This may take the form of additions to the guideline models, or the allowance
of modifications to certain models.
As data may be the limiting factor in improving the accuracy of models in
UNAMAP, improvement in the data available will contribute to the goal of
advancing research in modeling. Also, as data contributes to improved
accuracy, it will increase the credibility of modeling to industry and the
public.
111-76
-------
10.9.3 Estimated Cost and Time
The initiatives described above do not have a directly quantifiable cost or
level of effort. They represent a certain direction in which to focus both
research and management communication resources. Although the resources of the
Meteorology Division are extremely limited, the importance of this area
dictates that EPA implement some program to improve the data available for air
quality modeling.
10.10 Recommendation 10: Support the Electronic Transfer
of UNAMAP Documentation
10.10.1 Description
Battelle recommends that EPA make UNAMAP documentation available to modelers
through electronic file transfer. The electronic documentation could be
supported on the bulletin board system (See Section 10.1). Users could
download a copy of any UNAMAP model's current documentation whenever needed.
Therv, paper output and reproduction would be under the control of the user.
Updates could be incorporated into the user's copy as needed.
10.10.2 Benefits
Electronic availability would greatly speed the dissemination of documentation,
both in the original form and for updates. This improvement increases the
extent of UNAMAP support offered through a central location. Indirectly, the
use of non-regulatory models is facilitated through increased awareness and
accessibility of their documentation.
10.10.3 Time and Cost Estimates
The major effort associated with this improvement will be to develop the format
specifications and any format conversion programs which may be required. If
this recommendation is implemented in concert with the documentation changes in
Section 4.2, the UNAMAP documentation will be revised and will then exist in
111-77
-------
some electronic form. The costs attributable to this recommendation are then
limited to selecting and publicizing the standard packages.
o Select standard graphics and Labor Cost: $5,000 - $6,000
word processing packages. Software Cost: $l,000/copy
Time: 60 days
o Publicize standards to users Labor Cost: $2,400
(distribute newsletter to users Time: 20 days
and put on bulletin board).
111=78
-------
11.0 ALTERNATIVES NOT RECOWENDED
Several suggestions and alternatives are not included in our recommendations.
These are briefly discussed in this section.
11.1 Graphics Output for Models
The graphics standards and output capability of users vary widely. This issue
is most easily resolved in the PC environment, where the EGA display and HP
7475 plotter are defacto standards. Bowman, Cleary, and Trinity are providing
good products in this area.
11.2 Improving Response Time
Modeling on microcomputers will increase and the computers will become faster.
These factors will make response time for UNAMAP-scale AQSMs less of an issue.
Larger, more complex models, however, must address the execution time issue as
they are developed.
111-79
-------
12.0 SUMMARY
This Interim Report has reported the results of the UNAMAP interviews and
questionnaire, described the users' concerns in detail, and specified
Battelle's recommendations for improvements in all aspects of the UNAMAP
program.
The variety of problems and suggestions provided by UNAMAP users served to
identify areas where the UNAMAP program needs improvement to provide better
service to the air quality modeling community. The list of major problems
included both technical concerns (such as model accuracy) and non-technical
issues (such as user support).
Ten recommended improvements were described in the report. For each
recommendation, its benefits over the present system were listed. The
disadvantages of some recommendations were identified where significant. Time
and cost estimates were generated for each improvement. Taken in their
entirety, the recommendations included in this report can greatly increase the
accessibility of UNAMAP and improve the models' ease of use.
111=80
-------
Appendix A: Interview Guide
-------
Name:
Organization:
UNAMAP INTERVIEW GUIDE
BACKGROUND
1. Would you give us your job title:
Title:
2. What is the nature of your involvement with UNAMAP or other AQSMs:
User Developer Software support
Data provider Remarketer Validator
None Other
3. Have you worked for other organizations where you were involved with AQSMs?
yes no
If "yes", what were the organizations and what was your involvement (use the
categories above):
ORGANIZATION INVOLVEMENT
4.How long have you been involved with AQSMs?_
A-l
-------
5. Which UNAMAP models have you been involved with:
USED HOW LONG?
1 yr. or less
2-4 years
5 years or less
UNAMAP MODEL
MODEL NAME:
VERSION #:
SUBROUTINES USED:
Briefly describe any modifications you've made to the model:
I TIMES YOU
USE MODEL
per week
per month
per year
USED HOW LONG?
1 yr. or less
2-4 years
5 years or less
UNAMAP MODEL
MODEL NAME:
VERSION I:
SUBROUTINES USED:
Briefly describe any modifications you've made to the model:
* TIMES YOU
USE MODEL
per week
per month
per year
USED HOW LONG?
1 yr0 or less
2-4 years
5 years or less
UNAMAP MODEL
MODEL NAME:
VERSION is
SUBROUTINES USED:
I TIMES YOU
USE MODEL
per week
per month
per year
Briefly describe any modifications you've made to the model:
-------
USED HOW LONG?
1 yr. or less
2-4 years
5 years or less
UNAMAP MODEL
MODEL NAME:
VERSION I:
SUBROUTINES USED:
Briefly describe any modifications you've made to the model:
I TIMES YOU
USE MODEL
per week
per month
per year
6. Which non-UNAMAP models have you been involved with:
USED HOW LONG?
1 yr. or less
2-4 years
5 years or less
NON-UNAMAP MODEL
MODEL NAME:
VERSION #:
SUBROUTINES USED:
Have you modified the model? Yes No
Main purpose of the model ( e.g., predicts CO concentrations):
Who developed the model:
Briefly describe any advantages this model has over UNAMAP:
I TIMES YOU
USE MODEL
per week
per month
per year
USED HOW LONG?
1 yr. or less
2-4 years
5 years or less
NON-UNAMAP MODEL
MODEL NAME:
VERSION I:
SUBROUTINES USED:
Have you modified the model? Yes No
A-3
f TIMES YOU
USE MODEL
per week
per month
per year
-------
Main purpose of the mode] ( e.g., predicts CO concentrations)
Who developed the model:_ ____________
Briefly describe any advantages this model has over UNAMAP:_
USED HOW LONG?
NON-UNAMAP MODEL
I TIMES YOU
USE MODEL
1 yr. or less
2-4 years
5 years or less
MODEL NAME:
VERSION I:
SUBROUTINES USED:
Have you modified the model? Yes No
Main purpose of the model ( e.g., predicts CO concentrations):
per week
per month
per year
Who developed the model:
Briefly describe any advantages this model has over UNAMAP:
INTERVIEWER:If interviewee is not a user of AQSM, skip to 2.4.b on page
7. Following are some of the processes involved in using AQSMs. Review with us
the steps you follow and the information you use in these proceses:
What are the factors which influence your decision to use AQSM:
% of projects where you use AQSMs:
A-4
-------
INTERVIEWER: Be sure the following points are included for each:
% TIME STEPS INFORMATION
PROCESS SPENT FOLLOWED USED
* * ************************************************************
Choosing a model
Collecting data:.
Waiting/Arranging for
computer access:
Inputting data/
model parameters:_
A-5
-------
Waiting for result/
printing:
Verifying
output format:
Us i ng/1nterpret i ng
output :
8. Is the product of the AQSMs worth the costs (time, money, etc.)?
9c How important is UNAMAP or other models to your job?
11. How would you do your job without UNAMAP or other models?
What is the hardest part of using the model s?_
A-6
-------
INTERVIEWER: Use this section for all categories other than USERS of AQSMs
otherwise skip to page *
13. Describe the different processes you go through in fulfilling your role with
UNAMAP or other models.
9. How important is UNAMAP or other models to your job?_
11. How would you do your job without UNAMAP or other models?_
12. How did you learn about the models(s):
A-7
-------
INTERVIEWER: Use the following codes for training methods:
T = Training course, seminar, workshop S = Self-taught
U = University course while pursuing degree 0 = One-on-one training by co-worke
MODEL
TRAINING METHOD
DIFFICULTY IN LEARNING
T
T
T
T
T
T
T
T
U
U
U
U
U
U
U
U
S
S
S
S
S
S
S
S
0
0
0
0
0
0
0
0
Rank the models on the difficulty you experienced in learning them (i.e., which
model was easiest to learn, which was the next easiest, etc.).
INTERVIEWER: If the following question is answered YES skip to the questions for
that response; otherwise, skip to the question for NO respondents.
14. Do you use manuals or other items of documentation for the models you are
involved with? YES NO
YES respondents;
a. How useful are the manuals or other documentation to you?
b. What difficulty do you experience in using the manuals or other
documentation?
NO respondents:
a. Why do you not use the documentation?
A-8
-------
15. If you have a problem or need information about a model, which organizations do
you talk to:
INTERVIEWER: Be sure the following points are covered:
a) Model name
b) Organization providing the support
b) Type of support you request
c) How the support is made available
d) User satisfaction with the support on the following points:
correctness of answer, promptness in returning calls, timeliness of the
answer, hours when availabile, any difficulty in understanding, and
promptness in following up (documentation, software corrections, letters,
etc.).
The following abbreviations are used in the chart:
Type of support: How support is. make available:
C - Choosing a model to use
0 = Supplies data T = Telephone
S = Software problems S = Your site
R = Interpreting results 0 = Other
P = Programming questions
TYPE OF SUPPORT
ORGANIZATION:
HOW AVAILABLE
C 0 S R P
MODEL NAMES
00000
00000
00000
How satisfied are you with the support:
T
0
0
0
S
0
0
0
0
0
0
0
TYPE OF SUPPORT ORGANIZATION:
C D S R P
MODEL NAMES
00000
00000
00000
HOW
T
0
0
0
AVAILABLE
S 0
0
0
0
0
0
0
How satisfied are you with the support:
A-9
-------
TYPE OF SUPPORT
C D S R P
ORGANIZATION:
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
MODEL NAMES
How satisfied are you with the support:
HOW AVAILABLE
T S 0
000
000
000
16. Do you give support to AQSM users? YES NO
IF YES:
TYPE OF SUPPORT
C D S R P
000
000
000
0
0
0
0
0
0
TO WHOM:
MODEL NAMES
What are the most common problems or questions:
HOW AVAILABLE
T S 0
0
0 0
0 0
0 0
17. What kind(s) of computer hardware are available to your organization for AQSM?
INTERVIEWER: Be sure the following points are covered:
a) equipment available at the site and any equipment you use offsite (e.g.,
through telecommunications)
b) If they use that equipment for AQSMs
c) The manufacturer and model for the equipment. If the equipment is located
elsewhere and they don't know the manufacturer, put the organization where
the equipment is located.
A-10
-------
AVAILABLE
ONSITE OFFSITE HARDWARE MANUFACTURER & MODEL
/ / / / Mainframe computer
/ / / / Mini -computer
/ / / / Microcomputer
/ / / / CRT Terminals
/ / / / Printers
/ / / / Graphics plotters
/ / / / Telecommunications
YOU USE
FOR AQSMs
YES NO
YES
YES
YES
YES
YES
YES
NO
NO
NO
NO
NO
NO
18. Do you pay for your present hardware use?_
If so, how much?
19. If UNAMAP models were available on another computer for no cost, would you
object to additional/new hardware you might have to learn to use? Yes No
20. How do you assess your ability to get your job done with the current models and
computer systems?
21. We will suggest a number of possible problem areas. If these seem to
characterize your involvement with UNAMAP, briefly describe how:
Buying/locating AQSMs and their updates:
A-ll
-------
Choosing a model and/or subroutine for a project:
Documentation for the model:
Accuracy of model:
Data collection:
Data/Parameters input:
Hardware problems (access, compatibility,reliabi!1tyeetc):
Output format:
A-12
-------
Usability of output:
Lack of software support:
Organizational snafus (list the organizations and the problems):
Lack of standards:
Excess costs:
Other:
A-13
-------
22. For those problem areas you mentioned, what would remedy your biggest headaches
in those areas? -
23. Can you estimate the value of the benefits that would be achieved by your
suggested changes?
