oEPA
United States
Environmental Protection
Agency
Environmental Sciences Research
Laboratory
Research Triangle Park NC 277 11
EPA-600, 9-79-041
November 1 979
Research and Development
Workshop on
Atmospheric
Dispersion
Models in
Complex Terrain
-------
RESEARCH REPORTING SERIES
Research reports of the Office of Research and Development, U.S. Environmental
Protection Agency, have been grouped into nine series. These nine broad cate-
gories were established to facilitate further development and application of en-
vironmental technology Elimination of traditional grouping was consciously
planned to foster technology transfer and a maximum interface in related fields.
The nine series are-
1. Environmental Health Effects Research
2 Environmental Protection Technology
3 Ecological Research
4 Environmental Monitoring
5. Socioeconomic Environmental Studies
6 Scientific and Technical Assessment Reports (STAR)
7 Interagency Energy-Environment Research and Development
8. "Special" Reports
9. Miscellaneous Reports
This document is available to the public through the National Technical Informa-
tion Service, Springfield, Virginia 22161.
-------
EPA-600/9-79-041
November 1979
WORKSHOP ON ATMOSPHERIC DISPERSION
MODELS IN COMPLEX TERRAIN
by
Einar L. Hovind
Max W. Edelstein
Victoria C. Sutherland
North American Weather Consultants
Goleta, California 93017
Contract No.
68-02-3223
Project Officer
George C. Holzworth
Meteorology and Assessment Division
Environmental Sciences Research Laboratory
Research Triangle Park, North Carolina 27711
ENVIRONMENTAL SCIENCES RESEARCH LABORATORY
OFFICE OF RESEARCH AND DEVELOPMENT
U. S. ENVIRONMENTAL PROTECTION AGENCY
RESEARCH TRIANGLE PARK, NORTH CAROLINA 27711
-------
DISCLAIMER
This report has been reviewed by the Environmental Sciences
Research Laboratory, U. S. Environmental Protection Agency, and
approved for publication. Approval does not signify that the contents
neccessarily reflect the views and policies of the U. S. Environmental
Protection Agency, nor does mention of trade names or commercial
products constitute endorsement or recommendation for use.
11
-------
PREFACE
During the period July 16-20, 1979, an EPA sponsored Workshop was
conducted in Raleigh, North Carolina, to address problems associated
with plume dispersion modeling in complex terrain. Similar topics
have been the subject of other workshops sponsored during the past
three years, such as the 1977 AMS Workshop on Stability Classification
Schemes and Sigma Curves; the 1977 Specialists' Conference on the EPA
Modeling Guideline; and the 1976 Albuquerque Workshop on Research
Needs for Atmospheric Transport and Diffusion in Complex Terrain.
Since the Albuquerque Workshop the Clean Air Act Amendments of
1977 were signed into law. These amendments have placed a heavy
burden upon regulatory agencies and industry in that they call for
regulations which specify the use of dispersion models to achieve air
quality limits pertinent to prevention of significant deterioration
and attainment of National Air Quality Standards. The problem
is particularly critical regarding dispersion in complex terrain
since EPA, so far, has not specified models with demonstrated
high degrees of reliability for use in refined modeling analysis in
such settings. Furthermore, the need for these models continues to
grow with the increasing demands for energy resource development in
mountainous areas.
In response to this dilemma, EPA authorized the organization
of this Workshop through the Meteorology and Assessment Division
(MD) of the Environmental Sciences Research Laboratory (ESRL) at
Research Triangle Park, North Carolina. The planning and coordina-
tion of the Workshop activities, and the preparation of this report,
have been carried out by North American Weather Consultants (NAWC)
under contract to EPA, with significant assistance from the Meteor-
ology and Assessment Division.
The Workshop was organized into five panels on the following
topics: Model Development and Analysis; Model Evaluation and Applica-
tion; Experimental Design; Measurement Techniques; Data Management and
Quality Assurance. The number of Workshop participants totaled 59.
Of these, 47 were invited specifically to serve on the Workshop
111
-------
panels. They represented a cross-section of environmental organiza-
tions, control agencies, industry and the scientific community with
technical background and expertise in complex terrain modeling
and field studies. The other attendees were invited speakers,
guests, and personnel from the Contractor and EPA contracting
offices.
The Workshop agenda included four days of alternating plenary
and work sessions. The opening plenary session began with presenta-
tions by invited speakers who presented summaries of related complex
terrain dispersion programs currently being sponsored by industry
and by government agencies other than EPA. This plenary session
also included opening statements by EPA on their anticipated future
programs and presentations by each Panel Leader on the goals of the
Workshop and the tasks that the panel would attempt to accomplish
over the next four days.
At the end of the Workshop these accomplished tasks were pre-
sented orally during the last plenary session in the form of recom-
mendations by each panel how EPA should best proceed with the complex
terrain program. The Panel Leaders handled the tasks of assembling
the written statements from their respective panels and presented
edited material to the contractor for publication following review by
the panel members. These statements are published, unabridged, in
this report together with summaries of stimulating discussions
which took place during the other plenary sessions.
IV
-------
ABSTRACT
A Workshop, sponsored by the Environmental Protection Agency's
Environmental Sciences Research Laboratory (EPA/ESRL), Meteorology
and Assessment Division (MD), was held in Raleigh, North Carolina,
during July 16-20, 1979. The goals of this conference were to focus
on complex terrain modeling problems, and to develop recommendations
to EPA with respect to design of a workable multi-year program to
start in Fiscal Year 1980. Personnel from the Meteorology and
Assessment Division (MD) of EPA provided "strawman" material for
discussion at the conference. Based on this material, the conferees
agreed that considerable benefit would be derived by focusing the
conference on impingement problems.
Invited experts on various aspects of field studies, model
development, model evaluation, and data management prepared recom-
mendations and technical support documents on these topics and re-
lated budgetary allocations. The participants addressed in con-
siderable detail questions on how to collect a representative data
base to develop credible models for regulatory applications that would
adequately handle plume impingement problems in elevated terrain.
The Workshop participants agreed in principle that EPA should
adopt a two-phased field program approach, starting with a controlled
experiment on a small, isolated hill of simple geometric setting, and
then proceeding with a large scale program of increased complexity.
It was also recommended that the modeling development program follow
multiple development phases. The initial effort should be oriented
towards improvements in Gaussian-based models, while the final effort
should be aimed towards new model development incorporating complex
flow fields in rough terrain with either Gaussian-based or "K theory"
based models. Physical modeling programs should be an integral part
of the above efforts.
The Workshop recommendations address the need for improved
model evaluation techniques, statistical analysis methods, sensi-
tivity analysis, comparative field data analysis, and independent
evaluation techniques. Furthermore, informative recommendations
were provided on problems concerning data management and quality
assurance, as well as on specific costs relative to experimental
design with suggested budgetary allocations for future programs.
v
-------
CONTENTS
Preface iii
Abstract v
Figures viii
Tables ix
Workshop Participants x
Acknowledgements xviii
1. Introduction 1
2. Conclusions 3
3. Workshop Organization 8
4. Plenary Sessions 11
5. Panel Recommendations 25
Model development and analysis panel ... 26
Model evaluation and application panel . . 62
Experimental design panel 73
Measurement techniques panel 92
Data management and quality assurance
panel 115
Appendices
A. Pre-Workshop material prepared by EPA 136
B. Pre-Workshop material prepared by NAWC .... 174
C. Post-Workshop comments by panel members .... 184
VII
-------
FIGURES
Number Page
1 EPA Workshop schedule 10
2 Diagram of flow of information within program .... 30
3 Suggested array of samplers for 100 m hill
experiment 34
4 Perspective sketch of field program 57
5 Estimation of ambient plume behavior and the
degree of plume impaction upon elevated
terrain 75
6 Schematic depiction of near terrain profiles
of tracer concentration 76
7 Conceptual depiction of characteristic zones
of tracer plume transport and diffusion 84
8 Experimental arrangement using laser
illuminators 102
9 Downlooking airborne lidar showing plume
contact with ground 110
10 Interaction between quality assurance and
data management 116
11 Example of outlier detection 130
A-l Diagram showing flow pattern of stack
plume - nighttime 157
A-2 Diagram showing flow pattern of stack
plume - daytime 157
A-3 Experiment for determinging p(C|P, M) 164
A-4 Experimental design for studying the micro-
mesometeorology of complex terrain 169
B-l Flow diagram for Workshop ! 179
viii
-------
FIGURES (Cont'd)
Number Page
B-2 Flow diagram for specific complex terrain
model study recommendations 180
B-3 Suggested schedule for Workshop as detailed
by Panel Leaders in their letters to Panel
Members 181
TABLES
Number Page
1 Summary of measurement techniques and their
probable applications to experimental scales ... 94
2 Single-sensor error checks 129
IX
-------
WORKSHOP PARTICIPANTS
1. INVITED PARTICIPANTS AND GUESTS
Dr. A. Paul Altshuller
Environmental Sciences Research Laboratory (MD-58)
Environmental Protection Agency
Research Triangle Park, NC 27711
Dr. David S. Ballantine, DOE
Office of Health and Environmental Research
Mail Stop E 201 (Germantown)
Washington, DC 20545
Dr. Norman E. Bowne
The Research Corporation of New England
125 Silas Deane Highway
Wetherfield, CN 06109
Mr. Edward Burt
Office of Air Quality Planning and Standards (MD-14)
Environmental Protection Agency
Research Triangle Park, NC 27711
Dr. C. Shepherd Burton
Systems Application, Inc.
950 Northgate Drive
San Rafael, CA 94903
Dr. Jack Cermak
Fluid Dynamics & Diffusion Laboratory
Colorado State University
Ft. Collins, CO 80523
Mr. John Clark
Meteorology and Assessment Division (MD-80)
Environmental Protection Agency
Research Triangle Park, NC 27711
-------
Mr. Jesse H. Coleman
TVA-Air Quality Branch
E and D Building
Muscle Shoals, AL 35660
Dr. Harry E. Cramer
H. E. Cramer, Inc.
540 Arapeen Drive
Salt Lake City, UT 84108
Mr. Loren W. Crow
Consulting Meteorologist
2422 South Downing Street
Denver, CO 80210
Mr. Gerard A. DeMarrais
Meteorology and Assessment Division (MD-80)
Environmental Protection Agency
Research Triangle Park, NC 27711
Dr. Vernon Derr
NOAA-Environmental Research Laboratories
Wave Propagation Lab
Boulder, CO 80303
Dr. Marvin Dickerson, L-262
Lawrence Livermore Laboratory
Post Office Box 808
Livermore, CA 94550
Mr. John Eckert
EPA Monitoring and Support Lab
Post Office Box 15027
Las Vegas, NV 89114
Dr. Bruce A. Egan
Environmental Research and Technology, Inc.
3 Militia Drive
Lexington, MA 02173
Mr. Roy Evans
EPA-Environmental and Support Laboratory
Post Office Box 15027
Las Vegas, NV 89114
XI
-------
Mr. Gilbert Ferber
NOAA-Air Resources Laboratory
8060 13th Street
Silver Spring, MD 20910
Dr. Douglas G. Fox
U. S. Forest Service
240 West Prospect Street
Fort Collins, CO 80521
Dr. Dan Golorab
ORD-Energy Effects Division (RD-682)
Environmental Protection Agency
Washington DC 20460
Dr. Philip Gresho, L-262
Lawrence Livermore Laboratory
Post Office Box 808
Livermore, CA 94550
Dr. Freeman Hall
NOAA-Environmental Research Laboratories (R45X9)
Wave Propagation Lab
Boulder, CO 80303
Mr. Donald Henderson - AIR
National Park Service
Post Office Box 25287
Denver, CO 80225
Dr. Glenn R. Hilst
Electric Power Research Institute
3412 Hillview Avenue
Post Office Box 10412
Palo Alto, CA 94303
Mr. Norman Huey
EPA Region VIII-Air Branch
1860 Lincoln Street
Denver, CO 80295
Dr. Julian C.R. Hunt
Dept. of Applied Mathematics & Theor. Physics
University of Cambridge
Cambridge, England CBS 9EW
xii
-------
Mr. John Irwin
Meteorology and Assessment Division (MD-80)
Environmental Protection Agency
Research Triangle Park, NC 27711
Dr. Warren Johnson
SRI International
333 Ravenswood Avenue
Menlo Park, CA 94025
Mr. Robert C. Koch
Geomet, Inc.
15 Firstfield Road
Gaithersburg, MD 207GO
Dr. Robert G. Lamb
Meteorology and Assessment Division (MD-80)
Environmental Protection Agency
Research Triangle Park, NC 27711
Dr. Ron B. Lantz
Intera
11511 Katy Freeway, Suite 630
Houston, TX 77079
Mr. Thomas F. Lavery
Environmental Research and Technology, Inc.
2625 Townsgate Road
Westlake Village, CA 91360
Mr. Vince Mirabel la
Southern California Edison Company
Post Office Box 800
Rosemead, CA 91770
Mr. Donald W. Moon
Salt River Project
Post Office Box 1980
Phoenix, AZ 85001
Ms. Joan H. Novak
Meteorology and Assessment Division (MD-80)
Environmental Protection Agency
Research Triangle Park, NC 27711
Xlll
-------
Dr. Robert Papetti
ORD-Air Quality Staff (RD-682)
Environmental Protection Agency
Washington DC 20460
Mr. William Petersen
Meteorology and Assessment Division (MD-80)
Environmental Protection Agency
Research Triangle Park, NC 27711
Dr. Francis Pooler Jr.
Regional Field Studies Office (MD-84)
Environmental Protection Agency
Research Triangle Park, NC 27711
Dr. Rudy Pueschel
NOAA-Environmental Research Laboratories
Atmospheric Physics and Chemistry Laboratory
Boulder, CO 80303
Mr. Alvin Rickers
Utah State Division of Environmental Health
Salt Lake City, UT 84110
Dr. Courtney Riordan
ORD-Office of Air, Land, and Water Use (RD-682)
Environmental Protection Agency
Washington DC 20460
Dr. Ronald E. Ruff
Atmospheric Science Laboratory
SRI International
333 Ravenswood Avenue
Menlo Park, CA 94025
Mr. Frank Schiermeier
Meteorology and Assessment Division (MD-80)
Environmental Protection Agency
Research Triangle Park, NC 27711
Mr. Maynard E. Smith
Meteorological Evaluation Services
134 Broadway
Amityville, NY 11701
xiv
-------
Dr. Ted Smith
Meteorology Research, Inc.
Post Office Box 637
Altadena, CA 91001
Dr. William H. Snyder
Fluid Modeling Section (MD-81)
Environmental Protection Agency
Research Triangle Park, NC 27711
Mr. Timothy C. Spangler
North American Weather Consultants
2875 S. Main Street
Salt Lake City, UT 84115
Mr. Gene Start
NOAA-Environmental Research Laboratories
550 Second Street
Idaho Falls, ID 83410
Dr. Ivar Tombach
AeroVironment Inc.
145 North Vista Avenue
Pasadena, CA 91107
Mr. James B. Tommerdahl
Research Triangle Institute
Post Office Box 12194
Research Triangle Park, NC 27709
Dr. Jeffery Weil
Martin-Marietta Laboratories
1450 South Rolling Road
Baltimore, MD 21227
Dr. Michael D. Williams
John Muir Institute
Route 5, Box 229A
Santa Fe, NM 87501
Mr. Dean Wilson
Office of Air Quality Planning and Standards (MD-14)
Environmental Protection Agency
Research Triangle Park, NC 27711
xv
-------
Mr. Robert Wilson
EPA Region X (MS-329)
1200 Sixth Avenue
Seattle, WA 98101
Dr. William E. Wilson
Regional Field Studies Office (MD-84)
Environmental Protection Agency
Research Triangle Park, NC 27711
2. NON-ATTENDEE CONTRIBUTORS
Dr. William R. Goodwin &
Dr. Paul D. Gutfreund
Dames and Moore
1100 Glendon Avenue
Los Angeles, CA 90024
Dr. Don Grey
Utah Power and Light Company
Post Office Box 899
Salt Lake City, UT 84110
Dr. Gary M. Klauber
Environmental Measurements, Inc.
1445 Old Annapolis Road
Arnold, MD 21012
Mr. Thomas J. Lockhart
Meteorology Research, Inc.
Post Office Box 637
Altadena, CA 91001
Dr. John E. Pinkerton
National Council of the Paper Industry
for Air and Stream Improvement, Inc.
260 Madison Avenue
New York, NY 10016
Mr. Robert N. Swanson
Pacific Gas and Electric Company
77 Beale Street
San Francisco, CA 94106
xvi
-------
3. EPA PROJECT OFFICE STAFF
Meteorology and Assessment Division (MD-80)
Environmental Protection Agency
Research Triangle Park, NC 27711
Mr. Lawrence E. Niemeyer
Mr. George C. Holzworth
Mr. Gerard A. DeMarrais
4. NAWC CONTRACTOR STAFF
Mr. Einar L. Hovind
Mr. Max W. Edelstein
North American Weather Consultants
600 Norman Firestone Road
Goleta, CA 93017
Ms. Victoria C. Sutherland
North American Weather Consultants
2875 South Main Street
Salt Lake City, UT 84115
xvi i
-------
ACKNOWLEDGMENTS
More than 50 individuals were involved in the Workshop pro-
ceedings presented in this report. The contributions from this
group, and in particular the unstinting efforts of the panel members
can best be described as exemplary. Several participants in the
Workshop deserve special acknowledgement, as indicated below.
Advance planning, assistance in selection of panel members,
and preparation of "strawman" pre-conference documents by Mr.
Lawrence E. Niemeyer, Mr. George C. Holzworth, Dr. William H.
Snyder, and Dr. Robert G. Lamb of the Meteorology and Assessment
Division of EPA at Research Triangle Park, are acknowledged with
gratitude.
The five Panel Leaders, under whose guidance and motivation
the Panels functioned, were notably effective in synthesizing large
quantities of complex technical material into coherent panel reports.
These Panel Leaders are: Dr. Bruce A. Egan (ERT), Mr. Roy Evans
(EPA), Dr. Ronald E. Ruff (SRI International), Mr. Maynard E. Smith
(MES), and Mr. Gene Start (NOAA-ERL).
Valuable background information was provided by special pre-
sentations made by invited speakers who summarized the related
complex terrain programs currently sponsored by industry and by
government agencies other than EPA. These speakers were: Dr.