24. How would your effectiveness improve?
25. Why might your suggestions not be implemented?
A-14
-------
26. How do the regulatory requirements affect your involvement with UNAMAP and
other models?
27. What changes 1n regulations would have an effect on your involvement with
UNAMAP and other models?
28. Are you aware of significant coming changes in AQSM systems or requirements
(regulations, AQSM trends, computer/graphics technology, other)?
A-15
-------
29.Do you have personal observations about the job, systems, or the organization you
feel would be useful to the UNAMAP evaluation?
A-16
-------
Appendix B: Questionnaire
-------
QUESTIONNAIRE INSTRUCTIONS
WHAT: This questionnaire is part of an evaluation of the UNAMAP models
undertaken by Battelle on behalf of the U.S. EPA. The objective of the
survey is to discover how extensively various models and subroutines are
being used, the types of computers utilized, and any perceived needs which
are not being met by EPA's UNAMAP models. We are interested in the
capabilities and advantages of non-EPA models as well.
WHO: The questionnaire has been sent to a large number of organizations,
including state air pollution control agencies, federal agencies,
contractors, consultants, and industrial groups. You have been selected
either because you have purchased UNAMAP or because your organization is
known to be a part of the air quality modeling community. Your involvement
is important to shape improvements in the UNAMAP models.
HOW: Please answer as many of the questions as possible. Although most of
the questions are multiple choice, some will ask you to "briefly describe"
some aspect of your involvement with models. These answers increase the
quality of the information because they reflect your specific situation.
Please use the margins or back of pages if there is too little space for
your answer. Since the questionnaire will be going to many types of users
of UNAMAP, the exact wording may not always be applicable to you. In those
cases, answer the questions in a way which best expresses your involvement
with air quality models. A completed sample questionnaire has been
included to help you.
WHY AND WHEN: The questionnaire is intended to take a minimal amount of
your time. Please complete it promptly and return the questionnaire in the
postage-paid envelope included. We would like to have the questionnaire
returned by February 20, 1987. It must be returned by February 28, 1987 to
be included in the survey. All respondents will receive a copy of the
survey results. If your name is not on the label below or if the address
is incorrect, please enter the correct information in the space provided:
Name:
Address:
This questionnaire has been submitted for approval to the Office of
Management and Budget (OMB) under the Paperwork Reduction Act of 1980, 44
U.S.C. 3501 et seq.
B-l
-------
UNAMAP QUESTIONNAIRE
1. Circle the term below which best describes the nature of your current work with
UNAMAP or other models:
User Developer Software support
Data provider Remarketer Validator
Other None
2. If you worked with UNAMAP or other models in a previous job, please circle a_H
of the terms below which apply to that job:
User Developer Software support
Data provider Remarketer Validator
Other_
Organization Approx. Dates_
3. Check the time period which covers your involvement with air quality models:
/ / 1-4 years /_ / 5-10 years /___/ 10 years or more
4. The following section asks for information about the air quality simulation
models with which you currently work. The models you use most should be
described first. NOTE: Separate answer areas are provided for UNAMAP and non-
UNAMAP models. UNAMAP models are on pages 2 and 3, Other models begin on page
4.
a) Check the box which describes how long you've used the model;
b) Fill in the blank with the number of times per week, month, or year that you
use the model;
c) Please use the back of this page to describe more than four models.
I TIMES YOU
USED HOW LONG? UNAMAP MODEL USE MODEL
1-4 years MODEL NAME: per week
5-10 years VERSION I: per month
H°w did you obtain/access the model:
10 years or more -^.^_^_=_^=______m.m_________^m___. Per
Briefly describe any modifications you've made to the model:
B-2
-------
USED HOW LONG?
1-4 years
5-10 years
10 years or more
UNAMAP MODEL
MODEL NAME:
VERSION I:
How did you obtain/access the model?
Briefly describe any modifications you've made to the model:
I TIMES YOU
USE MODEL
per week
per month
per year
USED HOW LONG?
1-4 years
5-10 years
10 years or more
UNAMAP MODEL
MODEL NAME:
VERSION I:
How did you obtain/access the model?
Briefly describe any modifications you've made to the model:
I TIMES YOU
USE MODEL
per week
per month
per year
USED HOW LONG?
1-4 years
5-10 years
10 years or more
UNAMAP MODEL
MODEL NAME:
VERSION I:
How did you obtain/access the model?_
Briefly describe any modifications you've made to the model:
I TIMES YOU
USE MODEL
per week
per month
per year
B-3
-------
5. Why did you choose to use the particular UNAMAP model(s) you described above?
USED HOW LONG?
1-4 years
5-10 years
10 years or more
NON-UNAMAP MODEL
MODEL NAME:
VERSION I:
How did you obtain/access the model?_
Have you modified the model? Yes No
Main purpose of the model ( e.g., predicts CO concentrations):
I TIMES YOU
USE MODEL
per week
per month
per year
Who developed the model:
Briefly describe any advantages this model has over UNAMAP:
USED HOW LONG?
1-4 years
5-10 years
10 years or more
NON-UNAMAP MODEL
MODEL NAME;
VERSION I:
How did you obtain/access the model?
Have you modified the model? Yes No
Main purpose of the model ( e.g., predicts CO concentrations):
f TIMES YOU
USE MODEL
per week
per month
per year
Who developed the model:
Briefly describe any advantages this model has over UNAMAP:
B-4
-------
# TIMES YOU
USED HOW LONG? NON-UNAMAP MODEL USE MODEL
1-4 years MODEL NAME: per week
5-10 years VERSION I: per month
How did you obtain/access the model?
10 years or more ___ per year
Have you modified the model? Yes No
Main purpose of the model ( e.g., predicts CO concentrations):
Uho developed the model:
Briefly describe any advantages this model has over UNAMAP:
6. Why did you choose to use the particular (non-UNAMAP) model(s) you specified
above?
7. Draw a circle around the number on the scale below to indicate the importance
of UNAMAP or other AQSMs to your job. The farther to the right the mark, the
more important they are:
1 2 3 4 5 . 6 7 8 9 10
couldn't care less mid point no job without them
8. How would you do your job without UNAMAP or other models?
B-5
-------
9. Use the chart below to indicate how you learned to use the models with which
you are involved:
a) Give the model name
b) Circle the abbreviation for the training methods by which you learned the
model. The abbreviations are:
T = Training course, seminar, workshop S = Self-taught
U = University course while pursuing degree 0 = One-on-one training by co-worker
c) Rank the models on the difficulty you experienced in learning them (i.e.,
put a "1" by the model easiest to learn, a "2" for the next easiest, etc.):
MODEL TRAINING METHOD DIFFICULTY IN LEARNING
T
T
T
T
T
T
T
T
T
T
T
U
U
U
U
U
U
U
U
U
U
U
S
S
S
S
S
S
S
S
S
S
S
0
0
0
0
0
0
0
0
0
0
0
If you use manuals or other documentation for the models, circle the number on the
scale which best describes your response to the following two questions.
10. Are the manuals or other documentation easy to understand and use?
0
extreme
1
ly hard
2
3
4
5
6
7
8
9
10
very
easy
11. How essential are the manuals or other documentation to your work with the
models?
0
not at al
1
! 1
2
3
4
5
6
7
8
9
10
totally
-------
12. Specify below the organizations from which you request support on the models
you work with:
a) List the models you ask that organization about
b) Circle the letter(s) to specify the type of support you request
c) Circle the letters(s) to specify how the support is made available.
d) Rate your satisfaction with the support on the following points:
correctness of answer, promptness in returning calls, timeliness of the
answer, and follow-up (documentation, software corrections, letters, etc.)
Specify your level of satisfaction on each point by using a scale where 0 '
not satisfied and 10 = completely satisfied.
The following abbreviations are used in the chart:
Type of support:
C = Choosing a model to use
0 = Supplies data
S = Software problems
R = Interpreting results
P = Programming questions
How support is made available:
T = Telephone
S = Your site
0 = Other - Specify:
TYPE OF SUPPORT
ORGANIZATION:
C
C
C
C
0
0
0
D
S
S
S
S
R
R
R
R
P
P
P
P
MODEL NAMES
Satisfaction: Correctness
Timeliness
Return calls
HOW AVAILABLE
T S 0
T S 0
T S 0
T S 0
Followups
TYPE OF SUPPORT
C
C
C
C
0
D
D
0
S
S
S
S
R
R
R
R
P
P
P
P
ORGANIZATION:
HOW AVAILABLE
MODEL NAMES
Satisfaction: Correctness
Timeliness
Return calls
B-7
-------
TYPE OF SUPPORT ORGANIZATION:__ HOW AVAILABLE
MODEL NAMES
C D S R P T S 0
C D S R P T S 0
C D S R P T S 0
C D S R P ISO
Satisfaction: Correctness Timeliness Return calls Followups
13. Use the chart below to specify the kinds of computer hardware that are
available in your organization for AQSM:
a) Put a check in the box(es) to show where the equipment available (e.g., at
your site or through telecommunications)
b) Circle "YES" or "NO" on the right to describe your use of that equipment for
AQSMs
c) Give the manufacturer and model for the equipment. If the equipment is
located elsewhere and you don't know the manufacturer, name the organization
and location where the equipment is located.
AVAILABLE
ONSITE OFFSITE HARDWARE MANUFACTURER & MODEL YOU USE
FOR AQSMs
/_/ / / Mainframe computer YES NO
/__/ /_/ Mini-computer YES NO
/_/ /__/ Microcomputer YES NO
/_/ /__/ CRT Terminals YES NO
/__/ /__/ Printers YES NO
/_/ /__/ Graphics plotters YES NO
/__/ / / Telecommunications YES NO
14. Rate your preference of the sources of UNAMAP models which are listed below.
1 = most preferred, and 4 = least preferred
Rating
1. Mainframe computer (not UNIVAC) at RTP
2. Tape from NT IS
3. PC-compatible diskettes from NTIS
4, Other ^.^^^^^^^^^^^^^^^^^^
B-8
-------
15. Below we list a number of possible problem areas with AQSMs.
problems seem to characterize your involvement with UNAMAP or
models, rate them using the following scale:
If any of these
other other
0 1 2
not a problem
10
PROBLEM:
RATING:
PROBLEM:
severe problem
RATING:
1. Choosing a model
2. Accuracy of model
3. Buying/accessing the model
4. Documentation for the model
5. Support for the model
6. Data collection
7. Data input
8. Hardware compatibility
9. Hardware access
10. Unreliable hardware
11. Response time
12. Other
16. For the four most significant problems which you identified, briefly describe
what would remedy your biggest headaches. Use the following scale to estimate
the value of the suggested change and how the change would improve your
effectiveness:
1
8
10
only a little
very much
Problem
No.
Suggestion:
Rate the suggestion as to its impact on your effectiveness.
Problem
No.
Suggestion:
Rate the suggestion as to its impact on your effectiveness.
B-9
-------
Problem
No.
Suggestion^
Rate the suggestion as to its impact on your effectiveness
Problem
No.
Suggestion:
Rate the suggestion as to its impact on your effectiveness
17. On the following scale, circle the number which best characterizes the impact
of the "EPA Guidelines on Air Quality Models" (EPA-450/2-87-027R) on your
involvement with UNAMAP and other models:
no
0 1
impact
2
3
4
5
6
7
8
9
10
major
impact
18. Briefly describe any changes in air quality regulations which would increase or
decrease your involvement with UNAMAP and other models. Also, describe any
changes which would improve the quality of model data in EIA's.
19. Are you aware of significant coming changes in modeling technology or systems
(AQSM trends, computer/graphics technology, other) which will have an impact on
your work with air quality models?
B-10
-------
20. Do you have personal observations about the job, systems, or the organization
you feel would be useful to the UNAMAP evaluation?
B-ll
-------
EVALUATION AND ASSESSMENT OF UNAMAP
PART IV: IMPLEMENTATION PLAN
February 1988
-------
TABLE OF CONTENTS FOR PART IV
Page
1.0 INTRODUCTION IV-5
2.0 PROJECT BACKGROUND IV-8
2.1 Technology Assessment Overview IV-8
2.2 Data Collection and Analysis Techniques Overview IV-10
2.3 Problem Area Overview IV-11
2.3.1 Accuracy IV-13
2.3.2 Documentation IV-14
2.3.3 Support IV-14
2.3.4 Data Collection IV-15
2.3.5 Data Input IV-16
2.3.6 Hardware Compatibility IV-16
3.0 RECOMMENDED IMPLEMENTATION APPROACH IV-18
3.1 Changes in the UNAMAP Program IV-18
3.2 Research vs. Technology Transfer IV-19
3.3 Strategic Objectives IV-20
4.0 IMPLEMENTATION OF RECOMMENDATIONS IV-23
4.1 Recommendation 1: Establish an Electronic
Bulletin Board..... . IV-23
4.1.1 Implementation Tasks.......... IV-24
4.1.2 Estimated Costs and Time IV-25
4.2 Recommendation 2: Produce, Distribute, and Support
a Series of End-User Documentation IV-26
4.2.1 Implementation Tasks IV-28
4.2.2 Estimated Costs IV-29
IV-1
-------
Page
4.3 Recommendation 3: Provide UNAMAP Code, Data,
and Documentation for Multiple Computers IV-31
4.3.1 Implementation Tasks IV-32
4.3.2 Estimated Costs and Time IV-33
4.4 Recommendation 4: Improve the Accuracy and Technology
of Models Included in UNAMAP IV-34
4.4.1 Implementation Tasks IV-34
4.4.2 Estimated Costs and Time IV-35
4.5 Recommendation 5: Develop a Consistent Set of
User Interfaces IV-36
4.5.1 Implementation Tasks .... IV-38
4.5.2 Estimated Costs and Time.. IV-39
4.6 Recommendation 6: Provide Centralized Support
for All Models IV-41
4.6.1 Implementation Tasks IV-42
4.6.2 Estimated Costs and Time IV-43
4.7 Recommendation 7: Establish a Data Clearinghouse.... IV-44
4.7.1 Implementation Tasks IV-45
4.7.2 Estimated Costs and Time IV-46
4.8 Recommendation 8: Develop Specialized Models IV-46
4.9 Recommendation 9: Support the Collection of
Additional and More Accurate Data IV-47
4.9.1 Implementation Tasks...... IV-48
4.9.2 Estimated Costs and Time. IV-49
4.10 Recommendation 10: Support the Electronic Transfer
of UNAMAP Documentation......... IV-49
4.10.1 Implementation Tasks IV-49
4.10.2 Estimated Costs and Time IV-50
5,0 SCHEDULE IV-52
5=1 Detailed Schedule for Phase I Tasks....... IV-53
5.2 Budget and Schedule Summary for Phase I IV-55
IV-2
-------
Page
6.0 SUMMARY IV-58
APPENDIX IV-60
IV-3
-------
LIST OF FIGURES FOR PART IV
Page
Figure 1. Problem Areas Ratings of Seriousness.... .... . IV-12
Figure 2. Source Characteristics Data Screen.. IV-37
Figure 3. Schedule for Phase I Improvements.. IV-56
Figure 4. Estimated Cost and Start Times for
Each Phase I Recommendation.......................... IV-57
IV-4
-------
EVALUATION AND ASSESSMENT OF UNAMAP
PART IV: IMPLEMENTATION PLAN
1.0 INTRODUCTION
The User's Network for Applied Modeling of Air Pollution (UNAMAP) is a software
library of air quality simulation models provided by the Environmental
Operations Branch (EOB) of EPA's Atmospheric Science Research Laboratory
(ASRL). The Evaluation and Assessment of UNAMAP project was designed to
facilitate EOB's ongoing efforts to improve the utility and availability of
UNAMAP to the public. The project was divided into four major parts, of which
this report is the last.
The purpose of this report is to describe a plan for implementing a series of
recommended improvements to the UNAMAP program. The plan consists of a
definition of strategy consistent with EPA's objectives for UNAMAP, and the
schedule and budget to implement the specific recommendations.
The improvements were derived during the course of the investigation, and were
based on two major sets of requirements. Technology requirements determine
what computer, data, and modeling technology are available to UNAMAP users now
and in the near future. User requirements determine what areas of UNAMAP
utilization are most difficult and determines where improvement would be most
beneficial. Each recommendation has been formulated both to take advantage of
the current technological environment and to help meet expressed and implied
user requirements.
Because the recommendations are dependent on previous phases of the project,
Section 2 of this report provides a summary of the information collected in the
earlier portions of the evaluation effort. Each of the following areas will be
summarized: (1) assessment of the technological environment, (2) data
collection and analysis to determine user requirements, and (3) the major
problem areas where user requirements are not being met.
IV-5
-------
The UNAMAP program has great visibility and impact on the public, including
industry, state air pollution control agencies, and community groups. To take
advantage of this visibility, UNAMAP needs to provide a centralized modeling
service of high quality. The recommended improvements, when taken as a whole,
will allow UNAMAP to achieve the following goals:
o To continue as the public source of newly developed and refined air
quality models.
o To distribute models which are easily executed on a variety of
commonly-used computers of all sizes.
o To provide a wide set of models which are relatively easy to execute,
even for the novice or occasional user.
o To offer modelers a central source of technical information,
meteorological data, and user support.
To attain these goals, a long-range strategy and milestones are required.
Therefore, the first part of the implementation plan is a strategic framework.
This will serve as a guideline for the assignment of priorities to the
recommendations consistent with the long-term objectives of the UNAMAP program.
The strategy is designed to strike a balance between the two objectives of
UNAMAP: advancement of research in dispersion modeling, and the effective
transfer of modeling technology to the public. This strategic framework is
presented in Section 3 of the report.
The strategy will be executed through a set of specific improvements. A
description of the tasks and estimated costs to implement the recommended
improvements will comprise the majority of this report. Battelle is making ten
specific recommendations for improvement to UNAMAP- They are the following:
o Establish an electronic bulletin board on a multi-user computer system.
o Provide a set of end-user documentation for all UNAMAP models.
o Provide models which execute on IBM mainframes, DEC VAX, and IBM PCs,
o Improve the accuracy of models,
o Develop consistent user-friendly interfaces for all models.,
o Consolidate all support for all UNAMAP models.
IV-6
-------
o Establish a meteorological data clearinghouse.
o Include more special purpose models in UNAMAP.
o Support the collection and use of additional and more accurate
meteorological data.
o Support the electronic distribution of UNAMAP documentation and
updates.
Section 4 of the report summarizes each recommendation, and discusses
implementation tasks and estimated costs.
Based on its contribution to the strategic framework, each improvement can be
assigned a priority and a completion time estimate. These two factors are used
to generate an implementation schedule. The schedule is presented in Section
5.
IV-7
-------
2.0 PROJECT BACKGROUND
The conclusions and recommendations described in this report are based on
research and analysis which were performed in three phases. The first was a
technology assessment to determine technology available to UNAMAP. The second
was data collection (by interview and questionnaire) and analysis. The third
was the derivation of recommendations. These three phases are summarized in
this section.
2.1 Technology Assessment Overview
All products follow a pattern of growth which involves changes to the product,
the users, and the market. UNAMAP has followed such a growth pattern. The
Technology Assessment Report evaluates the technological environment in which
UNAMAP exists today and that in which it will function most effectively in the
future. The information was used in subsequent stages of the project to
provide evaluation criteria for system alternatives and final recommendations.
The technology assessment found that all components of modeling technology have
changed dramatically since the development of the first computerized air
quality models in the 1960s. The technology is continuing to change at a fast
pace. The report covers three distinct phases in air quality model
development: (1) the technology as utilized by UNAMAP Version 6; (2) the
technologies used by models currently under development as well as adaptations
made to UNAMAP to utilize current technology; (3) the technological
requirements for future models to solve current problems.
In Version 6, the technology applied to modeling has progressed, while that of
the computer systems used to run them has not. The UNAMAP models have become
more sophisticated and consistent, while adaptations and processors have
improved their usability. The UNIVAC 1100 utilized to support UNAMAP, however,
represents no advances since UNAMAP was begun in 1973.
IV-8
-------
New modeling developments have grown from the modeling community's attempts to
deal with today's air quality modeling applications. Model developers are
creating models for more challenging environmental problems. UNAMAP users have
adapted the models to respond to situations that are different or more complex
than those for which the software was designed. Consultants and third-party
vendors have contributed more sophisticated data collection and input methods
and have facilitated the use of new computer technology.
Modeling software must advance to meet both application and user needs.
Modeling has become a successful and important tool in protecting the
environment. Because of this success, models are needed for even more complex,
real-world situations which need representation. Large, complex regional
models are needed by county, regional, and state agencies to study dispersion
and transport of pollutants over larger distances. More realistic complex
terrain models are needed to represent geographical areas where terrain is a
factor.
Users need software which not only fits the application, but is also easier to
use. In judging ease of use, the user-friendly, microcomputer products
available for other applications will be used as criteria against which the
UNAMAP software will be judged.
The data used by the models will continue to be provided by the National
Weather Service (NWS) or collected at the site in question. Therefore more
consistency of data input and output among the models will be needed to
facilitate the novice or infrequent user.
Computer technology will continue its trend toward distributed processing with
microcomputers used for an increasing amount of modeling activity, including
graphics output. User expectations will also require that data gathering and
file transfer technologies in UNAMAP match those available in other application
areas.
The typical model user has changed since the advent of UNAMAP. Today, a user
may not be a "modeling expert". He may use the models only a few times a year
IV-9
-------
or lack the computer expertise to use mainframe versions easily. Services such
as modeling consultants and developers of menu-based microcomputer versions
have grown to meet the needs of today's user.
2.2 Data Collection and Analysis Techniques Overview
A major portion of the Evaluation and Assessment of UNAMAP project was the
collection of data from UNAMAP users and other members of the dispersion
modeling community. The data helped define who uses UNAMAP and how the models
are used. The data also identified the areas of the modeling process which
users felt should be improved.
There were two parts to the data collection activity: personal interviews and
mailed questionnaires. In-depth interviews were conducted with 22
representatives of the following organizations'
o EPA (ASRL and OAQPS)
o EPA Regional Meteorologists
o State air quality agencies
o Local or county air quality agencies
o Private industries (as users of the models)
o Educational institutions
o Modeling consultants
o Modeling software marketers
The interviews provided an overview of the UNAMAP system as well as information
on the modeling process. The data from the interviews were used to design a
questionnaire which was mailed to a larger segment of the UNAMAP community (256
organizations).
Usable data were returned in 112 questionnaires. These questionnaires were
analyzed, and certain types of data were extracted including a user profile,
which models are used, and problems encountered in the modeling process. A
typical UNAMAP user:
IV-10
-------
o Belongs to one of four major industry groups: consultant, state
government, private industry, or local government;
o Classifies himself as a user of the models;
o Has an experience level which varies by industry group.
The largest portion of the questionnaire dealt with the problems perceived by
the users and their suggestions for improvement. Twelve potential problem
areas were identified by Battelle project team members based on conversations
with EPA. Questionnaire respondents were asked to rank the problem areas as to
the severity of the problem. User responses were grouped into low, medium, and
high categories. Problem areas scored as medium or high on the severity scale
by all major user groups were further analyzed. User suggestions to correct
these problems were categorized, and a percentage of suggestions in each
category was computed.
Alternatives were generated from the suggestions and from the experience of the
Battelle staff in solving similar problems. Evaluation criteria and
constraints were identified, and used to determine the strengths and weaknesses
of the alternatives.
The suggestions and alternatives were evaluated using the technological context
described in the Technology Assessment and Battelle1s previous professional
experience. Through the evaluation process, Battelle developed a list of
recommendations. The recommendations were formulated to address multiple
problem areas wherever practical, and to utilize technology which is available
to the UNAMAP support staff and model users. The implementation of these
recommendations comprise the bulk of this report (Section 4).