Glenn R. Hilst, Electric Power Research Institute; Dr. David S.
Ballantine, Department of Energy; and Dr. Vernon Derr, National
Oceanic and Atmospheric Administration.
XV 111
-------
SECTION 1
INTRODUCTION
At the request of EPA a Workshop was held in Raleigh, North
Carolina, during July 16-20, 1979, to tackle the pressing problems
associated with the development of improved mathematical models for
predicting plume dispersion in complex terrain. Invited experts, with
broad technical and scientific backgrounds, were assigned the tasks of
generating recommendations and guidance to EPA for use in the planning
of dispersion model development and associated field experiment pro-
grams over the next 3-5 years.
The manner in which the Workshop was organized, with five panels
on topics essential to achieving the goals of the- conference, is des-
cribed in Section 3, Workshop Organization. The Workshop schedule
allowed for periodic interchange of technical information through
frequent plenary sessions.
Brief summaries of opening statements, presentations by invited
guest speakers, and highlights of the technical discussions, which
took place during these sessions, are presented in Section 4,
Plenary Sessions.
Under the able leadership and direction of five invited panel
leaders, the panels addressed a number of technical aspects of
future EPA programs, as viewed from their respective specialties.
Not only did the panels prepare these written recommendations
under a very limited time frame, but the participants also prepared
highly technical reports which the panel leaders arranged into compre-
hensive support documents to their panels' recommendations. These
statements and recommendations are presented unabridged-in Section 5,
Panel Recommendations, in the format presented to the contractor by
the Panel Leaders. The proceedings also include some excellent
"strawman" scenarios prepared by EPA and distributed prior to or at
the opening day of the Workshop. The scenarios served the stated pur-
pose of introducing the participant to some of EPA's internal thoughts
on suitable approaches to the problems at hand. They also served to
quickly focus the panels' attention on critical concepts which needed
to be examined in an open forurn before the formulation of their recom-
mendations. The "strawman" plans and other material and information
-------
distributed to the participants prior to the Workshop, are presented
in Appendices A and B.
The Workshop proceedings have also provided a forum for addition-
al comments by participants generated during their review of the final
panel recommendations subsequent to the Workshop. These letters and
comments are summarized in Appendix C.
Finally, it was recognized that the Workshop could only accomo-
date a limited number of invited participants, and that there were
many highly qualified experts in the disciplines relevant to the
Workshop topics who could not participate in person. Invitations
were distributed to over fifty experts outside those attending the
Workshop, providing an opportunity for them to submit ideas and
topics for panel considerations. Several excellent technical inputs
were received and utilized by the Workshop panels and recognition
of these contributions is acknowledged in the list of participants.
-------
SECTION 2
CONCLUSIONS
The Workshop provided a timely forum for technical discussions on
topics pertaining to plume dispersion in complex terrain. The dis-
cussions covered a wide range of subject matters dealing with field
study designs, data handling, and measurement techniques relevant to
the tasks of model development and evaluations. The EPA "strawman"
plans, which were prepared for consideration by the Workshop partici-
pants, provided an effective means of focusing the panels' attention
on specific approaches to future EPA field studies and model develop-
ment programs.
The panel recommendations and associated technical discussions
clearly demonstrate the complexity of the problems, and that there
are no simple solutions. There was general agreement on the over-
all approach regarding model development programs. This agreement
was based on acceptance by the conferees of the concepts presented
by EPA in their "strawman" documents (Appendix A) which recommended
focusing on plume impingement for the initial effort. However,
divergent viewpoints were frequently expressed by individual par-
ticipants, and at times by panel leaders, regarding the details
of the required experimental studies that would be most appropriate
within the stated program constraints.
The wealth of technical information and panel recommendations
generated by the Workshop relative to the specific goals expressed
by EPA at the opening plenary session have been examined by the
Workshop coordinators (NAWC staff), and the following seven con-
clusions were drawn.
1. THE OVERALL PLAN
All panels agreed in principle with an approach to model develop-
ment starting with a controlled experiment on a small isolated hill of
simple geometric setting. Once an initial understanding of plume im-
paction with elevated terrain is achieved, larger scale experiments
should be conducted, and should include systematic investigation of
-------
phenomena with increased complexity. General agreement was also ex-
pressed on the need for a closely coordinated model development pro-
gram, including the use of physical modeling techniques, to accompany
the field studies. The total EPA program should be designed to
facilitate coordination with other related efforts, such as the EPRI,
DOE, and NOAA programs.
A lack of concensus emerged on the question of what physical
hill size would be preferable for the small-scale program. Whereas
the modeling panels considered a 100 m hill to be a desirable size for
simple field experiments on non-buoyant tracer plume impaction, the
panels on experimental design and measurement techniques expressed,
during the plenary sessions, a preference for a medium size hill of
400-500 m. However, their written recommendations presented herein
are in the form of pro and con assessments with respect to different
field program scenarios, such as small scale (100 m hill), medium
scale (400-500 m hill), and full scale (real plant) experiments in
order to provide EPA with maximum guidance for reaching a final de-
cision on this issue. The overriding concerns in this decision-
making process must be that the scale of experiments can be extended
to "real world" situations, and they must be consistent with the
ability to measure the required parameters.
2. MODEL DEVELOPMENT APPROACH AND IMPLEMENTATION
There was general agrement on the recommended approach that the
initial model development effort should be oriented towards reasonable
"fix-up" to Gaussian-based models currently used in the regulatory
practice. The subsequent model development activities should concen-
trate on developing and using better descriptions of the flow fields
expected in rough terrain for a variety of meteorological and topo-
graphic settings. The resultant flow field descriptions could be used
with either Gaussian-based models or with "K theory" based models.
3. MODEL TESTING AND DEMONSTRATION OF CREDIBILITY
There was general agreement on several recommendations stating
the need for adequate model testing and evaluations. These procedures
should include examination of model performance based on observed
versus predicted comparisons. Emphasis was placed on the need for
having model development data to meet requirements of model evalua-
tion with adequate measurements to verify model performance. The
recommendations also included the need to define the applicability
of the result. Furthermore, the demonstration of credibility would
be greatly improved by the use of flow visualization. This technique
enhances understanding of the basic processes occurring during
-------
plume interaction with the terrain, which cannot be readily described
from ground measurements of tracer concentrations alone.
4. MODEL APPLICATION IN NEW SETTINGS
The discussion of model transferability to other sites centered
around the results to be obtained from the small scale versus large
scale field programs. It was agreed that the basic process of plume
impaction in simple, small scale experiments should be readily
transferable to other settings. However, the more complex physical
processes associated with large scale terrain features, such as com-
plex plume trajectories, thermally induced mountain slope flows, and
vertical wind shear, are so site specific as to make the transfer of
results to new settings very difficult. Again the flow visualization
techniques will greatly aid in the acceptance of transferability
of the model.
Specific recommendations regarding model applications include the
need for thorough documentation of the experiments, the data, and the
model development. It is essential that, to assure proper use of the
models, the documentation must include a list of limitations, special
conditions, and confidence bands for future model applications.
5. THE SCIENTIFIC FEASIBILITIES OF PROPOSED FIELD AND LABORATORY
STUDIES
The discussions on small versus medium and full scale ex-
periments raised many questions on similarities in plume behavior.
Doubts were expressed by some whether the surface boundary layer
and vertical atmospheric structure associated with the small scale
study would be applicable to "the real world". Since turbulent
motion is very important in the plume impaction processes with
elevated terrain, questions were also raised whether these processes
can be simulated on a scale significantly different from that found in
large scale complex settings. Other questions were concerned with
plume impaction; whether it is dominated by surface induced turbulence
or by turbulence at plume elevation. If the former persists, the size
of the hill doesn't matter, but if the latter dominates, then the
small hill experiment might be inappropriate.
The technical information and discussions generated by each
panel, and presented in these proceedings are highly illuminating
since they address these questions from different technical and scien-
tific viewpoints. Recommendations were made that field programs
should allow for testing the effects of scale, and that the impaction
problem be studied- as a function of terrain and meteorology. There
-------
was general agreement on the recommendation that physical modeling
techniques be incorporated into the model development program. The
applications and limitations of this technique must be defined.
6. EFFICIENT MEANS OF DATA HANDLING
The primary inputs on data handling were given by the Data
Management and Quality Assurance Panel. The recommendations from
this panel provide an informative overview of major phases of
data management and information feedback as a quality control
measure. They suggest that, following implementation of a Data
Acquisition Plan, the following three actions should occur: 1) Field
operations should provide for field data through an acquisition
phase with quality assurance feedback; 2) data compilation processes
should follow field data review and bring the data into a final
check phase, again with quality assurance feedback; and 3) an
independent data review should follow after preparation and delivery
of reports to be accompanied by final quality assurance feedback.
Although field activities are distinguished from off-site activities,
this panel recommended that the same groups conduct these activities
through the final data archiving process.
7. BUDGET ALLOCATIONS
The recommendations prepared by the panels dealing with field
experimental design, instrumentation, and data handling contain a
number of cost estimates and budgetary allocations. These estimates
must be viewed relative to EPA's expected budgetary constraints of
about $2.0 million per year for three years. The Measurement
Techniques Panel presented specific cost estimates for the 100 m hill
and larger scale projects as follows: For the 100 m hill, hardware
costs alone were estimated to be near $2.6 million and the operation
of the ground-based instruments could be around $40,000-$50,000 per
month. Estimates for gaseous tracers, sampling equipment, laboratory
analysis, towers without equipment, and plume visualization equip-
ment are also included. For the larger scale experiments, hardware
costs for Doppler Acoustic Systems could be around $40,000-$50,000,
aircraft tracer measurements about $8,000 per day, lidar costs about
$50,000 per month, and airborne lidar with ground based data systems
about $4,000 per day.
The Experimental Design Panel prepared a budget estimate
for 50 days of sampling during two 40-day research periods with con-
tinuous 18 months monitoring (without centralized data logging equip-
ment) on a 400 m hill in the Western U.S. The cost estimate was
nearly $2.9 million dollars based-on direct costs and labor with a
-------
moderate contractor's charges (General and Administrative expense, and
fee or profit).
The Data Management and Quality Assurance Panel submitted cost
figures based on upper-bound requirements for two field study periods.
The two-phased program would require approximately 12-13 person years.
Personnel and support costs would consume about one million dollars,
and field data facilities about $130,000. The estimated costs were
considered consistent with the following budgetary allocation sug-
gested by the Measurement Techniques Panel during panel session dis-
cussions: management (10%); field work and initial data processing
(55%), data analysis and quality assurance (15%), and model develop-
ment (20%). These proposed allocations were not discussed in depth
during the Workshop.
-------
SECTION 3
WORKSHOP ORGANIZATION
On March 9, 1979, North American Weather Consultants (NAWC) was
authorized by the Environmental Protection Agency (EPA) to organize
and conduct a Workshop to Develop Recommendations on Atmospheric Dis-
persion Models in Complex Terrain. The contract award was originally
based on a task requiring NAWC to organize "a Workshop of invited ex-
perts on atmospheric transport and diffusion over complex terrain who
will make specific recommendations for multi-year Government research
that will significantly improve the reliability of practical mathe-
matical dispersion models over complex terrain". Furthermore, the
original plan specified two types of emission configurations:
"1) Tall chimneys such as exist at fossil-fueled power plants and
smelters and 2) assemblages of emitters relatively near ground level
such as are expected in connection with oil shale resources
development".
During the first planning meeting between the EPA contracting
office and the contractor (NAWC) in Research Triangle Park, NC, on
March 26-29, 1979, the overall Workshop effort was modified to address
specifically the problem of plume dispersion from large fossil-fuel
power plants in complex terrain. Agreement was also reached on the
Workshop structure with plans to have five panels on the following
topics: Model Development and Analysis, Model Evaluation and Applica-
tion, Experimental Design, Measurement Techniques, and Data Management
and Quality Assurance.
At the same meeting a list of potential participants was de-
veloped to include experts in applied and theoretical modeling, physi-
cal (laboratory-scale) modeling, field (full-scale) experimentation,
instrumentation, data archiving, and quality control. Consideration
was also given to obtaining a balance between the panel members' pro-
fessional background with representation from federal and state
agencies, universities, the electric power industry, and the private
consulting industry. Finally, the assignment of the 47 participants
to specific panels was made in such a manner as to provide some
mixture of expertise, where possible, on each panel.
8
-------
A pre-conference meeting, scheduled at the EPA facility in
Research Triangle Park on June 6, was attended by the five panel
leaders (or their substitutes) and the contractor. By this time,
Raleigh, NC, had been selected as the site for the Workshop, follow-
ing considerations of the cost and convenience of several other loca-
tions. The meeting covered technical apects of the Workshop, the
respective roles of the panels, and final recommendations for
the list of invited participants.
Subsequently, NAWC and EPA representatives prepared and distrib-
uted pre-Workshop material to panel leaders and members. These docu-
ments are included as part of the Workshop report. Invitations were
also sent to over 50 experts in the fields listed above who could not
be included in person at the Workshop, but who were given the oppor-
tunity to contribute construcive ideas and concerns to the panels for
their deliberations. Those who responded to this special invitation
with material for discussion are included in the list of participants.
The Workshop schedule (Figure 1) was designed to provide maximum
interaction and exchange of information between the panels through
alternating panel sessions and plenary meetings. The opening plenary
session provided the opportunity for representatives from EPRI, DOE,
NOAA, and EPA to present to the participants various programs which
are currently planned or under way by their respective agencies on
topics dealing with dispersion in complex terrain.
The recommendations prepared by each panel during the Workshop
were presented to the participants during the final plenary session.
Draft copies of the statements were distributed the following week to
the respective panel members for final review. These statements
represent the panel recommendations to EPA and are presented without
editing by the contractor, as agreed during the conference.
-------
EPA WORKSHOP SCHEDULE
JULY 16-20, 1979
TUESDAY WEDNESDAY THURSDAY .FRIDAY
OJJV.
10:00-
10:30
12tOO
12:30-
1 • ACV
•a. 1C.
4:OO
A. "^rv
7 • nrv.
10:00-
i
REG i!
T,
1
\
5TRA-
tON
7
f
PLENARY
SESSION
V
CCFFEE
PLENARY
SESSION
INITIAL
PANEL
MFFTING<7
BREAK
CASH
BAR
DINNER*
PANEL
MEETINGS
COFFEE
PANEL
MEETINGS
BREAK
LUNCH*
PLENARY
MEETING
COFFEE
PANEL
MEETINGS
BREAK
PANEL
LEADERS
MEETING
PANEL
MEETINGS
COFFEE
PANEL
MEETINGS
BREAK
LUNCH*
PLENARY
MEETING
COFFEE
PANEL
MEETINGS
BREAK
PANEL
LEADERS
MEETING
PANEL
MEETINGS
PANEL
MEETINGS
COFFEE
PANEL
MEETINGS
BREAK
LUNCH*
FINAL
PLENARY
MEETING
PANEL
LEADERS
MEETING
PANEL
LEADERS
MEETING
*ARRANGED TO FACILITATE THE CONDUCT OF THE WORKSHOP.
Figure 1. EPA Workshop schedule,
10
-------
SECTION 4
PLENARY SESSIONS
The plenary sessions were arranged throughout the Workshop to
serve as a forum for exchange of information, as a means to stimulate
discussions, and as a catalyst for the individual panels to achieve
the primary goal of the conference, namely, to develop constructive
guidance and recommendations for EPA. Section 4 summarizes the high-
lights of these sessions, starting with the opening plenary session
and later focusing the attention upon major topics under discus-
sion during subsequent sessions.
The first session was devoted partly to the opening statements by
EPA, EPRI, DOE, NOAA and the contractor (NAWC). A brief summary of
these statements in order of presentation is given below. During the
second half of the first session, each panel leader presented opening
remarks describing the tasks and the goals of their respective panels.
A brief synopsis, with a list of the main topics for discussion by
each panel, follows.
The third portion of this section deals with questions and
issues that were generated by the panel leaders' progress reports dur-
ing the subsequent plenary sessions. The "strawman" plans developed
by EPA served the purpose of immediately focusing the participants'
attention upon specific issues. The material presented herein shows
these issues to be primarily the scale of future field studies rela-
tive to parameters to be measured, considerations in experimental de-
sign and data management, measurement techniques, and cost factors.
1. OPENING STATEMENTS
1.1 Summary of Opening Remarks by L. E. Niemeyer, Environmental
Protection Agency
Considerations of complex terrain modeling problems were pio-
neered in the early 1940's by Hewson and Gill and reached a new
prominence a few years ago when NOAA and the predecessors of DOE and
EPA conducted the Southwest Energy Study. Other efforts have included
a recent complex terrain study funded by the EPA and contracted by
Geomet, Inc., and the Albuquerque Workshop organized by DOE three
11
-------
years ago. Standard Gaussian models have been adopted, but the
improvement of the models has been difficult without the necessary
funding for a comprehensive evaluation program. However, in light of
the recent emphasis on oil shale and coal resource development in the
West, EPA management has earmarked approximately $2.0 million per year
for three years in order to make a significant impact on the problem
of modeling plumes from large sources in mountainous terrain.
In a strawman approach prepared for the Workshop, EPA has sug-
gested three main directions for studying the modeling problems:
1) A small hill study to focus on the key question of plume
impingement,
2) A full scale study to verify the resulting plume impinge-
ment module developed under the small scale study,
3) Other aspects of the modeling problem such as plume trans-
port, frequency of plume impact ion for comparison with the
standards, and turbulence or dispersion statistics.
The goal of the Workshop is to provide recommendations for a
workable program considering EPA ideas but extending and revising them
based upon the experience and background of the participants.
1.2 Summary of Opening Remarks by Einar L. Hovind, North American
Weather Consultants
The planning and organization of this conference have been under-
way since March 1979 when EPA formally awarded North American Weather
Consultants (NAWC) the contract to conduct this Workshop. EPA and
NAWC agreed that the best way to achieve the stated purpose, namely to
develop recommendations to EPA on the problem of plume dispersion from
large fossil fueled power plants in complex terrain, was to arrange a
five-panel Workshop with about 47 experts on the following panel
topics: Model Development and Analysis, Model Evaluation and Applica-
tions, Experimental Design, Measurement Techniques, and Data Manage-
ment and Quality Assurance.