2.3 Problem Area Overview
Six out of the original twelve potential problem areas were rated as major
problems by the users. These are the problems which our recommendations
attempt to rectify. They are discussed in this section. Figure 1 shows the
areas specified as major problems as well as the potential problems not
considered significant by most users.
IV-11
-------
C
-5
O
cr
CD
m
10
8
MAJOR PROBLEM
Documentation
Support
Data Collection
Accuracy
Data Input
IBM
VAX
PC
Hardware
Compatibility
70
01
3
w»
o
-t»
t/>
-1
o
3
ro
Buying/Accessing
a Model
NOT A MAJOR PROBLEM
TOOLS
Unreliable
Hardware
Choosing
a Model
Enter
Hardware
Access
Response Time
-------
2.3.1 Accuracy
A totally accurate model would produce concentration estimates which exactly
match actual measurements of air quality. Models, of course, can only strive
to meet this standard. A majority of users consider the current models to fall
too far short of this goal. Model users also want a better method for
interpreting and using information on the degree of model accuracy and its
limits. It is particularly important that management decision-makers know how
to evaluate the accuracy of the model results.
While most of the obvious improvements to the accuracy of UNAMAP models have
been made, some alternatives which can improve model accuracy were identified:
o Conduct additional research
o Establish an improved dissemination process for those evaluations and
validations
o State the model results in terms of confidence intervals
Alternatives were evaluated on cost, degree of improvement, and breadth of
impact within the user community. Some constraints had to be considered as
well. First, only a limited number of thorough model evaluations have been
conducted. Second, the EPA objective to protect the environment by preferring
over-prediction to under-prediction can sometimes act to constrain model
accuracy by eliminating models which may be highly accurate in some cases, but
may under-predict in other cases.
IV-13
-------
2.3.2 Documentation
The data indicated that documentation in the form of user's guides and system
documentation is very important to the modelers. Many users stated that the
manuals were inconsistent, not oriented for the novice or infrequent user, and
changes to the manuals did not get to those who needed them.
There are two alternatives to improve ease of use. First is a separate set of
user manuals for the novice or infrequent user with a "cookbook" approach and
more examples. Second is to rewrite the existing manuals adding new ease-of-
use features and improving the consistency. The distribution of documentation
would be quickened and simplified by use of electronic media.
The documentation alternatives must be considered in relation to several
constraints.
o Preparing documentation is time-consuming and expensive;
o Existing distribution channels are not conducive to frequent updates;
o EOB does not have the personnel resources available to assume
distribution responsibilities.
The proliferation of user-friendly, microcomputer products for other
applications has led to a generally increased level of expectation in terms of
documentation clarity and ease of use. The UNAMAP system must recognize these
expectations and employ current documentation techniques and standards.
2.3.3 Support
User support consists of information and assistance provided to answer
questions and give guidance as to the proper application and successful
execution of the UNAMAP models. Currently UNAMAP support service
IV-14
-------
responsibilities are divided. EOB helps users with the modeling software while
the Office of Air Quality Policy and Standards (OAQPS) answers questions on
regulatory issues.
Overall, support was rated as at least a medium-level problem by all user
groups. The quality of support was not the issue; users indicated that UNAMAP
support is too fragmented and decentralized.
There are three alternatives to standardize UNAMAP support:
o Support activities could be integrated into the Regional Offices;
o Centralized support could be provided by a Modeling Support Branch
within EPA;
o Standardized model components and a help facility could be developed to
reduce support requirements.
All support alternatives must be judged on the level of service given to users
and cost effectiveness. Support alternatives should also be evaluated in the
context of other planned improvements such as documentation changes and user
interfaces. The primary constraint on support, of course, is staff resources.
2.3.4 Data Collection
Collection of meteorological data is a significant part of air quality
modeling. There are two sources of meteorological datas data from a National
Weather Service (NWS) installation near the site or data collected at the site
itself. Users complain that there is no comprehensive source of information on
what data are available, and how to get it.
One alternative is a data clearinghouse. Such a clearinghouse could offer a
variety of services from a simple newsletter to an electronic bulletin board.
The services to be provided and the medium of communication may be limited by
IV-15
-------
the initial effort required. The volume of data available to the clearinghouse
may constrain its services or medium.
2.3.5. Data Input
After data has been collected, the data input phase is required to resolve any
data problems, format the data, and enter items such as receptor locations and
other terrain data. Data input is often a manual function although
preprocessor programs can correct several types of data problems.
UNAMAP users want data input to be made easier. Novice or infrequent users,
especially, must expend a great deal of effort in determining how to enter data
for each specific model. There are three alternative approaches: change the
data, change the models, or provide an interface between the data and the
models.
The main constraint for data input alternatives is the level of effort required
to implement the various approaches. A full-screen, interface program, for
example, will require a separate version for each type of terminal protocol to
be supported.
2.3.6 Hardware Compatibility
There are many types of computer hardware used for modeling by the UNAMAP
community, but one brand (UNIVAC) has been the "standard". Users of other
hardware convert the software or find a company selling the version they need.
Although UNIVAC has been the standard version of the models since they began,
EPA's National Computer Center (NCC) is converting all UNIVAC operations to
IBM. Future models will be supported on hardware other than the UNIVAC.
Whatever the standard version, users of other hardware may be required to
convert the models. The ideal hardware compatibility solution would provide a
version of the UNAMAP models for every type of hardware used by the modeling
community. This ideal cannot be met with current resources.
IV-16
-------
Alternative approaches to increase hardware compatibility fall into two
categories: support multiple hardware or make the conversion process easier.
Resources are major constraints on hardware compatibility alternatives. Choice
among these alternatives must be based on cost effectiveness and usefulness to
the modeling community. Support of multiple types of hardware would benefit
the users, but would be extremely costly in terms of resources. Easing the
conversion process would also require a major effort.
IV-17
-------
3.0 RECOMMENDED IMPLEMENTATION APPROACH
UNAMAP is the primary mechanism used to promote two of the ASRL Meteorology
Division's major objectives: research in dispersion modeling and transfer of
the technology to the public. Research keeps the field of dispersion modeling
technically viable and publicly credible. Technology transfer is the process
by which the results of research are made available for use by the public.
Because these two objectives are interdependent, any strategy for improving
UNAMAP must consider both. To restate the conclusions of the Technology
Assessment (Section 2.1), changes have occurred in the UNAMAP user community,
and in the UNAMAP software library itself since UNAMAP was begun. The
questionnaire results indicate that these changes have resulted in a reduced
effectiveness of the UNAMAP program in meeting its two objectives.
3.1 Changes in the UNAMAP Program
UNAMAP began as a centralized, timeshared system used by a relatively small
group of experienced modelers whose needs were, for the most part, being met.
These modelers were users, developers, and evaluators. Their intimate
knowledge of the models, as well as their involvement in multiple aspects of
the dispersion modeling program, limited the need for support and ease-of-use
features. The UNAMAP program of 1987 is different in two significant respects.
The most significant change to the UNAMAP program since it was established in
1973 has been a growth in the number of users and in the purposes for which
they use the models. This trend has affected UNAMAP in several ways. First,
users are not as likely to be intimately involved in dispersion technology or
in model development as were users in the past. Novices and infrequent users
rely on experienced personnel to help them. As a result, support needs have
increased. Second, at the same time support needs have increased, the growing
size of the community has reduced communication between users and model
developers. This has placed an additional burden on EOB to provide all
support. Third, the increased number of users has resulted in models being
IV-18
-------
converted to and executed on a wide variety of computers. This has greatly
increased the complexity of providing usable code, data, and documentation.
The second significant change to UNAMAP is the inclusion of an increased number
of models produced by many different developers. In 1978, Version 3 of the
UNAMAP library consisted of 11 models, all directly supported by EOB. Version
6, released in 1986, contains 25 models, developed by a variety of consultants,
industry, and EPA contracts. Besides the increase in the number of models,
users find it difficult to adjust to the differences between models. These
differences can be significant because models are developed by individuals or
institutions to meet specific needs. As a result, consistency and
standardization among them has not been a goal.
These two changes have generated a requirement for the UNAMAP program to become
more user-oriented. This implies that (1) the novice user must get more
consideration, and (2) there must be greater emphasis on uniformity of
operation for the models.
3.2 Research vs. Technology Transfer
The new UNAMAP must emphasize both practical and research concerns. As air
quality modeling has become an integral part of the regulatory process
technology transfer has become more of a challenge. Modelers are under
pressure to run the models and produce results to be used for regulatory
purposes. For many users, these concerns take priority on a day-to-day basis.
On the other hand, there are certain aspects of the models that require serious
study. For example, the accuracy of models has not been evaluated for many
sets of conditions. Also, the technology utilized in many of the regulatory
models is outdated.
There are several possible interpretations of what constitutes technology
transfer. A narrow definition implies that the technology can be transferred
by transmittal. In such an approach, the technology is made available to the
public in more or less the same format as that in which it was developed.
Alternatively, a broader definition emphasizes the quality of the transfer with
IV-19
-------
varying levels of information and adaptation of the technology to meet the test
of usability. Because of the previous sophistication of its user community,
UNAMAP has always leaned toward a fairly narrow definition of technology
transfer. To be effective in the future, UNAMAP must meet the needs of a
larger, more varied user group.
3.3 Strategic Objectives
Any changes must address the current state of user requirements and the current
set of models first. Only from this base can advancements be made. Therefore,
approaches discussed in this section will first expand the users' ability to
use the models, and then serve to advance the modeling technology being used.
The initial changes must allow UNAMAP to evolve to meet the current
expectations of the modeling community. When using the regulatory models is
less difficult, modelers will be able to turn more of their attention to the
research aspects of modeling. As in the early days of UNAMAP, the user
community will become more involved in the process of testing and validating
new models as part of the research cycle.
The new "Strategic Vision" for UNAMAP, presented here, is also an acceptance of
its evolution toward a more distributed system with modelers using a variety of
computer hardware to meet different modeling requirements. UNAMAP began as a
centralized program, but changed to a distributed environment in response to
technological advances in computer systems and the growth of the user
community. Instead of being restricted to a timeshared system, modelers now
have access to computers at the regional and local level and to microcomputers
on their own desks. They run models on whichever computer hardware is most
accessible, easy to use, and cost-effective.
There are three objectives which, if achieved, will promote the growth of the
new UNAMAP:
1) To continue improving the models in UNAMAP through research
2) To provide a single source of information for modelers
3) To make all UNAMAP models easier to use
IV-20
-------
These objectives will promote the two functions of BOB: research and
technology transfer.
The research objectives of the UNAMAP program must continue. Battelle
recommends that the research focus on improving the accuracy of the general
purpose models and on the development of new special purpose models. Because
of the importance of the model results in the regulatory process, accuracy is
an important issue to users and decision-makers. More research will be
necessary to improve the accuracy of models being used in the permitting and
EIS applications.
New models are also important to the UNAMAP community. Again, largely because
of increased use of models for regulatory purposes, models are needed which
address new problems and new terrains. UNAMAP, as the largest single source
for air quality modeling software, is perceived by the public as the logical
origin for all dispersion modeling software.
The UNAMAP program must also take a lead role in providing a centralized
information source for all modeling needs. EOB staff members currently supply
some support. Users must, however, go elsewhere to order the models, purchase
meteorological data, and get certain types of support for some models and
applications. Users expect centralized support through UNAMAP and are confused
at the actual complexity of the support network.
Battelle's approach recommends that a new dimension be added to the technology
transfer by making UNAMAP a central source for all air quality modeling
technology. This "one stop shopping" concept should provide modelers with
information regarding the following topics:
o Meteorological data
o Technical advice
o Model documentation
o Modeling software updates
o Support services
IV-21
-------
The new UNAMAP will provide this technology transfer by using electronic
communications for improved transfer of model changes, documentation updates
and meteorological data to the modeling public. This capability will also
serve to transfer technology secondarily by facilitating communication among
modelers.
To serve the current modeling community, the models must be made easier to use.
Ease of use includes software which can be run on the various types of computer
hardware widely used for modeling. It also includes solving such user problems
as lack of consistency between models, the need for user interfaces, and
improvements in user documentation.