The Workshop schedule consists of a mixture of plenary sessions
and panel meetings, starting with this opening plenary session where
representatives from EPRI, DOE, and NOAA will provide information on
proposed and ongoing research programs conducted by the electric power
industry and by government agencies other than EPA. It is felt that
this information is highly relevant to the tasks placed upon each
panel of this Workshop.
12
-------
These tasks will be outlined later in this session by each panel
leader. They will follow a presentation by the EPA Project Officer on
the Workshop Goals, and the manner in which EPA expects to utilize the
results of this Workshop in the planning of future field programs and
modeling efforts on dispersion in complex terrain.
1.3 Summary of Related EPRI Research Programs by Glenn R. Hilst,
Electric Power Research Institute
EPRI projects that are directly involved with complex terrain
problems are organized under two main programs. The first is called
the Air Quality Studies program and has two primary features. One is
concerned with regional air quality studies of long range transport
and multiple source effects. The major problems considered under this
program have to do with fine particles and their chemical speciation,
source attribution, and effect on visibility. A network of air qual-
ity stations is being instituted in the western region for gathering
information on these problems. The second part of the Air Quality
Studies is the Sulfate Regional experiment which is now in the analy-
sis and modeling phase of its definition of SOg-sulfate relationships.
The second major program, which is more related to individual
plume behavior, is called the Plume Model Validation Project. The
primary goal is the collection of substantial experimental data for
the purposes of validating models for their general applicability and
validating individual predictive modules, such as plume rise, trans-
port, diffusive mixing, etc. Toward this end field projects are being
organized to study plume behavior in progressively more complex situa-
tions, beginning with a plains site, a coastal site, and then finally,
a complex terrain site. It is expected that more model development
will take place with the complex terrain data base, because knowledge
of plume behavior in these situations is limited.
In general, it should be recognized that any program dealing with
complex terrain questions is going to be extremely expensive, and that
the investment will have to be proved worthwhile. Nonetheless, deci-
sions must be made on what is necessary to answer these questions ir-
respective of the cost. Certainly one way to maximize the benefit
gained from the various research efforts is to promote coordination
and cooperation among the interested parties so that the total of what
is learned is more than the sum of the parts.
13
-------
1.4 Summary of Related DOE Research Programs by David S. Ballantine,
Department of Energy
DOE conducted a Workshop in Albuquerque in 1976 to address the
poor state of knowledge of transport and dispersion in complex ter-
rain. The Workshop outlined a recommended national effort at a level
of about $30M over a five year period. It included a discussion of
the need to study various complex terrain features and characteristic
meteorological systems.
When new funding did not materialize, DOE redirected ongoing re-
search efforts in the atmospheric sciences and in 1978 began a complex
terrain program in the Geysers geothermal area of California. The
Geysers area represented a site of opportunity where there was large
scale energy development and where considerable data on meteorology
and air quality had been collected because of environmental concerns.
The focus in Geysers is on a study of nocturnal drainage winds.
Dr. Marvin uickerson of LLL was selected as Project Director for
the DOE Atmospheric Studies in Complex Terrain program which bears the
acronym ASCOT. An initial two week field study was conducted in late
July 1979 with participation by LLL, LASL, ATDL, PNL, WPL/NOAA, and a
local contractor ES&S (Environmental System and Services). A major
field experiment will be conducted in September 1980.
A preliminary critical review of complex terrain models was con-
ducted by the ASCOT modeling group, which resulted in a recommended
model development program which includes various statistical, dynamic,
and physical modeling alternatives.
A longer range program is evolving for efforts beyond 1981 when
the Geysers study should be completed. Various options based on con-
siderations of different terrain features, meteorological phenomena,
and relevance to DOE energy development programs are being examined.
1.5 Summary of Related NOAA Programs by Vernon Derr, National Oceanic
and Atmospheric Administration
The responsibility of NOAA, and particulary the Environmental Re-
search Laboratory, in this Workshop, is to aid in assessing the role
of remote sensors and other sensors (such as aircraft instrumentation)
that NOAA operates or that come from other sources. The main objec-
tive is to design and perform experiments to provide a widely appli-
cable data set, which is acquired with a minimum of measurements and
14
-------
against which we can test models. In order to do this, several ques-
tions must be answered regarding such things as the representativeness
of the data, the resolution in time and space, and the continuity of
measurements required. It is also necessary to determine the basic
set of weather conditions most concerned with impact, and how to de-
scribe these conditions at different scales. Another question to be
addressed is the optimum configuration of remote sensing equipment.
Remote sensing equipment includes a wide variety of techniques
and measurement capabilities, some of which are briefly described
below.
1) A three-wave length lidar system (.347, .694, and 10.6 m)
with polarization capabilities to identify and trace parti-
culate pollutants
2) A new doppler-lidar system for measuring winds in clean air
3) Laser to measure wind or detect particulates
4) Microwave radiometric devices for measuring temperature,
and humidity profiles, and liquid water content
There are two needs for sensors in this kind of a program. One
to determine the prevailing meteorological conditions, and the other
is to determine the actual paths of the particulates or gases. For
these needs a combination of in situ, aircraft, and remote sensors is
required.
1.6 Summary of Workshop Goals by George C. Holzworth, Project Offi-
cer, Environmental Protection Agency
EPA has decided that its initial effort in its complex terrain
modeling program should focus on the problem of terrain impact ion be-
cause this has been a source of controversy for both the regulatory
agencies and the power industry, and because it is felt a significant
impact can be made on this topic within available resources. More
specifically, the problem has to do with a plume in a stable at-
mosphere encountering one or more terrain obstacles and particularly
the first obstacle.
The primary goal of the study will be to generate significant re-
sults as soon as possible to demonstrate what can be done with the
funds available. By significant results we mean the development of
models with a demonstrated higher degree of reliability than current
models in general use. These models must be reasonable and practical
15
-------
in terms of requirements for computing facilities, meteorological
input and plant operations data. However, one way to effectively
improve the modeling results is to specify more representative input
information than is currently required.
The study plans envision the development of stable impaction
models through a series of experiments on different scales, ranging
from physical modeling to full scale field studies, which are linked
together to demonstrate their credibility.
The strawmen proposed by EPA (Appendix A) are suggestions for the
direction of this work. The goals of the Workshop are to make recom-
mendations on the following aspects of the study:
1) The overall plan
2) Model development approach and implementation
3) Model testing and demonstration of credibility
4) Model applications in new settings
5) The scientific feasibilities of proposed field and labora-
tory studies
6) Efficient means of data handling
7) Budget allocations
In summary, the Workshop is designed to solicit scientific rec-
ommendations on how to most effectively and efficiently impact the
problem of mathematically modeling dispersion over complex terrain and
to do so with available resources.
2. PANEL LEADERS' DISCUSSION
The following outline is a synopsis of the major ideas presented
in the first plenary session by each panel leader while addressing
complex terrain issues in general and the scope of the proposed study
in particular.
2.1 Dr. Bruce A. Egan - Model Development and Analysis Panel
A. Some of the problem areas associated with complex terrain situa-
tions that might be addressed are:
16
-------
1) Regulatory issues (prediction of 3 and 24-hour worst cases).
2) Good Engineering Practice issues.
3) Near-surface releases.
4) Long distance transport.
5) Low level sources in a constraining valley situation.
B. The following are constraints on the scenario developed
in the Workshop:
1) Want a model developed that will be useful in regulatory
applications for complex terrain settings.
2) Need meaningful results on a timely schedule to assure
future funding.
3) Program must have a high probability of success which pre-
cludes some approaches which may not contribute to regula-
tory applications.
4) The approaches should complement related efforts.
5) The end products should be able to serve as a basis for
future efforts.
C. Within the constraints discussed, some directions are suggested:
1) Concentrate on a few high priority parameterization problems
(such as the impact of a nearby elevated source).
2) Emphasize the simple models such as the Gaussian models, but
collect data that would be more generally useful for other
models.
3) Relate the simple field study and physical modeling and
establish uses and limitations for the physical modeling.
4) Define the flow field and distortion effects, gradients of
turbulence levels, and trajectories for specific source
locations.
5) Separate the flow field model development and the disper-
sion model development.
17
-------
D. Questions and Discussions from the Floor are briefly summarized
below:
1) What averaging times are necessary to define the flow field?
Something considered consistent with the scale of the prob-
lem which implies 5-20 min in the case of Snyder's strawman.
2) In reference to the point about separating the flow field
and dispersion models, software has already been developed
for treating the problem in this way.
2.2 Mr. Maynard E. Smith - Model Evaluation and Applications Panel
A. Considerations of various aspects of the planned study are dis-
cussed below:
1) The focus should be on limited experiments which would pro-
vide maximum information. In other words, the study
shouldn't become so elaborate that one part of the program
has to be sacrificed for another.
2) It is desirable to know to what extent wind tunnel modeling
can be relied on, and so comparing this technique with the
small scale study is appropriate. However, this is not
likely to be the case in the large scale study due to the
additional complexities of the situation.
3) Input data requirements must be simplified as much as possi-
ble in order that data be effectively acquired under actual
field situations. This means data must be obtainable in
reasonable time periods and with reasonable instrumentation.
4) A panel of expert review may be valuable.
5) The budget should include funds for evaluating and comparing
the model with other field data.
6) It is important to consider and define how the model ought
not to be used, as well as how it should be.
B. Comments from the Floor:
1) In response to the comments on wind tunnel modeling it was
pointed out that progress has been made in modeling drainage
and convective transport. However, this kind of modeling
18
-------
may not be widely available to users involved in the regula-
tory process, and could not therefore be reasonably required
in that process.
2.3 Mr. Gene Start - Experimental Design Panel
A. The design of the experiment depends on answers to some of the
following questions:
1) What is the goal from the modeling standpoint? Are we trying
to test and discriminate alternate theories? Which models
should be considered?
2) What scale should be used - Lagrangian, Eulerian, or both?
3) What aspects of the problem should be emphasized - plume
rise, impaction, etc?
4) How should the problem be approached? This question relates
to what measurements should be taken and what scale and res-
olution are most appropriate.
B. Comments from the Floor:
1) The point was raised that it is important to design the
study so that results will be visualized and accepted by
others.
2.4 Mr. Roy Evans - Measurement Techniques Panel
A. The techniques selected will largely be dictated by the scenario
designed and the site chosen. A small scale tracer study and a
full scale power plant study imply different logistics, some of
which are outlined below:
1) Height rise - measurements will be simpler in the tracer
study because of lower altitude gained by non-buoyant
plumes.
2) Mixing height - not scale dependent.
3) Tracer species measurements - the tracer study would use a
fixed ground based network to measure tracer species, while
the power plant study would involve measuring several dif-
ferent pollutant species.
19
-------
4) Surface concentrations - simpler in the smaller study.
5) In-plume measurements - more extensive in full scale study.
Small scale study is less suited for remote sensing
techniques.
6) Source terms - more complex in full scale study.
7) Wind profiles, temperature, and stability - scale is
larger in the power plant study so measurements will be
more complicated.
B. Questions and Discussion from the Floor:
1) What is the value of aircraft sampling when such techniques
as lidar are readily available? The aircraft makes the
plume more accessible. Also, it is unlikely that any of
the techniques would be accepted without cross-comparison
and corroboration with other techniques.
2) It is important to distinguish between concentration fluc-
tuations and plume movements produced by variations in up-
wind velocity and by turbulence or separated flow over the
hill. It would be desirable to correlate concentrations on
the hill and upwind velocity fluctuations.
2.5 Dr. Ronald E. Ruff - Data Management and Quality Assurance Panel
A. Considerations related to data collection:
1) A data acquisition plan based on how data are to be re-
corded is needed. Techniques vary from manual to real-
time read-out.
2) The level of acceptable data accuracy must be established in
light of instrumental accuracies of about 10-20% and model
accuracies of about an order of magnitude.
3) Data should be flagged in some way to indicate its validity.
4) Field review and final checks are necessary in real-time or
short times in order to efficiently run the experiment.
B. Considerations related to quality assurance:
1) Instrument Selection
20
-------
2) Operational Procedures
3) Calibration and Audits
4) Checks on the overall system
5) Data Verification
C. Considerations related to the final product:
1) After the data base is assembled, final validation, correc-
tions, and changes must be made in an interactive way.
3. SUMMARY OF DISCUSSIONS DURING PLENARY SESSIONS
The primary objective of the experiment was established early in
the discussions to be that of an end product which would address the
regulatory problem by providing a better understanding of impingement
concentrations, particularly with respect to the 3-hour standard. The
principle controversy arose from considerations of what was the basic
approach best suited to achieving that end product (i.e., what size
obstacle to study). Other discussions explored the wide variety of
topics that could be studied within the framework of a given approach,
as well as the specific methods that could be used. The following is
a summary of some of the questions raised and comments made during the
plenary sessions.
3.1 100 m Hill vs. 400-500 m Hill
While examining the small scale study approach, the Workshop par-
ticipants became divided over studying a 100 m hill (as suggested in
Snyder's strawman, Appendix A) or a somewhat larger hill. In general,
those concerned with modeling favored the smaller hill, while those in
experimental design and measurements preferred the larger one.
The advantages of the smaller hill were seen to be its simple
geometry, its simpler dynamic influences, and alleged smaller costs
involved in studying it. In general, the ease of doing the study (for
example, being able to move the source around it) would be expected to
result in a more complete data base for addressing the impingement
problem. Another benefit lay in relating the results from the small
hill study with those from fluid modeling, while being one level
closer to the real world. It was also argued that the essence of sta-
ble conditions is that the flow is independent of height, so the size
of the hill is unimportant. Finally, recent experiments in Australia
and Great Britain have tied together turbulence and wind measurements
21
-------
on 150 m hills with calculations and laboratory experiments, suggest-
ing that achieving similitude is not out of the question.
The reservations about the 100 m hill centered mainly around the
transferability of the results to real world situations. In particu-
lar, it was felt that the surface boundary layer and the vertical
structure in the lower atmosphere would not be representative of the
meteorology at plume level in real cases. Turbulence processes were
viewed by some as being critical to the understanding of impaction,
and it was feared that the 100 m hill would not allow proper simula-
tion of them. Another point made was that, while the 400 m hill study
would involve additional physical complexities such as drainage flow
and thermal effects, those complexities are viewed by some partici-
pants to be essential in determining plume impaction. In addition,
the study of the larger hill would allow the use of more sophisticated
techniques and a preliminary estimate was given that the cost would
probably not be more than 20-50% higher than for the 100 m hill study.
The outcome of these discussions was that since no concensus
could be reached on this issue the decision on what size hill to study
was more properly the task of EPA management.
3.2 Small Scale and Large Scale Studies
The consensus of opinion was that both the small and large scale
studies ought to be done, with the small scale study coming first
chronologically. The discussion on this topic pointed out some of the
advantages, disadvantages, and suggestions for each.
The primary advantage of the small scale study (irrespective of
the 100 m or 400 m hill controversy) was that its simplicity offered
hope for success at a reasonable cost. There was also more flexibil-
ity in that the source could be moved, and the study could easily be
adapted to examine such questions as those related to GEP considera-
tions or the effects of plume meander. The scale would be small
enough that more real time feedback could be available. In addition,
it has a close relationship to physical modeling. Also important was
the idea that with a smaller obstacle, the opportunities for observing
stable flow in the proper direction would be greater.
On the other hand, the large scale study was viewed as close to
real world conditions. Many of the instruments could be more effec-
tively used at this scale which would be greater than their limits of
resolution. It was suggested that a site be chosen that was not so
complex as to preclude physical modeling. Another suggestion was that
22
-------
some of the advantages of the small scale study should be retained by
using a non-buoyant tracer and movable source.
In summary, it was expected that a combination of the small and
large scale studies would produce the most meaningful results as long
as the limitations of each were recognized.
3.3 Measurements
In general, the kinds of measurements desired were independent of
the scale of the study, while the specific instruments recommended
were heavily influenced by it. At any level, surface concentrations
were considered to be paramount. Wind field, stability, and tempera-
ture data were also critical, as were turbulence measurements.
Flow visualization techniques were emphasized heavily. Data related
to source terms and dispersion would also be necessary, and would be
more extensive during the large scale study.
As the discussion turned more to the full scale study, the empha-
sis on more sophisticated techniques increased. Lidar, both ground-
based and airborne, was expected to play an important role. In situ
aircraft measurements of various parameters also were considered to be
an integral part of the measurement arsenal.
While techniques would not always be adaptable to the different
scales, it was recognized that transfering as many as possible from
the first study to the next would be more cost effective.
3.4 Costs
Relatively little discussion was devoted to costs, however some
specific points were made during the plenary sessions on such items as
budget division and relative cost of small scale versus large scale
studies.
Early in the discussions a suggested budget division was made by
the Measurement Techniques and Data Management Panels as follows:
Project Management 10%; Field Work and Initial Data Processing 55%;
Data Analysis and Quality Assurance 15%; Model Development and Evalu-
ation 20%.
During subsequent sessions, the same panels presented estimated
costs for various field program scenarios. The instrumentation alone
for a 100 m hill experiment, as originally proposed by Snyder, was
estimated to be around $2.5 million, providing real-time read-out from
an array of towers with 8-level instrumentation. These figures were
23
-------
lowered somewhat as the need for the tower heights and real-time sam-
pling requirements were reduced. It was noted, however, that the
small hill experiments would employ costly in-situ measurements which
would have to be coarser to offset costs. Also, on the larger scale,
the argument was made that more information is returned for the type
of sensors in the field.
24
-------
SECTION 5
PANEL RECOMMENDATIONS
This section contains recommendations to EPA, which were
generated by this Workshop through the five panels. They represent
the viewpoints, expressed orally and in writing by the panel members,
which were organized into comprehensive documents by the Panel
Leaders. These statements have been circulated among the panel
members for comments before the Panel Leaders submitted them to the
Contractor for publication.
It should be noted that very limited time was available for
this process. The Panel Leaders would have preferred more time
for the demanding review, and organizational and editorial tasks to
assure overall continuity and fluency in the statements which follow.
However, it was the expressed desire of the Project Officer and
Contractor that as much as possible of the written inputs by the
panel members should be retained, and that the publishing of the
proceedings in a timely manner was more important than editorial
perfection of these statements.