Battelle's strategy for UNAMAP has two thrusts. The initial focus is to make
the system easier to use. The goal of this focus is to make the modelers more
efficient at performing their regulatory duties. Through increased ease of
use, modelers will be able to concentrate on modeling techniques rather than on
mechanics. The second stage is predicated on the assumption that once they can
execute the models more easily, users will attempt to use the newer models*
They will experiment with their data to compare how various models perform in
specific instances. This will increase the base of users knowledgeable in the
newer models. These users can become the impetus for improved technology in
the regulatory portion of the UNAMAP program, and can provide valuable input to
the future direction of air quality modeling.
IV-22
-------
4.0 IMPLEMENTATION OF RECOMMENDATIONS
This section describes the implementation of specific recommended improvements
which will promote the increased effectiveness of UNAMAP. The recommendations
do not individually correspond to specific problems described in Section 2.
Instead, they are designed to support the long-range objectives of the UNAMAP
program and remedy the underlying causes of current problems.
Achievement of the strategic objectives will help UNAMAP continue to fulfill
its leadership role in air quality modeling.
4.1 Recommendation 1: Establish an Electronic Bulletin Board
Battelle recommends that EPA establish an electronic bulletin board on the
Meteorology Division MicroVax computer. This bulletin board system would
provide the capability to send messages to a specific person or groups,
broadcast messages to all users, and transfer files to and from the system.
The bulletin board would increase the user-oriented aspect of UNAMAP. The
initial uses of the bulletin board would be to provide UNAMAP support (answer
questions via electronic mail) and to distribute information of value to the
UNAMAP community at large. Minor software patches and documentation updates
could also be provided over the bulletin board.
Later, the services provided could increase as computer resources and
communication lines allow. For example, the system could be expanded to allow
distribution of new versions or releases of the models, new manuals, and
possibly data.
The bulletin board could also increase use of the research models. Currently,
some research models are included on the UNAMAP tape in the hope that modelers
will use them and provide input into the testing and validation of these new
models. The models, however, go out indiscriminately to all purchasers with
little chance for "marketing."
IV-23
-------
4.1.1 Implementation Tasks
Battelle recommends the following task structure to provide a bulletin board
system for UNAMAP users. This workplan is based on Battelle's experience in
bulletin board development for another government agency. The plan contains
several tasks listed below:
Task 1: Software Evaluation
o Prepare a set of requirements and evaluation criteria to include such
functions as: bulletins, private electronic mail, distribution lists,
and file transfer.
o Develop a list of available multi-user bulletin board software.
o Determine which of the available systems meets the initial evaluation
criteria.
o Set up demonstrations of software chosen for further review.
o Develop a list of in-depth evaluation criteria (for example, ease-of-
use, flexibility, growth potential, etc.).
o Evaluate the systems demonstrated.
o Choose the software which fits UNAMAP's needs.
Task 2: Hardware Procurement and Implementation
o Order modems, cables and communication ports for MicroVax.
o Order telephone lines.
o Set up modems.
o Test all equipment together.
Task 3: Implement Software on Meteorology Division MicroVax
o Determine changes which must be made to the basic software.
o Tailor the basic software for EPA (welcome message, bulletin names,
etc.).
o Install the software system on the MicroVax.
IV-24
-------
o Test the system with the hardware.
o Provide system operator training.
o Prepare a means of advertising the bulletin board system to the user
community.
o Prepare an end-user tutorial/documentation packet.
4.1.2 Estimated Costs and Time
We have based our cost estimates for a bulletin board system on a similar
software system developed by Battelle for DOE. The software provides
bulletins, private (encrypted) electronic mail, distribution lists, and file
transfer functions.
o Labor (at Contractor rates):
Task 1: 40 hours
40 hours x $60 = $2,400
Time = 60 calendar days
Task 2: 20 hours
20 hours x $60 = $1,200
Time = 60 calendar days
Task 3: 40 hours
40 hours x $60 = $2,400
Time = 30 calendar days
TOTAL 100 hours x $60 = $6,000
o Hardware (list prices):
MicroVax communication ports $1,520 for 8 (DHV-11)
+ 330 installation = $1,850
Modems (Racal Vadic 2400 PC) 3 x $ 600 each = $1,800
Cables 3 x 10 = $ 30
TOTAL $3,680
-------
o Other costs:
3 telephones lines (GSA Cost)
Staff time = 12 hours/month
The necessary management time constitutes the system's operating cost. The
system manager should be an EOB staff member. Based on the DOE application, we
estimate approximately 12 hours per month will be required for file transfer,
update of bulletin boards, and message clean-up.
Battelle estimates that it will take 60 calendar days to select software and
tailor it to fit EPA's needs. The other major time constraint will be the time
required for delivery of the necessary hardware and telephone lines. The
MicroVax ports and modems have an estimated arrival time of 30 days after
order. EPA experience at RTP indicates that approximately 45 days are needed
for telephone line installation. Testing, training, advertising and the
tutorial will require about 45 calendar days after the hardware is installed.
4.2 Recommendation 2: Produce. Distribute, and Support a Series
of End-User Documentation
Battelle recommends that improved end-user documentation be developed to
enhance the user-orientation approach of UNAMAP. The existing documentation is
not sufficiently functional to play the important role users expect of it. As
the UNAMAP user community continues to expand, the demand for and reliance on
high quality documentation will increase, as will the importance of
documentation in supporting system acceptance among growing numbers of less
technically oriented users.
New documentation should be designed, developed, tested and distributed to end-
users. Battelle recommends an initial documentation design concept based on a
preliminary reader analysis that will ensure that user requirements are met.
Specifically, this design would result in documentation that:
IV-26
-------
o Acknowledges and addresses multiple types of users;
o Is easy to use for various needs of all user types -- specifically for
initial learning, routine reference, and additional background;
o Provides consistency from one model to the next, yet allows for the
variation inherent in each model;
o Exists for all models;
o Can be prepared and distributed in an efficient manner;
o Lends itself well to revisions and updates.
The documentation design and organization would be based on a functional or
task orientation. The manuals would be written for the non-technical reader,
tested by novice users. Since so many users depend on the manuals to learn the
models, manuals should include some self-paced teaching exercises. Manuals for
each Appendix A and B model would be produced, including Complex I.
Documentation can be provided using various media and through different
channels. The documentation could be produced in a paper version and
distributed through the EPA Regional Meteorologists. This alternative
addresses some of the current problems identified by users. The closer contact
between the regions and users would provide the documentation and revisions in
a timely manner to the actual users rather than to the name on the purchase
order.
A second alternative would be to produce the user manuals in electronic format
and distribute them through the electronic bulletin board. The distribution of
the documentation and updates would be much faster and paper output could be
controlled by the user. Section 4.10 discusses this alternative in more
detail. An advantage of either alternative over the current system is that
revisions could be made to the manuals more easily. The structure recommended
for the manuals uses a modular approach which will allow changes to be made in
individual sections without affecting the rest of the document.
The recommended organization for the revised documentation takes all these
requirements into consideration. The organization separates general overview
IV-27
-------
material, specific functional instructions and procedures, and reference for
various types of user needs and interests. The annotated outline included in
Appendix A identifies a general plan for the major sections in the proposed
revision to UNAMAP documentation.
4.2.1 Implementation Tasks
The implementation tasks are based on Battelle's well tested documentation
design and development methodology. The plan includes four tasks: project
design and planning, first draft, revised draft, and final copy. The
activities involved in the four tasks are listed here, with EPA review cycles
indicated.
Task 1: Project Design and Planning
o Define project scope and limitations (specific goals, schedule,
expectations, requirements).
o Collect resources (existing documentation, related materials, access to
software, contact with subject matter experts/experienced users, as
required).
o Learn use of model.
o Refine initial reader analysis.
o Develop document design (detailed working outline), to include:
- introductory and overview paragraphs
- notes about additional information required from EPA
- sample page designs
- binding recommendations.
o Deliver refined reader analysis and documentation design (working
outline) to EPA.
o Adopt or develop style guide.
o Secure EPA approval.
-------
Task 2: First Draft
o Building on working outline and incorporating EPA comments, complete
first draft, with all sections as complete as possible.
o Test first draft.
o Edit first draft to reflect feedback from testing effort.
o Deliver first draft to EPA.
o Secure EPA approval.
Task 3: Revised Draft
o Incorporate EPA comments on initial draft.
o Do all required revisions and editing.
o Retest revised draft.
o Edit revised draft to reflect feedback from testing effort.
o Deliver revised draft to EPA.
o Secure EPA approval.
Task 4: Final Copy
o Incorporate EPA comments on revised draft.
o Conduct final documentation test.
o Edit to incorporate feedback from final testing effort.
o Print final camera-ready art.
o Conduct final production proofreading.
o Deliver final camera-ready art to EPA.
o Secure EPA approval.
4.2.2 Estimated Costs
For the purposes of estimating the cost of this improvement, UNAMAP models have
been grouped into two general categories simple and complex.
IV-29
-------
Simple $17,000 x 15 simple models = $255,000
Complex $34,000 x 9 complex models = $306,000
TOTAL = $561,000
The time required to develop the documentation for one simple model is
estimated at about 200 hours for one FTE technical writer/documenter, and 66
hours for a modeling systems analyst. For one complex model, the estimate is
about 400 hours for one FTE technical writer/documenter and 132 hours for a
modeling systems analyst.
Costing Assumptions and Detail
The following assumptions were used to develop the cost/time estimates:
1. The examples we have (COM for simple, ISC for complex) are typical for
each type model.
2. Camera ready art (CRA) and electronic media in EPA-specified formats
will be the final product to EPA from the documentation developer.
3. The finished documentation will be used as a paper-based tool, but
may be distributed electronically for local printing by users.
4. Existing documentation will be provided by EPA to the documentation
developer in paper format only*
5. Technical writers/documenters will have access to software and to
subject matter experts and expert users as required.
6. Documentation revision begins after conversion of code to ANSI F77
(Section 4.3) and interfaces (Section 4.4).
30
-------
Time estimates in days, by task are as follows:
Simple Complex
Doc. Spec. Sys. Doc. Spec. Sys,
Task 1:
Task 2:
Task 3:
Task 4:
TOTAL
4.3
Project Design
First Draft
Revised Draft
Final Copy
Recommendation
and Planning 30
90
40
40
200
3: Provide UNAMAP Code,
30
30
6
0
66
Data, and
60
180
80
80
400
60
60
12
0
132
Documentation
for Multiple Computers
Battelle recommends that EPA actively provide UNAMAP software for the following
types of computers: IBM mainframes, DEC VAX, and IBM-compatible PCs. These
are the computers most frequently used for UNAMAP modeling, according to the
Battelle questionnaire responses. This improvement will directly serve the
long-range goals of UNAMAP in many ways. It will make the conversion process
either vastly easier or totally unnecessary, which will allow new versions of
the models to be executed with less expenditure of time and effort, and it will
make the non-regulatory models more available.
There are two possible approaches to implement this recommendation. The first
is to distribute distinct versions of UNAMAP for each computer type. Users
would specify which version they wanted when ordering UNAMAP. The second
alternative is to provide "universally executable" (ANSI Standard F77) code and
to provide data which can be loaded and accessed in the same way by all three
types of computers. Under this approach, UNAMAP documentation would need to
contain instructions for loading data and running the models on each type of
computer. To support the IBM PC version, a standard compiler should be
selected and supported.
Based on our recent analysis of UNAMAP models, Battelle recommends the second
approach. In our analysis, we performed experiments with two UNAMAP models,
COM 2.0, and ISCST. These experiments found that only very minor modifications
IV-31
-------
were required to produce code that executed in all three environments. The COM
documentation indicated that the model ran without modification on an IBM
mainframe under MVS and Fortran; the JCL for setting up the batch job was
included in the documentation as well. An OPEN statement for the punched
output file was added for the VAX version and for the PC version. On both VAX
VMS F77 and on Microsoft Fortran, we compiled with the compiler set to the ANSI
standard version (no extensions). Results from all three computers were
identical.