The following recommendations prepared by the five panels are
presented unabridged as agreed by the Project Officer, Contractor,
and Panel Leaders.
25
-------
1. MODEL DEVELOPMENT AND ANALYSIS PANEL
PANEL LEADER: Bruce Egan
PARTICIPANTS:
Norman Bowne Jack Cermak
John Clarke Douglas G. Fox
Philip Gresho Julian C. R. Hunt
Robert G. Lamb Jeffery Weil
Robert Wilson
1, REPORT OF THE MODEL DEVELOPMENT AND ANALYSIS PANEL
1,1 Overview of Deliberations
This panel Twas concerned with generating the basic approach
of the Workshop and the review and selection of the basic scenarios
which EPA should follow in their overall program. The panel dis-
cussions concentrated on defining what was needed to meet EPA's
most pressing needs regarding model improvement for regulatory-
oriented applications in complex terrain. The panel's approach
was from the perspective of knowing the modeling basis on which
EPA regulatory decisions were now commonly made (e.g. , on the basis
of VALLEY, CR8TER or similar models) and from the conviction that
there was considerable room for improving the physical representations
inherent in the model algorithms, especially in complex terrain
settings.
Implicit in the discussions was the recognition that a descrip-
tive and conclusive measurement program would be required to rectify
the discrepancies in the currently used and presented models. For
this reason the initial discussions centered on what sort of field
programs would be most useful. Later discussions centered on the
expected nature of the mathematical modeling effort which would be
required both in the short and the long term, the roles that physical
models could play in model development, and the uses to be made of
the measurement data to be collected.
26
-------
1.2 Perceived Goals of, and Constraints on, the EPA Program
The panel understood that EPA was primarily concerned with the
development of a credible model for regulatory applications involving
the near field impact of fossil-fuel fired facilities located in the
vicinity of rough terrain. The question of the magnitude of possible
plume impingement was of major concern. Other constraints on the
program discussed included:
• Overall budget in the range of two to two and a half million
dollars per year for about three years.
• A need to produce meaningful and useful end products during
the program to help assure the year-to-year funding.
• The program should .be designed to have a high probability
of success in meeting the goals of EPA.
• The program should be designed to complement other related
efforts.
• The program should result in modeling tools with a demon-
strated high level of reliability.
1.3 Recommended Approach
The Model Development and Analysis Panel had a consensus that,
to meet EPA's goals, objectives, and budgetary constraints, EPA pro-
ceed with a two phase research plan which involves:
(1) A detailed study of the interaction of a tracer gas plume
with an isolated hill located in otherwise relatively flat terrain.
This study should be performed on a physical scale small enough to
allow detailed tracer measurements at a large number (of order 100)
of locations on the hillside, have the possibility of testing a range
of release heights and different release locations, allow the
use of a variety of flow visualization techniques, and allow tower-
based measurements of wind speed and direction and measures of
turbulence at a number of levels above the surrounding plain and
above the crest of the hill. We suggest that the data need could
be met with a hill of order 100 m high.
(2) A further, larger scale experiment which would build upon
the findings of the first experiment, yet still be focused upon
obtaining a better understanding of the interaction of an elevated
plume with high terrain. This larger scale expriment would allow
27
-------
the systematic investigation of more phenomena, which in practice
further complicates the problem of predicting ambient air quality
levels in complex terrain settings. Specifically, this larger
scale experiment would require more detailed investigations of the
effects of locally-generated, thermally-driven circulations of
flow patterns, the magnitude and likelihood of plume impaction,
the effects of wind shear on dispersion rates, and the effects
that elevated inversions and other phenomena associated with atmos-
pheric stratification have on plume dispersion.
(3) A closely coordinated mathematical model development pro-
gram having the following elements:
• Initial use of existing and appropriate modeling techniques
in the detailed design of both the small scale and large
scale studies. This use would help assure that monitoring
sites are optimally located for maximum benefit.
• Initial model development oriented toward making maximum
use of data to provide reasonable "fix-ups" to Gaussian-
based models currently used in regulatory practice. This
was expected to take place during the first two years of
of the program.
• Subsequent model development activites concentrating on
developing and using better descriptions of the flow fields
expected in rough terrain for a variety of meteorological
and topographic settings. It is anticipated that this
effort would result in flow field descriptions which could
be used with either Gaussian-plume-based models (by
defining trajectories and distortion effects) or with
"K theory"-based models.
This was expected to take place in the later two to four
years of the overall program.
• Initial models developed to meet the requirements of
routine regulatory use which include; demonstrated improve-
ment over current methods, relatively easy to use and
inexpensive to operate, and capable of using a minimum
of routinely available input data.
(4) Data analysis efforts keyed to providing a better physical
understanding of the phenomena of interest. This will involve
qualitative analyses based in part on the results of flow visualiza-
tion experiments to be performed. The data analysis and model
28
-------
verification efforts should examine model performance (based on ob-
served vs. predicted comparisons) using different levels of input
information. For example, the model performance can be evaluated,
using only data from a nearby airport, only data from an on-site
meteorological tower, or using the more comprehensive data sets from
the detailed measurements to be taken around and on the surface of
the experimental hills.
(5) A coordinated and complementary physical modeling program.
This program should involve defining the applicability and limitations
of the use of physical models (wind tunnel, water towing tank) by
comparisons of results with the small scale field program. Based
upon the demonstrations of valid uses of the physical modeling tech-
niques, these methods might be applied to expand the data base for
testing mathematical models and to better understand the flow be-
havior for different terrain configurations. The results of these
techniques presumably can also be used to help design the larger
scale field program.
Figure 2 depicts the expected flow of information among the
program elements.
Several other comments about the panel's deliberations and
the reporting herein appear important to include at this point:
(1) Some of the initial "scoping" discussions touched upon prob-
lems of achieving reliable performance with even significantly im-
proved mathematical models when these were used to predict the highest
or second highest concentration values expected during a period
of a year, as required by present National Ambient Air Quality
Standards (NAAQS) and Prevention of Significant Deterioration (PSD)
rules. Panel members agreed that a re-definition of such standards,
which would require predicting the highest values occuring a few
percent of the time, would make more sense from a modeling point of
view. The error bands between modeled and observed values would be
expected to be much smaller in this latter case and greater re-
liability could consequently be put on modeling results.
(2) It was decided that, given the constraints of available time
for creating this report and the nature of the material to be covered,
it was preferable to not attempt a major editorial effort to make
the submission totally non-redundant or consistent in style. It was
also felt that such an editorial effort would tend to obscure some
of the flavor of individual opinions vs. those better described as
panel consensus, and that the report would thereby lose some value
to future readers.
29
-------
Parameterization
and Verification
MODEL
DEVELOPMENT
Design
REGULATORY
APPLICATIONS
Design
SMALL SCALE
STUDY
Data
Data
DATA ANALYSIS
Design
PHYSICAL MODEL
STUDIES
Design
LARGE SCALE
STUDY
JData
DATA ANALYSIS
Parameterization and Verification
Figure 2. Diagram of flow of information within program
30
-------
The remaining sections of this panel's report are basically
comprised of contributions by individual panel members who volun-
teered to summarize specific topics addressed by the Model Development
and Analysis Panel. The following topics are addressed:
1. The Small Hill Experiment
2. Model Development Activities
3. Uses of Field Data
4. The Role of Physical Modeling
5. Description of the Full Scale Experimental Program
1.4 The Small Hill Experiment
This panel unanimously agreed that a small scale experiment
(100 m) similar to that proposed in Snyder's "Strawman", and focusing
on the issue of parameterizing the highest air quality impacts of
an upwind source during stable, neutral, and unstable conditions,
would be a good choice to meet EPA's immediate goals and constraints.
The specific advantages of this experiment, which were cited, include
the following:
• The results of this experiment would address the major
modeling problems associated with permitting sources located
in rough terrain.
• The data offered real hope of success in improving upon
presently used modeling methods.
• Significant results could be achieved at reasonable costs.
• The experiment can be inexpensively expanded to also consider
issues of Good Engineering Practice (GEP) in the presence of
a terrain object.
• The scale and scope of the experiment allowed the possi-
bility of providing real-time or, at least, short-turnaround
feedback on the nature of the findings. Along these lines
it was strongly felt that having a relatively simple experi-
ment, which was focused on investigating the most important
technical issues, would avoid some of the problems which
have plagued other large scale experiments attempting to
meet very broad objectives.
31
-------
• The scale of the experiment was sufficiently small to allow
extensive use of flow visualization techniques (such as
multiple smoke releases on the hill surface) and portable
meteorological measurement devices (such as anemometers on
short masts) to investigate closely the boundary layer
flow on the hillside. The panel felt that such augmentations
to the basic measurement program would add greatly to the
overall worth of the program by supplying a better funda-
mental understanding of the flow phenomena of interest.
• The small scale study allowed the possibility of using a
relatively mobile source release. Specifically, the fact
that the tracer source height could be systematically raised
and lowered (perhaps even two sources could be released
simultaneously), and moved to varying distances from the
hill, gave the experiment both possibilities of examining
the relationships (such as source height to hill height
ratios), and would result in a higher data capture rate
(the source could be moved on a multi-hour basis to be
more-or-less upwind of the hill).
The panel identified priorities to be set for the measurements
to be made as follows:
(1) Near-ground surface (breathing level or less) measurements
of tracer material concentration values in order to provide a direct
relationship of emission rates to ambient air concentrations.
(2) Meteorological measurements from several levels on a tower
of at least hill height and preferably one and a half hill heights.
These measures should include anemometry for the three components of
wind speed and turbulence statistics (e.g., sigma theta, sigma E, and
standard deviations of vertical and horizontal wind speed fluctuations
averaged over 10 minutes or so). Differential temperatures and heat
flux measurements should be required. This tower should be located
on flat terrain away from the hill at a location where, under prevail-
ing wind conditions, the tower would measure the air flow unaffected
by the presence of the hill. A second, similarly instrumented tower
(50 m) should be on top of the hill. A few ten to thirty meter
towers should be located on the sides of the hill. Tracer measure-
ments should be made on each tower. Meteorological measurements
should also include sampling of the vertical temperature structure
of the atmosphere (temperature soundings and/or acoustic sounders)
and the possible supplemental use of a doppler radar for wind
measurements.
32
-------
(3) Measures (perhaps by remote sensing by lidar of fluorescent
particles) of the elevated plume concentrations and geometry upwind
of the region of interaction of the hill. Specifically, measures
of sigma z and sigma y were very desirable, provided they were taken
in sufficient detail or over a sufficiently long averaging time, to
produce estimates useful for averaging times as large as one hour.
(4) Flow visualization of flow phenomena - smoke releases, etc.
Surface smoke releases should be made simultaneously from 25 to 50
well chosen locations to ensure good definition of the mean flow
field. This is essential for reaching an understanding of the funda-
mental fluid dynamics of the flow over the hill and Reynolds Number
effects for comparisons with the physical model results. The smoke
can be photographed by aircraft and ground-based cameras.
The 100 m hill experiment ws envisioned to have the following
additional characteristics and details:
(a) An "ellipsoidal" hill should be found which allows study
of plume behavior approaching both the narrow and broad
sides in order to understand the effects of hill aspect
ratio on maximum ground-level concentrations.
(b) On the order of one hundred samplers should be positioned
on the hill side (not all of which need be run for each
experiment). A suggested array is presented in Figure 3.
(c) The panel did not feel that it was necessary to have all
concentration measurements in real time or with sampling
times of 10 minutes. On the contrary, the concern was
expressed that in order to properly understand the effects
of wind meanders, etc., a one hour averaging time would
be necessary to final interpretation and extrapolation
of results. However, it was desirable to have a few
continuous and realtime samplers in the experiment. After
an experiment is underway, the tracer release should con-
tinue from the same location for a duration of at least 3
hours.
(d) The panel also did not feel that it was absolutely es-
sential that the ground-based array be of such close
spacing so as to always be able to capture the plume or
to always allow direct determinations of the cross-wind
or vertical dimensions of the plume. The panel recog-
nized that very narrow plumes might miss the sensors
33
-------
100m Tower
T
Sampler-
TOP (PLAN) VIEW
1/Sfc
24 Sampters Each End
END VIEW
ITovrar
In1 ^ _*_
18 Samplers Each Side
(within dashed ttnas)
• •
SIDE VIEW
Figure 3. Suggested array of samplers for 100 m hill experiment
34
-------
during some experiments, but if the experiments were
performed a sufficient number of times, the largest ex-
pected impacts would be observed. The model validation
should be.focused on values at the high end of the frequency
distribution of concentrations - not just the highest
values.
(e) Experiments should be run for all meteorological conditions
and should not focus solely on behavior during stable
conditions.
(f) Several different tracers released at different heights
can have the utility of conducting multiple experiments
within a single experiment.
(g) Surfaca temperature fields could be obtained by indirect
techniques (IR measurements from aircraft).
(h) The statistics of wind flow onto the hill from a given
source location should be examined and related to the
statistics of local wind stations.
Specific limitations noted for the small scale experiment are
as follows:
(1) The wind and temperature structure that an elevated power
plant plume may encounter 300 to 800 m above the surface is not
likely to be similar to that through the first 100 m. This, however,
is not likely to affect any conclusions relating to the ratio of
maximum concentrations in the plume (immediately upwind) to those on
the surface.
(2) The small hill is not likely to be as thermally and thus
dynamically active as a large hill, i.e., upslope and downslope flow
will be unimportant. Also, the surface roughness and processes by
whic'u the plume interacts with the immediate surface will not be
adequately simulated on the small hill.
(3) Only a limited range of source heights relative to hill
height can be examined with a movable tower release. Specifically,
it may not be possible to have a release height greater than 70 m.
Other methods of releasing the tracer gas from higher elevations
should be investigated (e.g., use of a balloon).
(4) The hill is not tall enough to extend up through the
radiation inversion depth where variable temperature profiles and
35
-------
wind shear effects are important, i.e., processes that become im-
portant for plumes from large power plants will not be thoroughly
considered.
(5) Boundary layer structures, like those found on larger
hills of different slopes, will not be considered.
(6) Only flow on a hill having one aspect ratio can be explored
in the initial field study. (Note that the physical modeling
techniques may prove to be useful in expanding this data base).
Recognizing the limitations of the results of the small scale
experiment noted above, the panel believes it will be necessary
to also study impingement-related phenomena at a larger scale study
once the results of the small scale experiment have been analyzed.
However, the small scale experiment is an essential first step.
A debate arose in the general Workshop plenary sessions as to
whether the simple single hill experiment should be performed at a
site with a 400-500 m hill or at something as small as 100 m. This
panel had the point of view that some of the additional flow phenomena
to be expected at a 400 m site (local flows become more important,
wind shear effects, etc.) would tend to complicate the experiment
and the 100 m hill was slightly preferable from this point of view.
The major conviction of the panel of the merits of the 100 m size
was based, however, on the expectation that the flexibility and data
capture possibilities at a 100 m hill site would be severely compro-
mised at a larger scale site. The panel was specifically concerned
over the impracticality of constructing a tower of sufficient height,
and the logistical problems associated with performing the same
experiment on a large scale. The panel also felt that heavier
dependence on remote sensing techniques at the larger site would
not focus as well on the issue of actual ground-level concentrations.
2. MODEL DEVELOPMENT ACTIVITIES
2.1 Introduction
2.1.1 Objectives. As evident from numerous legal debates, a
serious need exists to develop a model capable of accurately predict-
ing concentration, especially maximum ground-level concentration,
in the vicinity of proposed sources located in complex topographic
environments; that is, locations where the topographic feature
height is a significant fraction of the source height (or much
greater than stack height).
36
-------
The proposed model development is strongly guided by existing
practice. Since existing practice includes use of VALLEY and CRSTER,
the model development should lead to either improved or substitute
models. The primary objective of model development is to improve
the accuracy of these models in predicting maximum ground-level
concentration.
2.1.2 Method. Perhaps the major area of controversy associated
with applications of air quality models is plume impaction on nearby
elevated terrain surfaces. Recent theoretical developments suggest
that under certain stable flow conditions, plumes can impact on
terrain. Physical models in stably stratified water channels and
wind tunnels of the flows around simple hills have demonstrated
that the impaction phenomena predicted does in fact occur, but
little information exists regarding the magnitude of this impact.
The model development program will seek to include results from
the theoretical developments, the field studies, and the physical
modeling experiments in order to provide model simulations of plume
impaction. This represents a step toward improving the scientific
justification for air quality modeling and a highly important step
with regard to application of the models in regulatory proceedings.
For neutral and slightly stable conditions, it is expected that
the conventional Gaussian model can be altered to yield acceptable
results. For the stably stratified case it may be necessary to
introduce a separate formulation.
It is also recognized that all significant situations in the
application of air quality models do not involve simple isolated
hills. Thus, the model will need to be developed in such a manner
that it can be applied to more complex situations. It will there-
fore need to be tested against other data previously collected in
addition to that collected in this program.
Increased complexity in the topographic situation will probably
require development of some form of meteorological or flow field
simulation in order to properly direct the plume in complex terrain.
Consideration should be given to the development and validation of
such a technique within the framework of this project and given the
constraints of acceptability for regulatory application.
2.2 Constraints on Model Development Program
It was suggested that the modeling program be divided into two
phases consistent with immediate and long-term objectives.
37
-------
2.2.1 Immediate (1-2 Years). This phase is closely coupled
to the small-hill impaction study and should observe the following
ground rules:
a. Should result in useable and valid improvements to the
VALLEY and CRSTER models.
b. Should be easy and inexpensive (computer resources) to use.
c. Should require a minimum of input data.
d. Should be successful in simulating surface concentrations
from the small-hill study over the range of plume impinge-
ment phenomena under very stable to neutral stability
conditions. Stability in this context is being character-
ized by the Froude number (Fr) of the upstream approach
flow given by Fr = U/Nh, where U is the velocity at hill
height, h, and N is the Brunt-Vaisala frequency.
The implication of these ground rules is that: (a) proposals
for changes in existing approved models be given priority; (b) a
variety of existing approaches should be studied; (c) little totally
new work can really be developed (i.e., the first phase modeling
will be based on the Gaussian-type models).
2.2.2 Long-Term (2-4 Years). This phase will build on the
first phase, generalize the model where possible, and be developed
simultaneously with the design and (perhaps) implementation of
the full scale study. The following ground rules apply.
1. Improved methods of estimating the transport wind field
for application to new "sites" are required. Locally driven surface
flows (upslope, drainage, etc.) should be considered for inclusion,
as well as a first estimate of other effects caused by complex
terrain such as vertical variation (shear) of the wind.
2. Should be reasonably economic in terms of computer resources
(storage/execution time).
3. Should also "succeed" in describing experimental results
from the earlier small-hill study.
4. Should be capable of treating different situations equitably
in regulatory applications.
38
-------
The implications of the above led to the following suggestions
for longer term model development activities:
(1) A review of the state of knowledge in diagnostic wind
modeling on scales up to 200 krn
(2) Development of a complex topography flow field data set
stable and neutral (perhaps also unstable) conditions for
for testing diagnostic wind models
(3) Early (within 2 years) selection of, at least, a generic
model to diagnose wind fields
(4) Utilization of the generic model approach to provide
guid?nee in design of the second phase field experiment
(5) Study of Gaussian and other parameterizations of plume
turbulence for coupling diffusion with the flow model
(6) Final verification and testing against field programs
of combined approach and initiation of the process to
include the new model in OAQPS guidelines.
2.3 Immediate Model Development
The following steps are suggested:
(1) Consideration of the state of knowledge (capability) in
modeling source emissions in complex topography including
recent theoretical developments of stratified flow theory
and results of wind tunnel and water channel experiments
simulating plume behavior in stratified and neutral flow.
This effort should focus on the potential use of:
(a) 2-D potential flow theory as an approximation of
the mean streamlines in the horizontal plane for
highly stable conditions (Froude number much less
than one);
(b) linearized theory (small amplitude, inviscid approxi-
mation for moderately stable conditions (Froude
number about one); and
(c) potential flow theory for the windward side of
terrain features during near neutral conditions.
39
-------
The ability of such approximations to adequately simulate
interactions between closely spaced topographic elements
where the flow is dominated by the shedding of vorticity
should be carefully studied. This effort should also
review recent findings regarding the motion of stagnation
points due to downstream vortex shedding, diffusion
rates when temperature gradients vary considerably over
the terrain object, the effects of slope winds on plume
centerline trajectories, and the most appropriate ways of
parameterizing plume dispersion in separated flow regimes.
(2) Critical review of existing "Recommended" (albeit defacto
recommendation) modeling algorithms, i.e., CRSTER and
VALLEY as well as other Gaussian-b?sed models. Compare
model algorithms and assumptions with theory and experi-
mental results as identified in (1) above. Develop a
"matrix" of proposed improvements which considers changes
in the parameterization in the Gaussian based models of:
horizontal and vertical diffusion, plume trajectory
alterations, stability classifications, and space and
time averaging effects.
(3) Proposed alterations of Gaussian-based models to include
results of 1-2. This would include consideration of each
component "parameterization" in models such as VALLEY and
CRSTER and the "matrix" of proposed improvements. This
effort could, among other alternatives, attempt to intro-
duce terrain correction factors into the Gaussian method-
ology and use combinations of diagnostic flow models with
various diffusion parameterizations to simulate transport
and diffusion in complex topography.
(4) Develop/adopt standards of performance for a complex
topography model consistent with those included in the
EPA modeling guideline and having a sound technical
basis.
(5) Testing of the matrix of improvements against the field
data collected by this study.
(6) Final recommendation of specific improvements.
40
-------
2.4 Long-Term Models
2.4.1 Diagnostic Wind Field Modeling. Since even the best
of the first generation (immediate) models may stumble in new and
seriously complex terrain situations, owing, in large part, to the
lack of an adequate (realistic) wind field, the next generation
of models should be directed toward ameliorating this deficiency.
Another concern is the increased opportunity for misapplication
of such model vis-a-vis flat terrain Gaussian models. Towards this
end, models which generate (diagnostically) some sort of a reasonable
3-D wind field should be developed and coupled to either (a) a
modified Gaussian model, or (b) a numerical solution methodology
for the advection-diffusion equation. This model could be based
on input data of varying degrees of detail, ranging from climato-
logical averages to a selection of site specific measured winds.
In all cases, the input data would be used to generate an improved
estimate of the 3-D wind field at a set of grid points, the bottom
surface of which represents the topography. Small-scale physical
models are expected to be useful in the evaluation of such models.
The spectrum of such modeling approaches is fairly broad,
ranging from those which simply generate wind fields conserving
mass (and do not penetrate topography) to those in which additional
physics is simulated. Hopefully, the best of these could include
effects of stratified flow, perhaps via a Froude number modification.
2.4.2 Toward the "Ultimate" Model. Again it must be borne
in mind, when examining the limitations of the second generation
models that even the best of these (which could be one combining
the best features of these currently available), even after being
"calibrated" via the full-cale field data, may not always perform
well in new and different topographic settings. The reason for this
is, as before, that such models contain insufficient physics to always
reliably generate wind fields which are strongly affected by details
such as terrain shape and small-scale driving forces (e.g., drainage
wind and upslope winds), or even large-scale effects such as mountain
lee waves and the strong downslope winds often assocated with these.
Although a fully comprehensive model would be a desirable end
product, and would be much more generally applicable, it is recognized
that such a model development program not only requires much more
time and effort than is allotted under the ground rules, but requires
a greater understanding of flow behavior under a variety of complex
conditions than is likely to be available. For these reasons, it is
prudent at this time to restrain model development as discussed
earlier.
41
-------
Nevertheless, it is appropriate to briefly discuss such compre-
hensive models in the event that future programs recognizing the
need are planned, or that simpler models are found to be inade-
quately extrapolative. The comprehensive model would be one which
solves a finite difference approximation to the equations of mass
momentum, and energy conservation on a set of grid points which
includes an accurate topographic description. The transport model-
ing could be performed, using finite difference or finite ele-
ment methods, and turbulence could be modeled either via first-
order (K-theory, with locally varying diffusion coefficients) or
second-order method, either of which must recognize and address
the current shortcomings of turbulence parameterization. Currently
such methods give no idea of the statistics of concentration fluctua-
tions (i.e., the effects of averaging times) and cannot be used for
unsteady flows. Random walk techniques may be as, or more, appro-
priate. Once such a model is developed, it could be calibrated
in the following ways: (1) Numerous simulations of wind tunnel
results covering a wide range of topographic and thermal variations
(some of which will have, hopefully, been verified against field
data) and (2) Several field study data sets, depending largely
on their availability (the data set obtained from the full scale
test of this project is a prime candidate) and completeness. Such
an approach, while significantly more complex, costly, and time
consuming (both in development and in computer time), would appear
to have a high probability of ultimate success and could lead to a
model which is truly general and confidently applicable to new and
different potential sites.
3. USES OF FIELD DATA
3.1 Summary
In this section of the report we consider the uses of the data
from the small and large scale field experiments. This section is
divided into two broad subdivisions.
1. Comparing the data with models, qualitative mathematical
and physical.
2. Using the data for improving the models (especially those
currently used by EPA (e.g., VALLEY, CRSTER)).
3.2 Comparing the Data with Models
3.2.1 Qualitative Models. Many air pollution regulatory de-
cisions are ultimately (explicitly or implicitly) based on qualitative
42
-------
models of the flow patterns and plume trajectories. Therefore, the
first use of the data, especially flow visualization data, must be to
help better define these qualitative models and to reduce the areas of
controversy surrounding them. Specific items for analysis are:
(a) Identify the range of temperature gradients and wind speeds
that the air flow passes around or over the hill.
(b) Compare the nature of the separated flow regions as a
function of Froude number with laboratory experiments.
(c) Identify conditions when plumes are prevented from imping-
ing into hills or valleys by downslope winds and when they fumigate
onto hills.
3.2.2 Mathematical and Physical Modeling. The general approach
governing the comparison between the field data and mathematical
or physical models should be to: 1) use the data governing the
approach flow [e.g., velocity field, turbulence measures, temperature
profile, height and quantity of plume release, nature of the hill,
etc.] to predict various aspects of the velocity and concentration
distributions over the hill, and 2) to then compare these with the
observations.
A list of some of the useful predictions and comparisons should
include:
(1) Comparisons of surface flow patterns over the hill (based
on smoke visualization) with model predictions as a function of
Froude number and surface heating/cooling rates.
(2) Mean plume trajectories. These can be calculated using
potential flow theory for simple hill shapes a priori from the hill
geometry and given Fr and compared with the observations.
(3) The frequency of impinging plumes and the local plume
meander in stable conditions. These can be related mathematically
to local wind statistics and perhaps the wind statistics from a
nearby meteorological station.
(4) If possible the velocity and turbulence structure over
the hill top should be compared with other neutral field experiments
and mathematical/physical model studies.
43
-------
(5) In the larger field experiment, significant measures
of up and down slope winds are expected including the depth thick-
ness, velocity, and turbulence of these winds (e.g., in the Navajo
Vermillian cliffs experiments large up and down slope winds were
observed with velocities of the order of 5-10 m/s and with layer
thicknesses of 20-30 m). These measurements should be compared
with theoretical values.
(6) The observed spread statistics, a , a , and maximum
concentration upwind of the hill, on the hill, and over the hill,
should be compared with complex terrain calculations based on:
(a) Pasquill-Gifford spread statistics.
(b) Local upwind measurements of velocity fluctuation sta-
tistics, and the upwind temperature gradient.
(c) Measurements over the hill of velocity fluctuation sta-
tistics, a(z), temperature and wind profiles and the
plume centerline locations.
(7) Compare the ratio of the surface concentration on the hill
at one point to the plume centerline concentration immediately upwind
of this point. Since the location of maximum surface concentration
varies quite rapidly as the wind direction changes, it is important
that these be compared as a function of the averaging times of the
plume measurements and of the ground-level concentrations. These
data will help considerably in interpreting previous field studies
where the averaging times were frequently different.
3.3 Using The Data for Improving Models
3.3.1 VALLEY Model
(1) Impact Ratio. One major use of the experimental data
should be to define the ratio of the maximum ground-level concentra-
tion on elevated terrain to the plume centerline concentration
immediately upwind of the terrain. These values can be obtained
from the small-scale study as a function of flow field conditions.
The EPA VALLEY model assumes total ground reflection of the
plume, resulting in a ratio of 2.0, which is thought by some to be
conservative.
The measured values of the ratio should be compared to values
calculated by mathematical models and measured by physical models.
44
-------
The satisfactory agreement of the model estimates with measured
values would allow the applicability of models to be extended to
flow, terrain, and heating/cooling conditions not measured in the
experiment(s).
(2) The Assumed Plume Width a . Currently the 24-hour average
concentration is based on assuming that the plume is spread out
over a 22 1/2 sector. This assumption must be compared with ob-
servations and possibly some improved way of estimating this sector
may emerge.
(3) Statistics of Plume Impingement. This leads to the
next major use for experimental data, which is to determine the
frequency of occurrence and persistence of, terrain impact flow
conditions in complex terrain. This should be accomplished through
long-term (at least one year) continuous meteorological measurements
to arrive at hourly average flow conditions. Such data should pro-
vide the necessary input to the mathematical model for determining
second highest concentrations for a given source during a year for
various averaging times, e.g., 3-hour and 24-hour. These estimates
will be site-specific. However, recommendations might then be derived
which would be generally applicable for specification of the minimum
(number and type) of meteorological measurements needed at complex
terrain sites to determine the frequency of occurrence and persistence
of terrain impact flow conditions. When a significant source of
pollutants is proposed for location in complex terrain, these
recommendations can be used to specify the necessary site-specific
data to show compliance with regulatory standards.
(4) Plume Downwind of a Hill. Currently the VALLEY model
assumes that a plume travels on downwind of a hill with no change
of dispersion rates. Physical models suggest that the plume widens
out to the hill's width, b, so that its concentration is decreased
by (b/a ) where a is the plume width prior to impingement. Hope-
fully this aspect of VALLEY could be improved. (a is also ex-
pected to increase downwind).
(5) Application of VALLEY Model. Currently this model is
applied when dT/dz is greater than a critical value. Physical and
mathematical modeling suggests very strongly that the criteria for
the application of this kind of model should be based on the value
of a dimensionless number such as the Froude number.
(6) Criterion for Flow Pattern. As indicated earlier, Fr
should be criterion for the flow pattern and for the plume impinge-
ment. Hopefully, this study should also indicate when strong up or
45
-------
downslope flows may alter this criterion and change the value of
the impact ratio.
3.3.2 Moderate Stability and Neutral Flow Models
(1) Terrain Following Models. In some models when the tempera-
ture gradient is less than a critical value, a neutral flow model
of the flow is used. The simplest of such models assumes that
the plumes follow a trajectory which remains at the same source -
displacement above the hill. Clearly the experimental results should
be compared with this model, and the differences noted. We anticipate
that this model will drastically underestimate maximum surface
concentration - especially for three-dimensional situations.
(2) Interpolative Models. In such models some allowance
is made for the trajectory approaching the hill surface (e.g., half-
height models). This approach is an improvement but is likely to
be insufficient in many applications.
(3) New Models. New models being developed under EPA contract
are oriented toward regulatory use and may form the basis of im-
proved techniques. If so, they should be carefully compared with
the data in both a regulatory manner as well as in a more scien-
tific manner.
4. THE ROLE OF PHYSICAL MODELING
In the last two decades physical modeling of the atmospheric
boundary layer has advanced to a state that permits accurate pre-
dictions of atmospheric transport and dispersion for many sites
and meteorological conditions. Two types of facilities have been
developed that enable thermal stratification of the lower atmosphere
to be simulated. These are the meteorological boundary layer wind
and the towing tank (water channel). Model studies of flow and
dispersion around an isolated hill are proposed to: a) complement
data obtained from the field study, b) check the degree of similarity
between small-scale model and full-scale data, and c) determine
the relative merits of the two physical modeling methods.
4.1 Background
The first attempt to study stably stratified flow over a
topographic obstacle in the laboratory was made by Abe (1929). Flow
over dry ice in a small wind tunnel produced a crescent lee wave
downwind of a 1:50,000 scale model of Mt. Fuji that was in qualitative
agreement with full-scale observations.
46
-------
In 1934 Prandtl and Reichardt (1934) designed and constructed
the first meteorological wind tunnel with a capability for development
of a thermally stratified boundary layer. This facility was destroyed
during World War II before being put to its intended use. Following
these early developments Cermak (1958) designed and constructed a
meteorological boundary layer wind tunnel at Colorado State Uni-
versity. This facility and details of the early developments are
described in a paper by Cermak (1975). The 30 m long test-section of
2 x 2 m cross-section is equipped with a floor that can be cooled (or
heated). Air heated (or cooled) in the return duct develops stably
(unstably) stratified boundary layers 1 m thick with temperature dif-
ferentials up to 200°F. This wind tunnel is well suited for physical
modeling of either stable or unstable flow over an isolated hill.
The history of towing tanks to study atmospheric motion is
somewhat obscuie. References to some early developments, and a
description of the towing tank proposed for physical modeling of
flow over an isolated hill, are given by Hunt, Snyder and Lawson
(1978). The channel is 1 - 2 m deep, 2.4 m wide and 25 m long.
The model is towed on a ground board through water that is stably
stratified by a linear salt-concentration gradient.
Physical modeling of flow and dispersion around the isolated
hill in both facilities is recommended, since each type of facility
offers distinct modeling advantages and disadvantages. Therefore,
comparison with full-scale results will help advance the science of
physical modeling while providing data to complement the field
study.
4.2 Modeling Criteria
Similarity criteria for physical modeling of the atmospheric
boundary layer are given by Cermak (1971) and Snyder (1972). A
discussion of similarity for stably stratified flow around an
isolrted hill is presented in Hunt, Snyder and Lawson (ibid).
Two types of similarity should be considered - similarity of
the atmospheric boundary layer approaching the hill, and similarity
of local flow around the hill. As discussed by Cermak (1971), the
boundary-layer characteristics (excepting for turning of mean wind
direction with height) will be similar if the following conditions
are met:
1. Equality of bulk Richardson No. (AT/T)gL/U 2 (L = vertical
distances used to determine temperature difference, AT).
47
-------
2. Equality of Relative Roughness K /L for Upwind Approach;
(L = length of upwind boundary; K = equivalent height
of roughness elements).
3. Reynolds No. UL /(v) in excess of value required to make
drag coefficient independent of K /L . On smooth hills
the location and extent of separation appears to be a
sensitive function of Reynolds number as well as K /L, .
so
4. Equality of relative boundary layer thickness 6/h, (h =
height of hill, 6 = boundary layer thickness).
Local flow around the hill is expected to be similar if, in
addition to satisfying the foregoing conditions, the following
conditions are met:
1. Equality of Relative Roughness, K /h.
s
2. Hill-Reynolds No. Uh/(v) is in excess of value required
to give similar location of separation for given value
of Kg/h.
3. Equality of product of Hill-Richardson and Reynolds Nos.
(This is a tentative relationship which needs experimental
confirmation).
Ouh\ AT_ hJ
A1s h /Uhh\ A1s h
g i i = g
T IT,. \ v / T IL
h
T = difference in temperature between hill surface and
air.
4. Model hill is geometrically similar to prototype hill.
5. Surface-temperature distributions are similar for model and
prototype hill. Frequently the Richardson No. for approach-
flow similarity is replaced by the square root of its
reciprocal - the Froude No. U/[(AP/p)gh] ' (Hunt et
al, 1978).
4.3 Similarity in the Meteorological Wind Tunnel
In the meteorological boundary-layer wind tunnel, the approach-
flow similarity parameters can be set to cover the following ranges:
48
-------
1. Bulk Richardson No. (for entire boundary-layer depth) -40
(unstable) to +20 (stable).
2. Relative roughness of upwind approach - as required for site.
3. Reynolds No. - usually well above minimum required since
L, = 30 m.
b
4. Relative boundary layer thickness - approximately equal to
prototype value. A length scale of approximately 1:300 is
suggested for the 100 m hill; therefore, the value of 6/h
will be about 3 for the model.
Local flow similarity parameters can be varied as follows:
1. Relative roughness of hill - equal to prototype.
2. Reynolds No. - approximately 7000 minimum for 100_m hill
or 1:300 scale (U = 1 ft/s; h ~ 1 ft; v* 1.5 x 10
ft2/s).