The major effort may be providing input data formats that can be run on all
three computers. A standard microcomputer configuration would require an ASCII
(not binary), preprocessed, disk-based (not tape) input file. Transfer and
execution of input data is on mainframes in this format as well. Currently.
mainframe-oriented UNAMAP preprocessors do not provide a disk-based ASCII file.
As part of the multiple computer support approach, Battelle recommends that EPA
provide, or allow for the use in all models, of preprocessed data in a form
that is directly transferable to an ASCII disk file.
4.3.1 Implementation Tasks
Task 1: Modify all models to execute on IBM, VAX and IBM PC
o Obtain models and install on all environments
o Obtain data
o Make known modifications in advance. Test. Make modifications
which emerge during testing.
Task 2: Make necessary modifications to data preprocessors and model code for
preprocessed ASCII disk file input
o Convert data from binary to ASCII.
o Modify preprocessor to produce ASCII characters to a disk-based
file.
o Convert model code to accept ASCII data from a disk-based file.
Test.
IV-32
-------
Task 3: Incorporate machine-specific execution instructions, any remaining
machine-specific modifications, and updated data input instructions
into documentation.
4.3.2 Estimated Costs and Time
Task 1 and Model portion of Task 2:
o Labor to fix and test
5 simple models at 8 hours = 40 hours
(ANSI F77 and IBM version is currently documented - e.g., CDM-2.0)
40 hours x $60 = $2,400
10 simple models at 32 hours = 320 hours
(No ANSI or IBM version currently exists)
320 hours at $60 = $19,200
9 complex models at 60 hours = 540 hours
(older, less structured code, more options, less we 11-documented -
e.g., ISC)
540 hours x $60 = $32,400
o Computer time: approx. $5,000
o Total cost = $59,000
o Estimated time = 150 calendar days (two persons)
Task 2: Preprocessor part only
o Labor to fix and test
Two preprocessors at 80 hours = 160 hours
160 hours at $50 = $8,000
o Estimated time = 30 days (two persons)
IV-33
-------
Task 3:
This estimate only considers the incremental portion necessary to add these
specific portions into the documentation as it is revised in accordance with
Section 4.2.
o Labor to update documentation
24 models at 16 hours = 384 hours
384 hours at $60 = $23,040
4.4 Recommendation 4; Improve the Accuracy and Technology
of Models Included in UNAMAP
Battelle recommends that EPA continue research to improve the predictable
capability of dispersion models used by the public. Improving the accuracy of
the regulatory models and advancing the modeling technology in UNAMAP are
necessary to continue the viability of the program. Although improvements to
the accuracy of UNAMAP models will not be cheap or easily accomplished,
Battelle recommends that the new UNAMAP include major efforts in this area.
4.4.1 Implementation Tasks
Although specific research activities are best defined by EPA researchers,
there are several alternatives available which will address one or more of the
accuracy issues raised by users. Each alternative requires different skills
and resources to accomplish.
The first alternative is to do what is required to improve the predictive
accuracy of UNAMAP models. This would be done in three steps. The first step
is to compile known accuracy problems from all sources. The second step is to
conduct additional research to correct these problems and improve the models.
The research will include more model validation and evaluation activities,
within specific situations, for a variety of models. Research will also be
done to improve the scientific basis of dispersion modeling. The third step is
to have any improvements incorporated into the regulatory process.
IV-34
-------
A second alternative is to provide the modeling community with information on
the accuracy of the models. This would require an improved dissemination
process. The information must be presented for two audiences: dispersion
modelers and decision makers. Technical discussions of model accuracy can be
provided through the bulletin board and included in the user's manual for each
model. Non-technical discussions of accuracy should also be provided through
the bulletin board.
Advances in modeling technology will occur through the research activities.
These advances must, however, go through testing, evaluation, and review
processes. Members of the UNAMAP user community must be encouraged to become
part of these processes.
A third alternative is to implement several changes in the ways models are
developed and tested to facilitate user participation. First, new research
models should function with a screen-oriented user interface, such as the one
described in Section 4.5. This will expedite their use and acceptance.
Second, the bulletin board could be used to recruit persons interested in
testing a new model. Currently, new models are put on the UNAMAP tape in the
hope that they will be used. The bulletin board could be used to publicize the
models. Third, new models could be downloaded via the bulletin board, run by
the modeler, and comments returned through electronic mail. This would greatly
speed the testing process.
4.4.2 Estimated Costs and Time
The activities described for improving the accuracy and technology of the
UNAMAP models are not easily quantifiable either by cost or level of effort.
They represent a direction in which to focus research resources and have the
greatest possible impact on UNAMAP users. The UNAMAP modeling community
considers these activities to be of the utmost importance. Resources need to
be dedicated to these activities.
IV-35
-------
4.5 Recommendation 5: Develop a Consistent Set of User Interfaces
Battelle recommends that screen-oriented software be added to all UNAMAP
models. The goal is to make all models in UNAMAP appear as much alike as
possible. By easing the data input and model setup burden, this improvement
will directly support the objective of making UNAMAP models easier to use. In
addition, by providing a similar interface across regulatory and research
models, the use of the non-regulatory models will be encouraged.
In general, the interface software would first allow the user to specify
execution options, then proceed through the input specifications (based on the
options selected), in the same order for all models. For example, after
options were specified, the software could display a "source characteristics"
data screen such as the one shown in Figure 4-1.
All conversions of units, formatting to the model's requirements, decimal
conversions, etc. could be performed by the interface.
Variations would be required in the software to allow for variations in source
characterization, such as end points for line sources, and diameters or corners
for area sources. The interface software would determine, based on the model
and the options selected, which data was required. The interface would ensure
that the data was put into the input file in the format needed by the model.
Because screen-oriented software is specific to a given type of terminal,
separate versions could be required for each type of hardware (i.e. IBM, VAX,
and PC) supported. Alternatives to distinct versions are: (1) to write all
screens in ANSI Cobol, which is portable between IBM and VAX, or (2) write the
interface to execute on a PC, which transmits the job stream and input file to
a VAX or IBM mainframe. Battelle recommends the latter approach. In any case,
the effort required to provide interfaces to all models and in all hardware
versions is significant.
IV-36
-------
c
(0
r\j
o
c
o
ii (D
I
GO
o
p>
P>
o
rt
n>
n
to
CO
n
n>
n>
CHARACTERISTICS OF EMISSION SOURCES
MODEL NAME
SOURCE SOURCE
ID NO. TYPE
1 1 qTTF 1 1
SOUR
HEIGI
(METEI
CE
HT SOURCE LOCATION
RS) LAT. LONG.
.CD CD. CD. CD CD. CD. CD
.CD CD. CD. CD CD. CD. CD
.CD CD. CD. CD CD. CD. CD
.CD CD. CD. CD CD. CD. CD
.CD CD. CD. CD CD. CD. CD
.CD CD. CD. CD 0.0. CH
.CD CD. CD. CD CD. CD. CD
.CD CD. CD. CD o.o.o
.CD CD. CD. CD CD. CD. CD
.CD CD. CD. CD CD. CD. CD
.CD CD. CD. CD O. O. O
.CD CD. CD. CD CD. CD. CD
.0 CD. CD. CD a. a. n
.CD CD. CD. CD CD. CD. CD
EMISSION RATE
KG UNIT TIME
-------
To maximize the benefit of the interfaces, the sequence of implementation must
be based on the numbers of users who would benefit. Responses to Battelle's
questionnaire indicated that the computers most commonly used for UNAMAP
modeling are, in order, IBM-compatible PCs, IBM mainframes, and DEC VAXes,
Microcomputer interfaces for regulatory models are currently available
commercially, and thus, would not need to be duplicated as a high priority.
IBM mainframe interfaces become the next candidate. The VAX interfaces should
then follow the IBM versions.
If the PC-based option is not selected, a screen-oriented interface for IBM
mainframes would require that users have a terminal capable of interpreting IBM
3270 protocol. These terminals are generally available. However, Battelle
recommends that EPA poll UNAMAP users to verify that 3270-compatible terminals
are used in the UNAMAP community before beginning the interface effort.
The sequence of model interfaces must also be planned carefully and in
accordance with the long-term objectives to be served by this improvement. On
the one hand, the objective of making the models easier to use would indicate
that regulatory models which are most difficult to set up (e.g., ISC) should be
the initial implementations. On the other hand, such an approach does not
encourage the use of new modeling technology. Battelle recommends that the
initial interfaces be built for 12 regulatory models in UNAMAP (Appendix A and
B) which are least likely to be replaced in the near future. Then, the
remaining regulatory models and the research models could be retro-fitted with
interfaces*
4.5.1 Implementation Tasks
Task 1: Develop Specifications for 12 Regulatory Models
o Select models
o Review models and documentation
o Determine possible options
o Analyze data input requirements for all option combinations
o Prepare logical flows and screen layouts for each model
IV-38
-------
Task 2: Develop Interfaces for 12 Regulatory Models
o Write IBM JCL
o Write screen formatting programs
o Write logic to link screens to options
o Test on IBM
o Write VAX "JCL"
o Test on VAX
Task 3: Develop Specifications for Remaining Models
o Review models and documentation
o Determine possible options
o Analyze data input requirements for all option
combinations
o Prepare logical flows and screen layouts for each model
Task 4: Develop Interfaces for Remaining Models
o Write IBM JCL
o Write screen formatting programs
o Write logic to link screens to options
o Test on IBM
o Write VAX "JCL"
o Test on VAX
4.5.2 Estimated Costs and Time
These costs are in terms of estimated contract rates for systems analysts and
senior programmers. Hourly estimates are given so that EPA can estimate the
level of effort to perform the work, regardless of whether or not it is
contracted.
IV-39
-------
Task 1: Labor for specifications for 12 models
o Six simple models at 20 hours = 120 hours
120 hours x $60 = $7,200
o Six complex models at 40 hours = 240 hours
240 hours x $60 = $14,400
o Total cost: $21,600
o Estimated time: 90 calendar days
Task 2: Labor for development of interfaces for 12 models
Obviously, this effort depends on the approach taken: multiple
versions, a portable Cobol version, or a PC-based version with VAX and
IBM options. The estimates below are for one version, and should be
adequate for any of the three approaches.
o Six simple models at 30 hours = 180 hours
180 hours x $60 = $10,800
o Six complex models at 60 hours = 360 hours
360 hours x $60 = $21,600
o Total: $32,400
o Estimated time: 180 calendar days
Task 3: Labor to Develop Specifications for 12 models
o Nine simple models at 20 hours = 180 hours
180 hours x $60 = $10,800
o Three complex models at 40 hours = 120 hours
120 hours x $60 = $7,200
o Total cost: $18,000
o Estimated time: 75 calendar days
IV-40
-------
Task 4: Labor to Develop Interfaces for 12 Models
o Nine simple models at 30 hours = 270 hours
270 hours x $60 = $16,200
o Three complex models at 60 hours = 180 hours
180 hours x $60 = $10,800
o Total cost: $27,000
o Estimated time: 120 calendar days
4.6 Recommendation 6: Provide Centralized Support for All Models
Battelle recommends that a centralized support group be established within EPA.
The staff of the "Modeling Support Branch" would have expertise on all the
UNAMAP models, a variety of hardware, and would be authorized to convey
regulatory requirements within certain conditions. Improved methods of
communication between support staff, users, and developers would increase the
speed and quality of the support services offered to the modeling community.
It is unlikely that the staff of the Modeling Support Branch would have no
other responsibilities. However, it is crucial that all staff members who
participate in the support group perceive themselves as members of a distinct
and important entity. Staff would need to participate in construction and
implementation of a quality assurance plan for all support activities. They
must also be able to see professional recognition for their work in this area.
The electronic bulletin board could be very important in the provision of
centralized support services. The bulletin board would allow users to call at
any time and describe their problem with the necessary depth of detail.