3. Richardson and Reynolds Nos. on the hill cannot be made
equal to prototype because of h dependence. Therefore,
gravity flows induced by heating and/or cooling the hill
may not be similar to full scale.
4. Geometrical similarity - as required.
In terms of the hill Froude number, the approach flow can be
adjusted in the range 0.4 to infinity for the stable case and the
same magnitude of parameter change for unstable flow. Thus, a wide
range of meteorological conditions can be simulated, including an
elevated inversion. The latter condition is created by cooling
the upstream wind tunnel floor and warming a length of floor upstream
of the model.
4.4 Similarity in the Towing Tank
The approach, flow similarity parameters, can be set to cover
the following ranges:
1. Richardson No. - in terms of the hill Froude No., the range
of values is 0.1 to infinity (neutral flow).
2. Relative Roughness of Upwind Approach - because of short
ground board, equality of this parameter is not achieved.
49
-------
3. Reynolds No. - L is ordinarily too small to achieve suf-
ficiently large values.
4. Relative boundary thickness - not equal (5/h
-------
For Froude No's, less than one, full simulation of the approach
boundary layer may be relatively unimportant for achievement of
representative flow around the hill. This is true because flow
passes largely around the hill with little change of elevation and
the approach flow is quasi-laminar. However, for Froude No's, greater
than one, the upstream flow will almost certainly be turbulent and an
upwind plume may be transported over the hill surface in the boundary
layer flow. Accordingly, the towing tank has some advantage for
study of plume impingement at extremely low Froude No's, (including
a range of 0.1 to 0.4 that the wind tunnel cannot achieve). On the
other hand, the wind tunnel has an advantage that plume impingement
can be studied for Froude No's, from 0.4 to 1 and from 1 to infinity
where simulation of the upwind boundary layer is essential. Thus,
through physical modeling in both types of facility almost all
conditions encountered in the field (excepting for some meandering
and low-level jets) can be simulated. Furthermore, for Froude No's.
in the range of 0.4 to 1 both types of facilities should provide good
simulation and thus provide a good basis for comparison of results.
3
The minimum hill Reynolds No. Uh/z v should exceed 10 for
both facilities. Therefore, unsteadiness of the impinging flow due
to vortex shedding from the hill should be representative of full-
scale behavior. At these Reynolds No's, it is possible for the
boundary layer on the hill (for a sufficiently rough surface) to
be quasi-turbulent and provide an accurate simulation of dispersion
near the surface.
Simulation of up-slope and down-slope winds on the hill created
by heating and cooling of the surface is still in the research state.
Full-scale values of the product of Hill-Richardson No. and Hill-
Reynolds No. cannot be attained by the small-scale models because of
dependence upon the length scale squared. Therefore, a satisfactory
modeling of this phase of flow can be achieved only if dependence upon
this product becomes weak at some attainable value of the product.
There are several potential benefits to be derived from a com-
parison of physical-model simulations of flow and dispersion about
the isolated hill to the field observations. Principally, the
comparisons would demonstrate the applicability and/or show the
limitations of physical models to this situation. This demonstra-
tion is especially important for strongly stable flows (F 0.1) past
hills, where plume impingement can occur. Assuming that satisfactory
correlation is achieved between the physical model simulations and
field observations, this study would support further use of physical
models in basic or generic investigations of flow about obstacles
(such as those of Hunt, Snyder and Lawson (ibid), Kitabayshi, et al,
51
-------
(1971), and Lin and Binder (1967). It would also lend credibility
to the use of physical models in site specific studies in complex
terrain where information is required on the maximum surface concen-
trations. Furthermore, the measurements on flow and surface concen-
trations from physical models could be used with confidence in
testing mathematical models.
Presently, the simulation of thermally induced upslope/downslope
winds in laboratory models is in the developing stage. A combined
simulation of these winds superimposed on a mean flow toward a
surface obstacle is not foreseen too likely in the early part of
the research program. However, in a few years such a combined
simulation may be possible. Simulation of thermally induced winds
would be more critical in laboratory modeling of flow about the large
scale (several hundred meters) terrain feature.
5. DESCRIPTION OF FULL SCALE EXPERIMENTAL PROGRAM
5.1 Objectives
The objectives of the full scale program are developed on the
assumption that a small scale experimental program is conducted at
an isolated hill to define the relationship between impingement
concentrations and those of the unimpeded plume. Objectives are
listed in order of priority.
5.1.1 Definition of Flow Fields. A major problem in complex
terrain is the definition of the wind field and the resultant tra-
jectory of material introduced into the air. Since the transport
of the plume must be known, at a minimum, the wind flow at plume
height must be defined in detail. Coupling the transport flow
fields at the complex site to the geostrophic flow field may permit
a method of transferability to be developed. At the other extreme,
the local flow field may be coupled to a physical model in a tank
or tunnel. There are simple models of terrain modified wind fields
in existence that provide a mean wind in the surface layer. Evalua-
tion of these models can be made from a suitable observation program.
If they provide satisfactory estimates of the transport flow field,
a major portion of the complex terrain problem will be solved.
5.1.2 Impingement. The major concern of the EPA is the impinge-
ment problem, that is what is the maximum concentration at ground
level when the plume impacts elevated terrain? The second objective
of the complex site study is to determine if findings regarding
impingement can be transferred from the small scale study to the
52
-------
full scale site. The study should develop ratios of maximum ground-
level concentration to free plume concentrations as a function of
differences in the physical and meteorological characteristics of
the sites.
5.1.3 Understanding Physical Processes. A comparison of im-
pingement effects could be made on the basis of statistics, but
that would preclude transferability of the results to other sites.
An understanding of the physical processes involved in the transport,
dispersion, and impaction of the plume is a primary scientific ob-
jective. The individual processes of concern include plume impinge-
ment, occurrence of upslope or downslope and valley winds, downwash
or wake effects, depth of slope winds as a function of surface
temperature field, local turbulence and effect on concentrations
and the influence of topography on the transporting flow field.
5.2 Measurements Required to Meet Objectives
The measurements required to meet the objectives are discussed
in the following sections in the same order as the objectives were
set forth.
5.2.1 Flow Fields. Measurement of the flow field may be
measured by a combination of conventional anemometry on towers near
the source and on prominent terrain features, doppler acoustic
sounders and perhaps tetroons.
Surface wind measurements will provide information on slope
winds. Since the primary objective of knowing the flow field is to
determine plume trajectories, it would seem that direct measurement
of actual plume trajectories would be most useful. This could be
accomplished by aircraft mounted lidar tracking.
The time scale of observations should be sufficient to evaluate
turbulence, but short enough to approximate straight line segments.
Average values of direction and speed for periods of 10 minutes,
to not more than one hour, are considered appropriate. Spatial
scales are unknown at this time and are dependent on how much the
topography actually influences wind flow at plume height. Comparison
of plume or tetroon trajectories, with mixed measurements provided
by towers and doppler acoustic sounders recommended here, should
provide answers to this problem. The outcome of an intensive plume
tracking experiment in this program should be directed to supplying
requirements for a minimum density of measurement devices. Rawinsonde
observations should be made to tie local observations to the routine
synoptic scale observations.
53
-------
5.2.2 Impingement. Direct measurement of impingement concentra-
tions must be made on prominent terrain obstacles subject to impinge-
ment. The time scale of concentration measurement must be appropriate
to the air quality standard. Consecutive hourly values that can be
averaged for periods up to 24-hours are required. In addition, a
measure of plume concentration in the free air prior to, but close to,
the impingement point must be measured. The time average of this free
air measurement must be the same as the basic time of averaging of
the ground-level concentration. The source term, that is the emission
rate of the quantity whose concentration is being measured, must be
known.
Spatial variability of concentration patternss will be high. It
is not practical to have enough sensors on the terrain obstable to de-
fine the maximum. A rule-of-thumb generated years ago states that
there should be 10 samplers in six standard deviations of the Gaussian
plume to define the maximum concentration with good reliability.
Even using tracer samplers such density is unlikely. However,
continuous sampling to achieve a large number of samples with time
rather than space can be substituted in an attempt to capture the
maximum concentration. Therefore, continuous sampling is recommended.
Physical modeling may be used for development of flow fields, fill-in
concentration and wind fields, and provide a basis for model-prototype
comparison to evaluate validity of model results.
5.2.3 Physical Processes. In the section on objectives, it
was noted that the ability to transfer results to other complex sites
is desired. In order to understand the physical processes listed
in Section 5.1.3, the necessary measurements must be made in adequate
detail. These are listed and discussed.
Impingement frequency - a combination of plume visualization and
surface measurements is required. The areas where impingement is
expected on the basis of climatology and physical modeling should
have measurement capability. Plume tracking by lidar or photography
of a visible plume will also provide such data. There must be enough
observations to describe the frequency of occurrence of impingement
with respect to specified flow patterns.
Slope winds - it is expected that upslope and downslope winds
will be major mechanisms affecting plume concentration at the ground.
The occurrence, strength, and depth of the slope winds should be
measured by local anemometry or with visual means, such as smoke
tracers and photography, to establish the relationship between
observed concentration and flow. Measures of ground temperature and
54
-------
surface heat flux are also required. Hourly averages should be
sufficient.
Downwash or wake effects - if the plume is in the lee of a
ridge or hill affecting the vertical position of the plume, the
magnitude of the displacement should be documented by indirect
measurements with lidar or DIAL systems. The measurements should
be sufficient to establish any wave pattern with distance, as well
as fluctuations in heights, at a fixed point. Hourly mean values
with estimates of the fluctuations, if the fluctuations affect the
final stabilization height or ground-level concentration, are re-
quired. Wakes behind obstacles, such as an isolated hill or ridge
lines, may seriously deform the plume. If such events occur, they
should be measured by indirect means such as lidar or COSPEC. Concen-
tration will probably be too low for successful use of a DIAL system.
The expended width and/or depth of the plume should be measured and
analyzed as a function of obstacle parameters and plume size before
encountering the obstacle.
Slope wind depth - if slope winds are found to affect ground-
level concentration, then it will be necessary to analyze them in
more detail. It has been suggested that slope winds may be cate-
gorized as to depth and strength as a function of surface temperature.
Radiometric measurement of surface temperature by aircraft mapping
or regular thermometry may be used. A need for hourly values over
a wide area would suggest standard methods rather than aircraft
sensing.
Local turbulence - the local turbulence will be responsible for
diffusing a laminar plume flow onto the hillside in stably stratified
conditions. But plume meander, which strongly affects the concentra-
tion at a point, is not caused by local turbulence, but by unsteadi-
ness in airflow around the topography (e.g., vortex shedding). If
local turbulence is a function of surface roughness and heating (and
it better be), then direct measurement of turbulence intensity should
be related to local concentration. If the relationship is measured
at several points to establish patterns and if patterns are found to
exist in a quantifiable form, then a method of transferring findings
to another site exists. The turbulence measures should be averaged
over the same time period as the basic concentration measurement,
i.e., one hour or less.
Influence of topography - in Section 2.1 it was noted that the
topography influence on plume trajectory was important and that
trajectories would be studied. The purpose of this section is to
provide measurements, either direct plume measurements or indirectly
55
-------
by wind measurement, of the influence of specific terrain obstacles
on the wind pattern at plume altitude. The purpose is to establish
standing patterns as a function of geostrophic wind direction and
terrain aspect ratio. Hourly averages should provide sufficient
time resolution.
5.3 Description of Field Program
The previous sections outlined the objectives and the measure-
ments necessary to meet those objectives. The measurements were
stated on an individual basis. In this section we provide an over-
view of our estimate of the field program necessary to obtain the
measurements in an integrated manner.
Figure 4 is a sketch of the field program except for pollutant
or tracer ground-level concentration measurement sites. Routine
measurement of some variables are called for all the time. Intensive
periods of measurements are interspersed for more detail.
5.3.1 Routine Measurements. A meteorological tower of at least
100 meters should be erected, if it doesn't already exist at the site,
to measure temperature, temperature difference, wind speed and
direction, and turbulence near the source on a continuous basis.
Additional shorter towers should be located on the top of the hill
to measure wind speed and direction, temperature, and turbulence.
Turbulence measurements should be three-dimensional. Other ane-
mometers and temperature devices should be placed in the valley and
on the valley sides to measure slope winds routinely. Two u,v doppler
acoustic sounders should be employed in the valley to profile the
wind above the tower. Rawinsonde observations should be made 2 to 4
times per day. These routine observations will establish the general
climatology of the temperature and flow fields of the area.
Routine measurements of ambient concentrations of stack gases
should be made to establish concentration patterns. Fifteen to
twenty fixed monitoring stations should be located on prominent
terrain features and in the valley to establish the annual distribu-
tion of pollution patterns and frequency distributions of concentra-
tions at a point. Hourly averages should be acquired. Pollutants
to be measured, in order of priority, are S02, NO-NO , and 0,,.
In addition, routine high volume air sampling for total suspended and
respirable particulates should be carried out on a 24-hour basis.
5.3.2 Intensive Periods. Intensive measurement periods of 4
to 6 weeks duration should be carried on as usual, but additional,
more detailed measurements would be added.
56
-------
57
-------
Two aircraft should be employed. One containing a particle
lidar will be used to determine plume position out to distances of
50 km or more for as many hours as the crew and equipment can be
used. The second aircraft will be employed to fly through the plume
to establish free air concentration patterns of gases. Sufficient
passes must be made over a fixed position to establish hourly
averages for relationship to ground-level monitors.
A ground-level lidar and perhaps DIAL could be employed to
examine near field plume behavior for downwash effect, trajectories
and plume position.
The ground-level monitoring network is not capable of providing
adequate spatial resolution of concentration patterns. Tracer
studies are recommended. Tracer material may be released either
continuously or for specified periods during each intensive period
and sampled at 200 locations. Samples should be collected for one-
hour periods each. Currently available samplers are capable of
collecting 12 to 24 samples on preset timing using battery power.
Suggested tracers are SFfi or perf luorocarbons utilizing gas chromato-
graphic analysis. A few (up to 5) continuous tracer monitors should
be employed to provide a time history of tracer concentration.
One member suggested the tracer not be injected in the stack
but released from a tower. Another member felt strongly that the
tracer be emitted from the stack because plume rise is a significant
part of the problem. An alternative of using both methods was also
suggested; the latter (tower release) being used to systematically
investigate the effects of different release heights.
5.3.3 Analysis Timing. Real time collection of some of the
meteorological information is required to plan special intensive
measurements such as tracer studies. Real time collection of other
data is not required. Unlike the isolated hill, fast turnaround,
i.e., the next day, is not likely to be available and should not be
necessary. However, it is important to complete the analysis of
data from one intensive period before going to the field for the
next period.
5.4 Model Development
The model development effort that is a part of the large scale
study should focus on the flow fields rather than the plume dispersion
and impingement processes. The main objective should be the develop-
ment of a technique by which the three-dimensional flow and tempera-
ture fields within the atmosphere's first kilometer can be described
58
-------
quantitatively given the geostrophic flow conditions, the specifica-
tions of the local topography, cloud cover, etc. Flow and temperature
information is required to model plume rise, dispersion, and center-
line trajectory; the Froude number of hills in the plume's path
(needed to make concentration estimates on the hill); and, perhaps
most importantly, the annual frequency with which plume impaction
on particular terrain features is likely to occur.
Flow models of several types should be developed. Physical
models should be developed for the terrain site investigated in the
field and its simulated flow and temperature fields compared with
those actually observed. A demonstration that physical models provide
reliable simulations of the microscale flow regime characteristics
in complex terrain would help make this type of model a credible
source of flow data for other sites.
Numerical flow models should also be developed. A number of
"linearized" models currently exist that could be validated or
perhaps refined using the meteorological data gathered during the
large scale field study. If these simple models prove to be inade-
quate, far more advanced (nonlinear) numerical models should be
developed. In this case a constraint on the model design must be
that the computer time and memory requirements be small enough that
the model can run on machines accessible to a majority of potential
users.
Finally, simple empirical models can be developed that would
provide for given geostrophic flow and topography specification
(for example, valley depth, width, length, etc.), estimates of
drainage flow speed, stability and depth; vertical profiles, wind
speed and direction over the valley floor, etc. This model would
be used to provide information for plume rise and spread estimates
and frequency of occurrence of given plume trajectories.
In addition to the flow field modeling, part of the effort of
the large scale study should be devoted to the development of plume
rise and spread models that are applicable to complex terrain
environments where extreme variations in wind direction and speed
and temperature can occur in the vertical. Two-particle, Monte
Carlo type models are the simplist models that are best suited for
this purpose. The same model can also simulate the plume centerline
trajectory and spread given the 3-D flow and temperature fields and
information on the flow meander frequency and intensity.
59
-------
References
Abe, M. , 1929: Mountain clouds, their forms and connected air
currents. Bull. Central Meteor. Obs., Vol 7, No. 3, Tokyo, 1929.
Britter, R. E. , J. C. R. Hunt, and J. S. Puttock, 1976: Predicting
pollutant concentrations near buildings and hills. Symp.
Systems and Models in Air and Water Pollution. Institute of
Measurement and Control, London.
Cermak, J. E. , 1958: Wind tunnel for the study of turbulence in the
atmospheric surface layer. Fluid Dynamics and Diffusion
Laboratory, Tech. Rept. CER58-JEC42, Colorado State University,
Fort Collins, Colorado.
, 1971: Laboratory simulation of the atmospheric boundary
layer. AIAA Jl., Vol. 9, No. 9.
, 1975: Application of fluid mechanics to wind engineering,
ASME Jl. of Fluid Engineering, 97, Series 1, No. 1.
Hunt, J. C. R. , W. H. Snyder, and R. E. Lawson, 1978: Flow structure
and turbulent diffusion around a three-dimensional hill. Tech.
Rept. EPA-600/4-78-041, Environmental Sciences Research
Laboratory, Research Triangle Park, North Carolina.
Kitabayski, K. K. , M. M. Orgill, and J. E. Cermak, 1971: Laboratory
simulation of airflow in atmospheric transport - dispersion over
Elk Mountain, Wyoming. Tech. Rept. CER70-71 KKK-MMO-JEC65,
Fluid Dynamics and Diffusion Laboratory, Colorado State
University, Fort Collins, Colorado.
Lin, J. T. , and G. J. Binder, 1967: Simulation of mountain lee waves
in a wind tunnel. Tech. Rept. CER 67-68 JTL-GJB24, Fluid
Dynamics and Diffusion Laboratory, Colorado State University.
Prandtl, L., and H. Reichardt, 1934: Einfluss von warmeschichtung
auf de Eigenschaften einer turbulenten Stromung. Deutche
Forshung, No. 21, Berlin.
60
-------
Puttock, J. S., and J. C. R. Hunt, 1979: Turbulent diffusion from
sources near obstacles with separate wake - Part 1. Atmospheric
Environment Vol. 13, No. 1, pp. 1-13.
Sherman, C. A. , 1978: A mass-consistent model for wind fields over
complex terrain. Journal of Applied Meteorology, Vol. 17, No. 3,
March 1978, pp. 312-319.
Snyder, W. H., 1972: Similarity criteria for the application of
fluid models to the study of air pollution meteorology. Boundary
Layer Meteorology, Vol. 3.
61
-------
2. MODEL EVALUATION AND APPLICATION PANEL
PANEL LEADER: Maynard E. Smith
PARTICIPANTS
Edward W. Burt John S. Irwin
Harry E. Cramer Vincent A. Mirabella
Loren W. Crow Alvin E. Rickers
Donald Henderson William H. Snyder
1. INTRODUCTION AND OBJECTIVES
Our panel reviewed the stated objectives and priorities of the
EPA group, and attempted to adapt its discussions to respond to the
Agency's needs. As we understand them, the objectives are:
A. To establish as accurately as possible the maximum concen-
trations found when a stable plume impinges directly on
a local terrain obstacle.
B. To determine whether the diffusion or distortion of the
stable plume differs from that over flat terrain prior
to its intersection with the terrain.
C. To consider what input data would be required to make use
of new or altered models developed for evaluating such
conditions.
Because our panel was expected to address the question of appli-
cations, we tried to keep the regulatory aspects of the problem in
mind during our deliberation. For this reason, we would require the
contractor to relate the experiments to the National Air Quality
Standards or PSD increments.
To implement these objectives and meet our responsibilities
with respect to the evaluation and application functions, we estab-
lished an outline of discussions which included:
62
-------
A. A set of end products or desired results of the studies.
B. Consideration of the role of modeling within the experi-
mental studies and in the application of the results.
C. Consideration of the specific field experiments needed.
D. The evaluation of the field data.
E. The application of the results to regulatory problems.
F. Additional topics which we believe EPA should consider
with respect to complex terrain and modeling.
2. END PRODUCTS
As stated in the Workshop "strawman", the air quality modeling
system is being designed "to produce atmospheric dispersion models
that are applicable to large sources in complex terrain and that
have a demonstrated higher degree of reliability than existing
models". This basic objective outlines the major end product ex-
pected from the modeling program. This end product will be the
result of either the refinement of existing Gaussian models or the
development of new models embodying advanced techniques and using
the data base collected during the field experiments.
In order to accomplish the study objectives, and to allow proper
interpretation of results for regulatory application, it is necessary
to determine the wind, turbulence, and concentration fields as
functions of both time and space.
In terms of time resolution, a required end product is to
obtain concentration measurements for averaging periods ranging
from ten minutes or less to as long as three hours. Consideration
of this time range will permit the interpretation and proper ex-
tension of modeling results to be compatible with applicable stan-
dards. These time average specifications should, to the extent
feasible, be augmented by continuous concentration measurements to
provide an accurate characterization of the fine-scale structure of
the three dimensional wind, turbulence, and concentration fields.
In addition, a methodology must be developed for assessing plume
impacts up to a twenty-four hour averaging period.
In terms of the spatial scale to be examined, strong emphasis
should be placed on developing the modeling capability for describ-
ing near-field impacts, i.e., source-terrain separation distances
-------
ranging up to several kilometers. Consideration of long-range
transport, and its associated problems, should not be within the
scope of these impingement experiments.
An additional and very important end product of the study
should be an assessment of the transfer value of the experimental
results to other locations.
We believe that this project should include the following:
A. Specific evaluation of presently used "off-the-shelf"
models with inclusion of confidence limits.
B. Delineation of model applicability and limitations of
modeling results.
C. Specification of minimum meteorology and source input data
necessary to provide reliabile impact estimates.
D. Determination of the concentration probability distribu-
tions of the test data.
E. Examination of the meteorological conditions likely to
cause the highest second-highest limiting concentrations
in complex terrain for the three-hour and twenty-four
hour averaging periods.
F. Recommendations for further research efforts to extend the
results of this study to other terrain configurations and
time-averaging periods.
3. MODELS
The details of the models that will evolve from these plume
impingement studies cannot be specified a priori. However, there are
certain model features and modeling criteria that should be addressed
prior to the experimental design to insure the adequacy of the design
and to facilitate the application of the results to real sources. We
believe that it is very important to develop at the outset a model
construct that contains, in as much detail as possible, all of the
basic terms and parameters that will be required to analyze the
measurements obtained during the experimental program. It should
also provide for the generalization of the modeling techniques
developed from the experimental data, so that these techniques can
be applied to specific sources and terrain situations.
64
-------
The basic model construct should include characterization of
the following:
A. Source
B. Stable Approach Flow
C. Plume Characteristics Prior to Impingement
D. Impingement
E. Hill Surface Boundary Layers
F. Ground-Level Concentrations
G. Time Averaging
A prototype model construct containing these factors should be
developed and used to calculate expected ground-level concentration
patterns for various combinations of stability, aspect ratio, wind
speed, Froude number and other critical parameters for the experi-
mental terrain factors. One should also remember that various
hypotheses may apply, and a single model construct may not be
sufficient.
The importance of the prototype model construct and calcula-
tions is that they will force specification of source, meteoro-
logical, terrain, and flow parameters that control the various
processes, and provide an objective framework for experimental
design, data analysis, and model development.
4. SPECIFIC EXPERIMENTS
Our panel decided that two field experiments are desirable,
one on a relatively small scale and the other on the scale of a
major terrain obstacle. The first at least should be replicated
in a fluid modeling facility.
It is important that flow visualization should be conducted
before either sampling network is made final. Such tests would
insure appropriate placement of the sampling equipment, and might
provide valuable information on the diffusion, aerodynamic and
transfer processes as well.
Real-time measurements of the meteorological variables, and
some of the key concentration measurements, would be helpful in
65
-------
managing the experiments, but much of the concentration data need
not be immediately available.
4.1 Small Scale
Partly because of concern about extrapolating the results of
a small-scale experiment to very large terrain features, and partly
because of the interchange of ideas with other panels, we debated
the relative merits of 100 m versus 400 m hill experiments at length.
The conclusion was that the 100 m experiment was a better choice for
numerous reasons, the primary ones being that (a) the work would be
easier to accomplish, (b) it appeared to have a much greater chance
for success in illustrating, measuring, and understanding the physical
mechanisms involved in the transport and diffusion of pollutants
through the atmosphere to the hill surface. We had no objection to
a larger hill experiment if all equivalent measurements were made.
The panel also debated the relative merits of the 100 m hill
(Snyder) versus the 25 m hill (Lamb) experiments. The latter appeared
overly complicated and was likely to be too often submerged within the
turbulent surface layer under very stable flow conditions. A natural
100 m hill would extend above the turbulent surface layer much more
often, and it was our first choice.
The details of the experiment received considerable attention,
especially with respect to measurements which would insure physical
understanding of the mechanisms involved. We recommend:
A. Meteorological Measurements. Wind speed and direction, tem-
perature, and three components of turbulence intensity up to at least
1.5 times the hill height on each of the main towers: these data to
be logged for the duration of the study periods (twenty-four hours per
day for three to six months) not just during tracer release periods.
B. Concentration Measurements. A sufficiently dense network of
samplers on the hill surface as well as a few above the shallow
(turbulent?) surface layer would show whether there may be sharp
concentration gradients near the hill surface because of stagnation, a
turbulent surface layer, or katabatic winds. It would also help in
understanding of the key processes involved in the transfer of pollu-
tants from the laminar approach flow to the surface itself.
C. Hill Surface Layer Measurements. Attempts should be made,
preferably through relatively simple means (possibly smoke releases,
66
-------
smoke rockets, wool-tuft grids, etc.). to determine the main charac-
teristics of the surface layer on the hill, i.e., its depth, speed,
and direction over the entire surface of the hill.
D. Source Placement. The placement of the source, vertically
and horizontally, is very important, especially since the stable plume
is likely to be exceedingly thin in the vertical. Indeed, upon
further discussion, it was deemed essential to "fatten" the source
artificially. The tracer could be released through rakes of small
tubes distributed in the vertical and horizontal directions to obtain
expanded initial plume dimensions, a zo and ayo.
E. Averaging Times. It is recommended that the source be fixed
during a given test period ( three hours) and that the concentration
measurements be made (in segments not to exceed ten minutes) during
the full three-hour period. The panel felt that averaging times of
ten minutes or less were essential in order to understand the physical
mechanisms involved (e.g., so that a "plume-meander factor" can be de-
termined), but it was concerned about possible misapplications or
extrapolations from ten-minutes to three-hour averaging times.
F. Hill Shape. A hill, ellipitical in plan view with an aspect
ratio of approximately 3:1, is preferred. Early investigations of
wind distributions and smoke plume visualizations are recommended to
insure that the winds frequently blow in a direction nearly perpen-
dicular to the major axis of the ellipse.
G. Synoptic Data. While synoptic data are believed to be un-
necessary for the conduct of the small scale experiments, they may
become valuable in specifying model input data or in transferring
the results to other terrain features. We therefore recommend that
they be obtained at the time of the tests and included in the data
package.
4.2 Large Scale
The basic types of meteorological and concentration measurements
required for the small scale experiment are also required for the
large scale experiment. The criteria, with respect to the terrain
geometry and the expected frequency of stable flow in the direction
of the terrain, are in principle the same for both scales. The large
scale experiment offers a choice of sources, a real source (power-
plant stack plume) or a simulated source (tracer). The use of a real
source has the advantage of incorporating source characteristics
that must be ultimately included in model applications, such as
plume buoyancy. On the other hand, plume buoyancy is not directly
67
-------
related to the impingement process, except as it relates to the
effective source height and the dimensions of the experiment. There-
fore, the disadvantages of using a real source are that the source
strength, effective source height, and source location with respect to
terrain cannot be controlled.
The use of a tracer allows a greater degree of control and flexi-
bility and may also be desirable for other reasons (e.g., simplifica-
tion of concentration measurements). The tracer source is therefore
our recommendation.
The large scale experiment also introduces complexities, with
respect to the structure of the flow in the stable layer and the
boundary layer, that are not present in the small scale experiment.
Characterization of these flow patterns may therefore be more
difficult.
Without considering the constraints that may be imposed by
logistical, budgetary, and measurement factors, the large scale exper-
iment probably should not be carried out until the small scale experi-
ment has been completed. This would allow for testing of the basic
concepts involved in the terrain impingement process and the develop-
ment of parameterization criteria.
It is recommended that the meteorological data be collected and
summarized (hourly) for at least a one-year period at the site.
Again, the terrain features of the large scale study should be
selected so that there is a reasonable frequency with which the plume
centerline will impinge directly on the terrain feature. This may
postpone experiments on very large two-dimensinal terrain features
(ridges) until comprehensive studies of large hills are completed.
5. EVALUATION
It is most important that the sufficient funds be allocated to
insure that the experimental work and the model development are
evaluated fully after they are completed.
5.1 Matrix of Tests
A matrix of meteorological, source, and receptor conditions
should be chosen to be sure that the processes are described adequate-
ly. For example, in the small scale experiment, one might decide upon
a minimum of three wind speed groups, three delta-temperature classes,
three source heights, and two approach angles.
68
-------
5.2 Repetitive Tests
Within each box of the matrix one would want more than one field
test in order to establish confidence limits of the result. It is our
opinion that approximately six to ten tests in each box would be
desirable.
5.3 Data Retrieval
It would be very helpful if most of the concentration data could
be reduced and carefully inspected within 24 hours. This would permit
a quick review and increase the efficiency of the test program. How-
ever, the key meteorological data and selected portions of the concen-
tration data should be available in real time to facilitate handling
of the individual experiments.
5.4 Statistical Analyses
We recommend that EPA define a minimum set of statistical analy-
ses needed to establish the adequacy of the model performance.
5.5 Sensitivity Analysis
The model resulting from the study should be subjected to a
sensitivity analysis to insure that it will produce reasonable results
over the full range of intended applications.
5.6 Comparison with Existing Field Data
While we recognize that there are few sets of field data suitable
for comparison with the new (altered) model, a serious attempt to
match model predictions with such data should be attempted.
5.7 Independent Evaluation
Because of the complexity of both the problem and the experi-
ments, we recommend that EPA arrange for an independent, concurrent
review and evaluation of the program. We do not mean quality
assurance of the data, for that is a separate problem, but rather,
review of the overall technical quality of the program.
5.8 Data
In conjunction with (5.7) it is most important that the entire
set of field data be taken, reduced, and described in such a way as
to permit independent evaluation.
69
-------
6. APPLICATIONS
The contractor should prepare a technical document which fully
describes the experiment, the data, and the model development. In
the interests of assuring proper use of any model, it is also neces-
sary to list limitations, special conditions, and confidence bands
for those using the model.
Because this study is designed specifically to define plume
impingement on elevated terrain, the results may not be applicable
or transferable to all complex terrain situations. Therefore, the
degree of flexibility in the model application should be described.
Flexibility in the model use is dependent on the skill of the user
in relating the measured flow patterns to the wind tunnel and tank
studies and in extrapolating the results to other source-terrain
configurations.
Potential limitations are:
A. Near-field application, approximately one hour travel time.
B. A source-terrain conf igurataion similar to that upon which
the model is based.
C. Restriction to those meteorological conditions which existed
during the tests.
The input data developed by the user must be compatible with the
model requirements. These data cannot be specified until the final
stages of model documentation, but it seems clear that certain
measurements will be required regardless of the type of model that
evolves. These are:
A. Onsite measurements of wind speed, direction, and stability
in the layers that affect plume transport and dispersion.
These data must be consistent with the averaging time of
the model.
B. Necessary meteorological and source input data which will
permit the user to convert from the model averaging period
to longer-term averages for direct comparison with regulatory
requirements.
C. Emission parameters.
D. Definition of the terrain.
70
-------
The funding required for even the "minimal" requirements cited
above will be beyond the financial capabilities of some applicants.
Such applicants should be guided as to the acceptability of meteoro-
logical data from beyond the environs of the pollutant source. A
first step toward such guidelines could evolve from a study of the
applicability of the experimental data obtained at various sites
during the field tests. The support document for such guidelines
should include discussion of the degradation of model results in
terms of the degradation of the input data.
7. SUGGESTED ADDITIONAL STUDIES AND ISSUES
7.1 Non-Impingement Maximums
The maximum three-hour concentrations in complex terrain often
are not associi ted with the direct impingement of stable plumes on
high terrain obstacles. Rather, the maximum values are frequently
found during transitional periods when the stable plumes are brought
close to valley walls by thermally-induced circulations generated by
unequal heating and cooling of major terrain features. Such phenome-
non may occur, for example, when the western wall of a valley is
heated by the sun in the early morning while the eastern wall is still
in the shadow; a cross-valley circulation develops, causing a transi-
tory fumigation situation.
Another situation which should not be ignored is the extremely
persistent moderate to strong wind neutral condition. This situation
may be more important than the stable case in producing 24-hour
maximum values in some configurations.
We recommend that EPA investigate these phenomena, using the
experimental data generated in this project, as well as other
available records.
7.2 Relation to National Air Quality Standards
Even after completion of these studies, few of our panel members
expect the predictions of the highest second-highest concentrations at
locations different from the test sites to be accurate within a factor
of two. Factors of three to ten are more likely.
An important reason for this large uncertainty factor is that
the meteorologists are being asked to predict an extreme value in
the probability distribution of concentrations. As Lamb points out
in his written comments on this program, the highest second-highest
concentration is extremely difficult to predict. Grey, in his letter
71
-------
to Hovind dated June 22, 1979, also recognizes this problem, but
neither he nor Lamb suggests one obvious solution; the meteorologists
could very well insist that the quantity which they are asked to pre-
dict be a reasonable one.
The highest second-highest concentration is an administrative
choice reflecting the idea that there are certain maximum concentra-
tion levels below which health and welfare effects are insignificant.
These same maximums could easily be achieved in most cases by relat-
ing them to a reference concentration in the probability distribution
nhich occurs much more frequently than the highest second-highest
value. A more predictable quantity would make our regulatory de-
cisions less capricious. It makes little sense to insist that the
emission limitation for an industrial plant be twice as stringent
if based on one set of five-year meteorological records than it
would be if based on a different five-year set. As one member of
our panel asked: "If the predicted 3-hour highest second-highest
S02 concentration for a given plant is 1,420 g/m (-10% over the
1,300 g/m standard) and the accuracy of the prediction cannot be
specified to better than a factor of five, should we deny him a
permit, or require 50 million dollars worth of control equipment?"
The field data needed to readjust the definition of the stand-
ards, without changing the desired maximum levels, exists. Ju-
dicious use of it would permit us to employ a much more orderly
system, based on more predictable values. This does not imply
abandoning the intent of the existing standards, such as allowing
only rare exceedances of 1,300 and 365 g/m SCX, concentrations for
the 3-hour and 24-hour averaging periods, but rather, achieving them
in a better way.
This question is, in a sense, beyond the scope of this Workshop,
but it is not beyond the consideration of the accuracy of complex
terrain (or any other) model.
72
-------
3. EXPERIMENTAL DESIGN PANEL
PANEL LEADER: Gene Start
PARTICIPANTS:
Shep Burton Frank Schiermeier
Warren Johnson Tim Spangler
Ron Lantz Dean Wilson
G. DeMarrais Mike Williams
Fran Pooler William Wilson
1. EXPERIMENTAL DESIGN
1.1 Introduction
A goal of this Workshop is the identification and selection of an
important and tractable aspect of the behavior of airborne effluents
within the setting of complex terrain. The primary focal point
selected for investigation is the concept of plume impaction upon
elevated terrain. Plume impaction is the behavior of a plume as it
encounters the terrain. Emphasis is directed toward understanding
this phenomenon for plumes which are emitted well above the ground and
which interact with higher terrain at downwind locations. Uncertain-
ties exist regarding which phenomena are significantly contributing
to the plume/terrain interaction and what magnitude of effect will
result from the action of these phenomena. There are several possible
features of the topographical setting and on-going physical processes
which may be important contributors to the cumulative behavior which
is termed "impaction". Certain broad categories of terrain and
meteorology may be considered and the experimental focus may be
selectively reduced to include a subset of them. This subset may be
examined more thoroughly than the full set. The qualitative and
quantitative understanding developed from the subset would provide
sound guidance for understanding these selected conditions and perhaps
guide related modeling for the remaining situations not investigated.