Support staff would not have their work interrupted by answering the telephone
calls. Instead, they could use specific blocks of time to get the questions
from the bulletin board, determine the answer (using other expert sources if
necessary), and post the answer to the user on the bulletin board. Users will
receive a better quality answer, often in less time, than in the "telephone
tag" approach.
IV-41
-------
The electronic bulletin board (or some other mechanism) should be used to log
questions and answers. A database of previously answered questions would
increase the ability of support staff to answer questions without further
recourse to outside experts. When a question has been answered once and
entered into the log, it is available for all other support personnel.
Everyone then benefits from the collective experience. The log can also be
used to gauge support activities for budget purposes and resource allocation.
4.6.1 Implementation Tasks
The following tasks are necessary to achieve the goal of centralized support.
This plan is based on Battelle's experience in setting up similar support
services to meet another client's needs.
The critical factor for support activities is the level of service given to
users. Goals and standards for the necessary level of service must be
determined. Standards could include time for initial response to user, time to
solve the problem, frequency of referring the user elsewhere, and other related
measures. Support Branch personnel should implement a quality assurance plan
for all support activities.
Task 1: Plan for Support Group
o Define support scope and limitations including goals, schedule and
resource requirements.
o Collect existing documentation, question/answer logss list of
outside experts, etc.
o Determine standards for support services, such as initial response
time and time to solve the problem.
IV-42
-------
Task 2: Establish Support Group
o Set up electronic bulletin board for support services.
o Develop a user packet/marketing brochure to publicize the new
services and tell modelers how to use them.
o Establish the question/answer log format.
o Obtain separate access to computers (IBM, VAX, PC).
o Obtain training on models and hardware as appropriate.
Task 3: Operate the Support Branch
o Respond to calls.
o Manage the question/answer log.
o Publish notes and notices.
o Initiate model software updates.
o Update documentation.
4.6.2 Estimated Costs and Time
Good user support obviously is worth some price. The primary constraint is
staff resources. Current staffing levels are inadequate to handle any
increased level of support. More staff time will be necessary. However, the
time necessary to provide support would decrease as staff members become more
expert in the new areas of knowledge and as the database of questions and
answers grows. A support staff training program would also be necessary in the
beginning which will increase start-up costs. Support staff should ideally be
personnel with UNAMAP experience. The estimates of staff time are based on the
assumption that the Model Support Branch will reside within the ASRL, and will
be initially staffed by three persons (not full time). Three is the minimum
recommended to assure coverage during all normal working hours, West Coast
hours, vacations, etc.
IV-43
-------
Task 1: Staff time to plan
o 2 staff members for 2 months = 4 person-months
Task 2i Staff time to establish support system and receive training
o 2 staff members for 4 months
o 3 staff members for 2 months (training)
o Equipment costs: approx. $5,000 (PC, software, terminal)
Task 3: Staff time to operate
o 1.5 FTE per year
Establishment of centralized support will significantly change the focus of
UNAMAP to a user-oriented system. Support staff will become more familiar with
user concerns and problems and will document them in the log. These factors
will increase the probability of correcting the problems in future models. The
improved accessibility and ease of use of the support system will enhance the
user's perception and use of UNAMAP.
4.7 Recommendation 7; Establish a Data Clearinghouse
Battelle recommends that EPA work with all data providers to facilitate the
location and procurement of data by users. Dispersion models are useless
without the data needed to simulate the meteorological conditions prevailing at
or near the site in question. Providing a central source of information to
assist in locating and procuring the data would bring improvement to one of the
most difficult activities in the modeling process.
Facilitating access to data will advance the ease of use objective of the new
UNAMAP. Providing data and models from a single source will also improve the
view of UNAMAP's position as a single, unified source of information. Full
consideration of the organizational and legal ramifications of this
IV-44
-------
recommendation are beyond the scope of this study. Such issues will need
resolution should EPA pursue the approach described below.
The electronic bulletin board could provide the medium to make the data
clearinghouse a functional entity. Orders for data could be placed through the
bulletin board to the data provider. After payment procedures are completed,
the data would be downloaded through the system.
Data from the various sources may be handled differently in terms of ordering,
payment and delivery. NWS data from NCDC could have the orders placed
electronically. NCDC could then access the bulletin board, download the orders
into their own system, and upload the data to the bulletin board or mail it out
to the modeler. OAQPS provides some NWS data which has been preprocessed.
This data is available on the EPA UNIVAC (the data will be moved to the IBM
with rest of the UNAMAP system). A list of this data would be on the bulletin
board. Users could place their order and have the data downloaded through the
system. This would require a translation of the data to ASCII coded
characters. Onsite data would be handled differently. Persons with data could
describe their data on the system. Users would contact the owner of the data,
either via the bulletin board or by some other means. Buyers and owners would
make purchase arrangements on an individual basis. The data could be mailed or
sent through the bulletin board.
4.7.1 Implementation Tasks
Task 1: Collect information from data providers
o Get lists of data available, formats, locations, costs (if any)
Task 2: Set up bulletin board
A separate bulletin board area would have to be established to provide
the data clearinghouse services.
o Develop bulletin board specifications for each different data
source
o Establish the new bulletin board area
IV-45
-------
o Load data (if applicable)
o Test the new bulletin board and data files
Task 3: Publicize the service to users
o Prepare initial newsletter
o Produce user documentation for new bulletin board area
4.7.2 Estimated Costs and Time
This recommendation is estimated as an internal EPA task.
Task 1: Staff time
1 person for 3 months
Task 2: Staff time estimates for a single source are:
o Develop specifications = 16 hours
o Establish bulletin board area = 16 hours
o Load data files = 24 hours
o Test the bulletin board and data = 8 hours
o Estimated time: 15 calendar days
Task 3: Staff time
o Initial newsletter = 32 hours
o Bulletin board documentation = 40 hours
o Estimated time: 30 calendar days
4.8 Recommendation 8: Develop Specialized Models
Battelle recommends that EPA provide specialized models as part of the UNAMAP
library to help the modeling community fulfill all aspects of their obligation
to protect the environment. Specialized models will be designed for use in
situations which are handled poorly or not at all by the current UNAMAP models,
These models may need different modeling algorithms or new functions such as
human exposure or health risk assessment.
IV-46
-------
UNAMAP has become the standard vehicle for dispersion modeling software, and
the modeling community expects UNAMAP to continue to fulfill all their modeling
requirements. UNAMAP1s credibility, accessibility, and support system make it
the obvious choice for the role of single-source modeling facility. Also,
EOB's past record in responding to modeling needs has fostered the expectation
that they will continue to provide new modeling services.
We have identified several types of models which are needed to meet current
application requirements. These models are to handle such specialized
situations as complex terrain, toxic gases, and offshore sites.
Development of the specialized models will involve both new model development
and evaluation of existing models. The development of these models will
require advances in modeling technology such as more intricate equations to
adequately represent the complex situations being modeled.
The tasks required to develop these models are familiar to EPA. Some of the
specialized models such as complex terrain are already under development by
EPA. Others, such as offshore models, are within the purview of the EOB but
not currently under development. Finally, toxic gas models are not currently
in the scope of EOB work. If provided, they would need to be acquired.
4.9 Recommendation 9: Support the Collection of Additional
and More Accurate Meteorological Data
Battelle recommends that EPA work closely with NWS, other government agencies,
and industry to promote the collection and dissemination of meteorological data
which will facilitate the advancement of dispersion modeling technology. Data
may be the limiting factor in improving the accuracy of models in UNAMAP.
Improvement in the data available will contribute to the goal of advancing
research in modeling. Also, as data contributes to improved accuracy, it will
increase the credibility of modeling to industry and the public.
IV-47
-------
4.9.1 Implementation Tasks
There are no specific quantifiable activities which will guarantee that
improved meteorological data is captured. Rather, this recommendation is for
EPA to press for data improvements through publicity, research, interagency
agreements and regulatory changes. Some examples follow.
By increasing awareness throughout the air quality modeling community, EPA
could increase public support for data collection improvements. Specific
activities could include:
o Explanation of other countries' data collection techniques in talks and
papers;
o Emphasis of improvements in model results when improved data is used;
o Analysis of cost-effectiveness (i.e., what economic benefits are
derived by collection and utilization of improved data);
o Making modelers aware of availability of the best data and its
utilization (an application for the data clearinghouse in Section 4.7).
A proportion of ASRL research could focus on the impact of data on model
results. Some specific comparisons could include:
o Running the same model with on-site vs. NWS data, or with other
specifications of atmospheric stability instead of current estimate;
o Comparing results of a model which utilizes improved data with a model
which does not.
Increased cooperation with the government agencies responsible for
meteorological data collection may eventually result in better data being
available to UNAMAP modelers. In addition to NWS, approaches to NRCS DOE. and
OoD may result in improved modeling data, at least for certain locations.
The Meteorology Division should continue to work closely with OAQPS to see that
the best possible data is applied to modeling in the regulatory environment.
IV-48
-------
This may take the form of additions to the guideline models, or the allowance
of modifications to certain models.
4.9.2 Estimated Costs and Time
The initiatives described above do not have a directly quantifiable cost or
level of effort. They represent a certain direction in which to focus both
research and management communication resources. Through our work on the
UNAMAP evaluation, Battelle is aware that the resources of the Meteorology
Division are extremely limited. Our interviews and questionnaire responses
have highlighted the importance of this area, however, and we recommend that
EPA implement some program to improve the data available for air quality
modeling.
4.10 Recommendation 10; Support the Electronic Transfer
of UNAMAP Documentation
Battelle recommends that EPA make UNAMAP documentation available to modelers
through electronic file transfer. This would greatly speed the dissemination
of documentation, both in the original form and for updates. The electronic
documentation could be supported on the bulletin board system (See Section
4.1). Users could download a copy of any UNAMAP model's documentation whenever
needed. Then, paper output and reproduction would be under the control of the
user. Updates could be incorporated into the user's copy as needed.
This improvement directly promotes ease of use, and increases the extent of
UNAMAP support offered through a central location. Indirectly, the use of non-
regulatory models is facilitated through increased awareness and accessibility
of their documentation.
4.10.1 Implementation Tasks
The key to the successful implementation of this recommendation will be the
extent to which users can produce paper output of both text and graphics in a
usable format. The challenge is to support the most common word processing and
IV-49
-------
graphics systems that users may have. For this reason, the major effort
associated with this improvement will be to develop the format specifications
and any format conversion programs which may be required.
There are two alternative approaches. The first is to support one standard
word processing format and one standard graphics package, most likely PC-based.
Users would be required to acquire the standard packages. After downloading,
they would use the standard packages' output drivers to produce paper copies on
any printer or plotter supported by the packages. The other approach is to
develop word processing and graphics "translator" software, and have it
accessible to users. This software would "translate" the text and graphics
documentation into one of several formats before downloading to the user.
Battelle recommends the former approach. Although some users would be required
to buy additional PC software, this approach is much more affordable to EPA.
Users would gain a valuable capability for a moderate (under $1000) price.
Task 1: Select standard graphics and word processing packages
o Evaluate capabilities of existing packages
o Determine current distribution in UNAMAP community
o Make Selection
Task 2: Publicize standards to users
o Distribute newsletter to users
o Put on bulletin board
4.10.2 Estimated Costs and Time
By using the approach which requires standard packages, the cost of this
recommendation is minimized. If this recommendation is implemented in concert
with the documentation changes in Section 4*2, the UNAMAP documentation will be
revised and will then exist in some electronic form. This could be the
standard format for electronic transfer. The costs attributable to this
IV-50
-------
recommendation are then limited to selecting and publicizing the standard
packages.
Task 1:
Task 2:
Labor for selection
100 hours at $60 = $6,000
Copy of selected packages
Word processing: approx. $500
Graphics: approx. $500
Estimated time: 60 calendar days
Labor for publicity (newsletter and bulletin board)
40 hours at $60 = $2,400
Estimated time: 20 calendar days
IV-51
-------
5.0 SCHEDULE
Battelle proposes a phased approach to the implementation of the
recommendations. The implementation schedule has three phases:
Phase I:
o
o
o
o
o
Establish an electronic bulletin board
Provide UNAMAP code for multiple computers
Produce end-user documentation
Develop a set of user interfaces
Support electronic transfer of documentation
Phase 2:
o
0
0
Improve accuracy/technology of UNAMAP models
Provide centralized support services
Provide specialized models
Phase 3;
o
o
Set up a data clearinghouse
Support collection of meteorological data
The recommendations were assigned to specific phases based upon certain
criteria. The first criterion is the breadth of impact upon UNAMAP users.
Phase I activities are those which will have the most influence on the largest
number of users.
The second criterion is the time and cost needed to implement the
recommendations. These recommendations require little long-term planning or
coordination outside the Meteorology Division. Therefore, the start-up time
will be short for these activities, and actual work can begin quickly. Also,
time and costs can be estimated with accuracy for these items. The preliminary
tasks for Phases 2 and 3 have a long lead time. They need to be started during
Phase 1 in order to facilitate their implementation in later phases.
IV-52
-------
The third criterion is ease of implementation. The recommendations in Phase 1
depend least upon the cooperation of sources outside of E08. The actions
needed to implement these improvements will be most easily handled since they
are dependent upon a single entity and do not have to deal with the
organizational issues inherent in multi-agency efforts.
The priority of implementation for the recommendations in Phase 1 is based on
interdependence of the activities. The first priority is the electronic
bulletin board. The bulletin board is an important mechanism used by many of
the other recommendations. Providing UNAMAP for multiple computers is the
second priority. Decisions made in the development of the new "universal" code
are needed for later improvements. For example, ASCII disk-based input file
specifications are part of the system requirements for the user interfaces.
Production of end-user documentation is the fourth priority because
documentation can only be completed after the previous activities. It will be
possible to include the new services in the documentation rather than having to
rewrite the manuals to include these services. Electronic transfer of the
documentation will follow since the specifications needed for the transfer
process cannot be written until the documentation is available.
5.1 Detailed Schedule for Phase I Tasks
The schedule for Phase 1, summarized below, makes assumptions based on staff
resources at EOB. It is assumed that each task will be accomplished by no more
than one professional staff member. Unless otherwise indicated, the time
stated is based on one FTE staff member.
Priority 1: Establish an Electronic Bulletin Board:
Tasks 1 & 2 - Software evaluation and equipment procurement - can run
concurrently
Task 3 - Implementation - cannot start until the other tasks are complete
Task 1: Start: Week 1 End: Week 9
Task 2: Start: Week 1 End: Week 9
Task 3: Start: Week 9 End: Week 13
IV-53
-------
Priority 2: Provide UNAMAP Code for Multiple Computers
(See Section 4.3):
This schedule assumes two programmer/analysts. It also assumes that Task 1
is 80% of the time estimated in the first paragraph of Section 4.3o2.
Task 1: Start: Week 1 End: Week 24
Task 2: Start: Week 24 End: Week 28
Task 3: Start: Week 1 End: Week 28
Priority 3: Develop a Set of User Interfaces
(See Section 4.5):
This schedule is based on the assumption that one person will work on each
tasks but that tasks will overlap.
Task
Task
Task
Task
1:
2:
3:
4:
Start:
Start:
Start:
Start:
Week
Week
Week
Week
28
36
41
52
End:
End:
End:
End:
Week
Week
Week
Week
40
62
51
64
Priority 4: Produce End-User Documentation:
The schedule assumes that model documentation would be revised in the same
sequence as interfaces are developed. Task 1 in this schedule is the first
12 models (six simple and six complex) as described in Section 4.5. Task 2
is the remaining 12 models. The schedule assumes that two persons will work
on each task.
Task Is Start: Week 63 End: Week 121
Task 2: Start: Week 65 End: Week 100
IV-54
-------
Priority 5: Support Electronic Transfer of Documentation
(See Section 4.10):
The schedule for this task is split, for the reason that the standard
packages need to be determined before the documentation upgrades are begun,
but the publicity of electronic transfer capability should be deferred until
documentation is forthcoming. Note that the schedule for Task 2 is designed
for incremental availability of electronic documentation. Task 2 should be
complete when new documentation is available for the first model, i.e., ten
weeks after the new documentation task is begun.
Task 1: Start: Week 54 End: Week 62
Task 2: Start: Week 73 End: Week 76
5.2 Budget and Schedule Summary for Phase I
The schedule for the Phase I recommendations is estimated to require 121 weeks,
or approximately 28 calendar months. The schedule is shown graphically in the
Gantt chart in Figure 2. The recommendations are shown in the chart in
priority order.
A summary budget for Phase I is shown in Table 1. The figure lists each
recommendation, its total estimated cost, and the estimated start time (in
weeks after beginning implementation).
IV-55
-------
I
en
to
c
-I
co
n
3-
CD
Q.
n>
-n
0
-8
-o
9 S
a «
3
T3
-s
O
tt>
o>
3
Bulletin Board
Task i
Task 2
Task 3
Multiple
Computer Code
Task 1
Task 2
Task 3
User Interfaces
Task 1
Task 2
Task 3
Task 4
Documentation
Task i
Task 2
Electronic
Transfer
Task i
Task 2
!
_j
1
M
i
i
1 1
i
i
i i
i i
i i
i i
i i
i i
i i
i i
t i
i i
i i
i i
h 1
H
i j i i j i i j III
!TI F 1 1 1 I S 1 if 1 II 1 I I 1 M S 1 II S ! 1 S I 1 1 II ! 11 1 II 1 1 1 1 1 1 1 1 1 1 I ! 1 1 | | 1 1 1 1
) 10 20 30 40 50 60 70 80 90 100 110 120
Weeks After Implementation Begins
-------
Recommendation Priority No. Estimated Cost Start Time
Electronic Bulletin 1 $ 9,680 Week 1
Board
Models for Multiple 2 $ 90,040 Week 1
Computers
Develop User Interfaces 3 $ 99.000 Week 28
Distribute Documentation 4 $ 9,400 Week 54
Electronically
Develop End-User 5 $561,000* Week 63
Documentation
TOTAL $769,120
* This amount is totally additive. No provision has been made for efficiency
gained by learning.
Figure 4. Estimated Cost and Start Time for Each
Phase I Recommendation Will Provide Timing of Funds Needed for Budget
IV-57
-------
6.0 SUMMARY
The results of the Evaluation and Assessment of UNAMAP project will facilitate
EPA's efforts to improve UNAMAP. This report is the last of four major parts
of Battelle's project.
The report described a plan for implementing a series of recommended
improvements to the UNAMAP program. The plan consists of a strategy to attain
EPA's long-range objectives for UNAMAP, the specific improvements which are
recommended by Battelle, and the schedule and budget to implement the
recommendations.
The conclusions reached in this report were based on research and analysis
which were performed in three earlier phases of the project. The first was a
technology assessment to determine technology available to UNAMAP. The second
was data collection (by interview and questionnaire) and analysis. The third
was formulation of recommendations. These three phases were summarized in this
report to give the reader the necessary background information.
The ASRL Meteorology Division's major objectives for UNAMAP are to advance
research in dispersion modeling and to transfer the technology to the public.
Because these two objectives are interdependent, Battelle's strategy for
improving UNAMAP considered both. Changes have occurred in the UNAMAP user
community, and in the UNAMAP software library itself since the UNAMAP was
begun. Battelle's research results indicate that these changes have resulted
in a reduced effectiveness of the UNAMAP program in meeting its two objectives.
The report described the implementation of specific recommended improvements
which will promote the increased effectiveness of UNAMAP. The recommendations
were designed to support the long-range objectives of the UNAMAP program and
remedy the underlying causes of current problems. Achievement of the strategic
objectives will help UNAMAP continue to fulfill its leadership role in air
quality modeling,
IV-58
-------
Based on interdependence of recommendation and on breadth of benefit, Battelle
divided recommendations into three phases for implementation. The
recommendations in each of the three phases are:
Phase I:
o Establish an electronic bulletin board
o Provide UNAMAP code for multiple computers
o Produce end-user documentation
o Develop a set of user interfaces
o Support electronic transfer of documentation
Phase 2:
o Improve accuracy/technology of UNAMAP models
o Provide centralized support services
o Provide specialized models
Phase 3:
o
o
Set up a data clearinghouse
Support collection of meteorological data
The schedule developed indicates that the Phase 1 recommendations can be
implemented, with a relatively low level of effort, within 26 months. After
that time, UNAMAP will have advanced a great distance toward becoming the air
quality modeling service with both well-researched models and a strong user
orientation.
IV-59
-------
APPENDIX A:
UNAMAP USER DOCUMENTATION
RECOMMENDED FORMAT
60
-------
The following annotated outline identifies a general plan for the major
sections of the proposed revised UNAMAP documentation.
UNAMAP Documentation
XYZ Model
Section Title
Note to Readers
Description
Includes an identification of the various types of
readers/users and presents a road map to the
various sections.
Overview
Tutorial
Purpose: To show each reader what parts of
the documentation will be useful to them,
depending on their unique needs.
Presents background on the model including any
key terms and concepts; features and
limitations; and a brief, but complete
technical description which identifies
assumptions used in the model.
Purpose: To provide all readers with a
context for applying the model.
Provides a step-by-step introduction to the
actual use of the model. Steps the user
through several examples. Identifies
techniques used in the model, such as menu
options, prompts, commands, etc. Introduces
any major points related to use of the model.
Where appropriate, relates to reference
material in the subsequent sections.
IV-61
-------
Reference/Job Aids
Purpose: To provide a first-use experience to
all types of users.
Includes reference material such as menu
illustrations, diagrams, and option
descriptions; prompts and
acceptable/unacceptable responses; data field
limitations and restrictions; data collection
and data entry job aids; and other checklists.
Also includes sample model output with
explanation to assist users in reading output.
Purpose: To provide functional reference aids
to assist all types of users in the actual use
of the model.
Scientific Notes
Additional reference material of a scientific
and/or meteorological nature, such as scientific
modeling concepts, sensitivity analyses, or point
source computations.
Purpose: To satisfy additional informational
requirements of dispersion meteorologists and
others interested in the scientific
foundations of the model.
Program Notes
Additional reference material of a computer
programming nature, such as the program
overview and structure, indexed listing of the
FORTRAN source statements, and other
programming details.
Purpose: To satisfy additional informational
requirements of data processing personnel and
IV-62
-------
others interested in the computer programming
aspects of the model.
Glossary Brief but complete definitions of technical
vocabulary used in the model, the
documentation, or the output.
Purpose: To provide ready reference on
technical terms users might encounter in
working with the model.
To further illustrate this design, the following diagram shows how the current
table of contents for the PTPLU model documentation would be reorganized around
this new design concept.
IV-63
-------
PTPLU Model
Comparison of Current Table of Contents and
Proposed Revised Organization
Notes Items that appear in bold in the right column indicate sections of
content that already exist in the current documentation* In some cases, these
will require some reformatting and new presentations. Other items listed in
the right column are new material.
IV-64
-------
Current Documentation
Proposed Revised Organization
Foreword
Preface
Abstract
Figures
Symbols and Abbreviations
Acknowledgments
1. Introduction
2. Data Requirements Checklist
3. Features and Limitations
4. Technical Description
5. Program Overview and
Structure
6. Input Data Preparation
7. Execution of the Model and
Sample Test
8. Example Calculation
References
Appendices
A. Modeling Concepts
B. Indexed Listing of
FORTRAN
Source Statements
C. Sensitivity Analysis
Glossary
Foreword
Note to Readers
abstract
preface
Table of Contents, List of
Figures, List of Tables
Overview
introduction
terms and concepts
features and limitations
technical description
Tutorial
tutorial demonstration
execution of the model and
sample test
example calculation
Reference/Job Aids
data requirements checklist
input data preparation
symbols and abbreviations
menu organization and use
prompts and responses
sample output with explanation
Scientific Notes
modeling concepts
sensitivity analysis
Program Notes
program overview and structure
indexed listing of FORTRAN
source statements
Glossary
glossary
IV-65
------- |