It is desirable to investigate impaction as a function of terrain
types. Four types of terrain have been identified and ranked in order
73
-------
of expected complexity. They are 1) isolated hill, 2) ridge line,
3) broad valley, and 4) canyon. The isolated hill appears to be the
simplest, most easily understood topography and the study of it is
recommended to be first. It also is the easiest to compare to con-
current wind tunnel and water tank simulations. Subsequent studies
should consider the other topography once understanding (some of which
should be transferable) is developed for hills.
Impaction needs to be studied as a function of meteorology.
Three atmospheric stability situations are considered most important;
stable, neutral, and transitional from stable to neutral conditions.
Unstable stabilities should also be investigated, but stable flow is
the selected focus of this Workshop due to its greater importance in
current modeling evaluations. Neutral cases also have significant
potential for impaction of high concentrations on elevated terrain.
Such impacts have been observed and documented.
The transitional period from stable to neutral flow is included
because it presents a potential for mixing very high concentrations to
the surface of elevated terrain. Studies of plumes originating during
stable conditions, with airflows over elevated terrain, may include
impaction during transitional periods (either a diurnal change or a
turbulent episode) and thereby provide insight on other features of
the impaction process.
The evaluation of plume impaction upon complex terrain must in-
clude many considerations in order to achieve probable success. The
experiment must include surface and upwind aerial measurements of
tracer concentrations. Central to the experimental design is the need
to discern how the tracer concentrations measured on the terrain
slopes compare to the aerial tracer concentration which would have
existed at the point in space were there no topographic features. The
essential ideas are summarized in Figure 5. Several observations of
plume centerline concentration must be collected at distances upwind
of the terrain in order to unambiguously specify the magnitude of
tracer concentration and the rate of change of concentration with
downwind distance. These observations must yield an extrapolation
(or projection) of expected aerial concentration for the distance of
the elevated terrain. A corresponding observation of tracer concen-
tration on the slope is measured. The ratio of observed to expected
(extrapolated) concentration is termed the impaction ratio. If the
ratio equals one, the terrain obstacle presents no detectable altera-
tion of plume concentration and complete or full-value plume impaction
has occured; if the ratio equals 0.1 the impacting plume concentration
is only 1/10 of expectations and some diluting or resisting mechanism
has reduced the magnitude of the plume impaction. It is very
74
-------
important that the same averaging times be used for all tracer
concentration prior to ratioing them.
c
o
§
o
o
Slope measured concentration
Measured concentrations
H Extrapolated concentrations
= Distance of topographic feature.
Distance
Figure 5. Estimation of ambient plume behavior and the degree of
impact ion upon elevated terrain.
If the ratio differs significantly from unity, then more detailed
measurements of magnitudes and profiles of micrometeorological and
tracer concentrations would be useful to understand the near terrain
slope processes which alter these plume impactions. Figure 6 sche-
matically depicts a possible observed near-slope profile of tracer
concentration and an extrapolated profile of tracer based upon upwind
measurements. A divergence of these profiles from one another is de-
picted. Two hypotheses for the differences in concentration profiles
are the following. The average observed tracer concentration values
(and therefore the profiles) are less than expected because of topo-
graphically induced transport and/or diffusion effects. These time
integrated (averaged) concentrations could be less, due to the tracer
samples being collected during alternating periods with full-value
and reduced-value (or even tracer free air) concentrations. Under
this situation, the tracer concentrations in time integrated samples
collected on the slopes would be less than the extrapolated tracer
concentrations; but the area of coverage on the slopes could be larger
than expected from extrapolations of the plume cross-sectional area
75
-------
determined at upwind points of observation if enhanced meandering of
the plume centerline occurs near the topography. Another manner in
which smaller concentrations might be sampled on the terrain could be
the result of the slope boundary layer (when it exists). The micro-
meteorological processes operative within the slope boundary layer may
either produce additional tracer dilution or impede the transfer of
plume mass through the layer. In either case the time-integrated
tracer concentration would be reduced in the terrain samples. In all
likelihood, the magnitudes of plume impactions upon elevated terrain
are the result of all of the above processes acting together. The
dominance of one or more phenomena (or the subordination of others)
probably accounts for the widely differing observed plume impaction
and postulated behaviors. The meteorological and topographic factors
and their magnitudes, which lead to the dominance or subordination of
certain phenomena related to wind direction meandering, turbulence,
and diffusion, are key relationships to investigate.
c
o
+5
2
o
o
— Q Extrapolated concentrations
• Near slope concentrations
Xj= Distance of topographic feature
Distance
Figure 6. Schematic depiction of near terrain profiles of tracer
concentration.
To concurrently evaluate and improve the performance of models,
many factors must be measured. Transport of the airborne plume may be
76
-------
a highly important factor and it is recommended that winds be deter-
mined in the form of a multilevel field of u, v, w wind speed com-
ponents. The dispersive power of the atmosphere should be parameter-
ized for each test through the measurements of turbulence, ambient
plume diffusion and atmospheric stability. Parameters describing
the tracer sources, such as its strength, height and horizontal
location must also be documented. An effort to establish the energy
budget (e.g., various turbulent fluxes) might also be included in the
experimental measurements.
The conduct of the experimental study may be significantly
affected by practical constraints. The most important constraint
could be the funding limits. The design and budget should be care-
fully prepared to ensure that the most imporant parameters are being
included and that the experiment isn't too limited in scope to provide
sufficient information to obtain meaningful results. The design also
must consider the availability of potential sites which satisfy the
experimental requirements. One type of idealized measurement site
would have a power plant about 5 km upwind of an ellipsoidal hill
which extends higher than the effective plume height; transport of
the plume, during stable flow, would occur toward the hill 365 days
per year. It is doubtful that such a site exists. Therefore, the
experimental design must consider availability of sites early in the
design process.
The experimental design must consider the timeliness and the
applicability of the results. Some findings (significant information)
should be provided after a reasonably short period of time in order to
demonstrate the desirability of continued funding of this program.
The information acquired must be applicable to the real world and have
direct relevance to the engineering design and regulatory problems
of plant settings within complex terrain. The selection of a "full-
scale" setting (typical of a number of operating fossil-fueled power
plants and/or smelter plants) should be considered.
A real concern is the risk of developing erroneous conclusions
due to an inadvertent (and unknowing) subordination of certain
physical processes. These physical processes, because of the selec-
tion of smaller than typical settings and topographic features, may
have an inordinately small relative contribution to the overall com-
bined effect by the full spectrum of transport and diffusion mecha-
nisms. If the relative contribution of a phenomenon to the transport-
ing and/or diffusion of an airborne effluent is scale dependent, then
conclusions based upon observations at a small hill may not translate
to conclusions for a large hill, etc. For example, the presence or
absence and size or magnitude of the effects of slope flows and
77
-------
slope boundary layers are believed to depend upon terrain obstacle
size. The ultimate transfer of airborne material to the terrain
obstacle surface (plume impaction) can be significantly altered by the
strength and size of slope boundary layers and flows. If a choice of
scale for the field study excludes or subordinates these phenomena,
there is a distinct risk that inapplicable conclusions (and formu-
lated models) would be applied to larger settings in which these
phenomena would contribute significantly. On the other hand, in-
vestigations of a specific setting will increase understanding of
that scale; a second investigation of a substantially different scale
will provide additional understanding. There may be partial agree-
ment between study findings and some distinctly different and perhaps
conflicting findings for studies of very different size of topographic
features.
Some other constraints will be in the areas of instrumentation
and expertise to handle the job. The measurement and design panels
jointly attempt to avoid the development of a plan for which no ade-
quate and proven instrumentation or methodology exists, or whose im-
plementation would be prohibitively expensive. Similarly, the
availability of qualified contractors must be considered when
establishing a time table for the experiment. (They may already be
committed to another activity).
From an advancement of science perspective, it seems desirable
to study the simple geometry and setting to gain a clear understanding
of the observed plume behaviors. This understanding may be a trans-
ferable module which is incorporated (largely intact) into either more
generalized models to be developed or used as a module for adjusting
some of the existing models to plume transport, diffusion, and impact.
Once an initial understanding is achieved, the more difficult and
diverse settings and processes may be examined through module by
module systematic investigations, as necessary.
The advantages and disadvantages of three scales of complex
terrain plume studies were considered. The scales are (1) small
(50 to 100 m high) hill study, (2) power plant scale plume study,
and (3) large scale (400-600 m high) hill study. Preliminary
scenarios for scales (1) and (2) are contained in Appendix A. A
summary of strengths and weaknesses for each scale of study follows.
3.1.1 Small Scale (50 to 100 m) Hill Study
• Advantages. Source mobility is a decided advantage
over a fixed source in guaranteeing plume impaction of the hill under
78
-------
study. The logistics of conducting such a small scale tracer study
are decidedly easier than either of the two larger scale studies.
Because of the small scale, this study has a closer relationship
to laboratory or physical plume modeling. The relatively low plume
heights would permit experiments to be conducted in the most stable
regimes which occur close to the ground surface.
* Disadvantages. The small scale of this study poses
several measurement problems (for example, plumes are of small size
and travel distances are relatively short) because measurements of
particulate plumes by lidar would be approaching or decreasing below
the lower limits of resolution. To alleviate this problem it has been
proposed to develop an initial volumetric source or to relocate the
source farther upwind. This tradeoff may permit improved lidar reso-
lution, but would increase the difficulty of aiming (impacting) the
plume on the small hill target; the releasing of tracer with a virtual
size would also be more difficult.
Even if lidar measurements of the particulate (e.g., oil fog)
plume were made possible by this adjustment, the determination of
absolute concentrations of plume tracer remains a difficult problem.
Airborne detection systems for tracer concentrations would be in-
effective in measuring axial values due to the small vertical extent
of the plume. Also, aerial sampling may significantly affect the
small plumes by the additional turbulence due to the aircraft.
Balloon-borne measurement packages may yield vertical plume dimen-
sions but appear to be unworkable due to a likelihood of horizontal
positioning errors.
The preliminary proposal (Snyder, Appendix A) using five 150 m
instrumented towers, is believed to be too costly. When these costs
are combined with other direct study expenditures, they are estimated
to about 75% of the $3.3 million tentatively allotted for the field
experiments. A subsequent proposal to reduce the tower requirement to
one 100 m tower eliminates the cost problem, but significantly
cripples the information content of the small scale study. To sample
the low-level wind fields near the obstruction, it may be necessary to
revert to low resolution pibals since remote sensors such as acoustic
dopplers cannot measure within the initial 30 to 60 m above the
ground-surface.
Finally, the appropriateness of the scaling of the small scale
study to the real world situation can be a problem. Phenomena such as
wind shears, strengths of the boundary layer, and the size of
mechanical/convective eddies are necessarily different between the
79
-------
proposed small scale and the larger scale in which power plant plumes
disperse.
* Conclusion. The panel feels that the small scale hill
study consists of many valuable experimental concepts, but that the
scale is probably too small to be workable, clearly representative, or
of immediate applicability. Therefore, these concepts have been em-
bodied into a large scale hill study.
3.1.2 Power Plant Scale Plume Study
• Conclusion. Without describing the advantages and
disadvantages, this option was passed over because of its dependence
on power plant emissions from a fixed source location. The occurrence
of useful test days in which the wind transport was acceptable might
be rare. The funding for the field experiments is too limited to
allow success to be dependent on the vagaries of transport wind direc-
tion and plant operational characteristics. Ideally, this scale of
field study would reflect the real world, but it is not affordable
at the anticipated funding level and no suitably situated plant has
yet been identified.
3.1.3 Large Scale (about 400-600 m) Hill Study
• Advantages. The basic premise of this tracer study
is to incorporate the source mobility of the small scale hill study,
but project it toward a larger terrain obstacle in an attempt to fa-
cilitate supporting measurements. For example, the upwind fetch is
sufficient to permit definitive transport and diffusion measurements
of the combined tracer/particulate plume by airborne lidar and tracer-
sampling aircraft. Similarly, the plume elevation is sufficient to
permit definitive measurements of the plume structure, the boundary
layer and the low-level wind fields by remote sensing techniques such
as ground-based lidar and acoustic doppler sytems.
More important is the closeness of this scale to real-world
situations and the minimizing of concern regarding scale subordina-
tion of phenomena affected by wind shear, boundary layer existence and
structure, and turbulent diffusivity. This resemblance to actual
power plant plume situations and dispersion provides for more imme-
diate applicability of results, and an immediate credibility to
non-technical and non-meteorologist recipients.
80
-------
• Disadvantages. Because of the increased scale, the
large scale hill study necessarily entails more logistical diffi-
culties in some ways (but possibly less in others). Although achiev-
able, this study would require more personnel, aircraft platforms,
remote sensors, and increased source strength. These additions not
withstanding, the large-scale hill study (field experiment) would
cost about half as much to the same as the costs for the small scale
hill study (as originally proposed).
• Conclusions. It is recommended that at least two
large-scale hill studies be conducted in the complex terrain setting.
For reasons cited above, this approach is deemed most cost effective,
more acceptable and representative, and would produce results of
immediate applicability. The conduct of the necessary measurements
appears feasible with existing measurement technology.
The small-scale hill study is believed to exceed (or extend to
the very limits) the resolution capabilities of most remote and air-
borne measurement systems. Fixed in-situ measurement techniques are
possible but the costs of these measurement arrays to define the
meteorology and the plume diffusion adequately are believed to exceed
greatly the costs for the large hill measurement configurations.
3.2 Design Concepts for Plume/Terrain Interaction Study
To examine the interaction between airborne effluent plumes and
elevated terrain near effective plume height, a setting is desired
with the height of the topography comparable to or greater than typi-
cal plume heights associated with fossil fueled power plants during
stable conditions in the Western United States (to avoid controversy
regarding scale size). The experimental design will be discussed from
the perspective of the 400-600 m hill.
The work required for this study can be conveniently separated
into components associated with site selection, frequency and duration
of experiments, and the measurements which must be made in character-
istic zones of pollutant behavior.
3.2.1 Site Survey and Selection. An initial task will be
selection of an appropriate site or sites which optimize the chance of
useful information within budget, time, and design constraints.
The design constraints include a 400 to 600 m hill reasonably
isolated from major terrain influences by fetches of relatively
flat land. In this context, releases may be made at distances of
up to 5-10 km from the base of the hill; tracer releases should
31
-------
be initiated at locations free of wake effects from other terrain
roughness influences.
The hill should also be relatively symmetrical (circular to
elliptical). For circular symmetry, releases may be made from a number
of directions without changing the cross-sectional shape of the
terrain; for elliptical symmetry the contribution of aspect ratio
might be considered. The intent of the circular symmetry constraints
is to provide many similar stable flow experiments with features as
consistent as possible during study periods. Some of these constraints
on symmetry of the hill could be relaxed if certain wind direc-
tions are precluded because of inappropriate upwind terrain. Also, if
crosswind shape variation does not significantly detract from the
measurements or does not diminish the number of acceptable testing
days which occur, or if aspect ratio does not significantly detract
from the measurements or does not diminish the number of acceptable
testing days which occur, or if aspect ratio variations are desired,
hill symmetry becomes less important.
An important consideration, although not essential, would be the
possibility that a given site (hill) may be situated within a region
of channeled flow so that the frequency of acceptable testing condi-
tions is greatly increased along certain preferred wind directions.
Then the logistics could be greatly simplified because appropriate
source release points and fixed measurement sites would be much easier
to locate, and more permanent.
3.2.2 Frequency and Duration of Experiments. An important re-
quirement of the site is a high frequency of stable flow with depths
up to and exceeding the height of the hill.
A goal is to achieve 50 stable flow experiments (total) during
conditions of stable flow over two periods of testing. Each experiment
should include at least 3 hours of tracer releases and measurements,
and 6 hours is suggested as a usual duration. (Impaction contributions
to the 3 hour air quality standard are to be considered). The experi-
mental periods would cover two or more seasons of the year during
which stable conditions are expected. A relatively large number of
experiments is required in order to provide information on the details
of plume interaction with high terrain considering variations of wind
speed, temperature gradient, height of release, and time of day. In
the case of time of day, nighttime and morning cases may be of special
interest. Morning cases are important. Field experience indicates
that the occurrence of high concentrations upon elevated terrain is
82
-------
frequently associated with plumes established during stable condi-
tions which are transported toward obstacles during daytime condi-
tions. Nighttime cases are representative of the highest frequency of
stable conditions. In order for a site to be acceptable, an examina-
tion of the local climatology should suggest an occurrence of a large
number of days with acceptable stable flow (exceed 30) over an ex-
perimental season or period (in which about 20 to 25 tests are
desired).
Once an individual experiment has been initiated (for an expected
6 hours of continued measurements), there is no assurance that the
atmospheric stability and flow condition will persist throughout the
experiment. In this event, the experiment should, in general, continue
since important transient behaviors and understanding of peak to
average concentration fluctuations may be documented.
3.2.3 Characteristic Zones. It is convenient to describe the
experimental measurements from a concept of charateristic zones within
which certain activities and changes in turbulent diffusion and plume
transport are expected. Four zones are identified and are shown in
Figure 7. Zone 1 contains the release of the tracer plume (and its
plume rise, stabilizations, etc., if applicable). Zone 2 is the trans-
port zone. Within Zone 2 the tracer plume is transported and diffused
by usual atmospheric conditions (those conditions not incorporating
topographical distortions of flow streamlines or alterations of turbu-
lence). The "ambient" behavior of the plume is established in this
zone. Zone 3 contains the terrain feature of interest. This terrain
feature is instrumented to measure possible plume impaction upon the
slopes; airflow distortions, changes in atmospheric turbulence, and
differing rates of diffusion developed here. Zone 4 is the region
"downwind" of the terrain feature. Within the wake zone the tracer
plume continues to be transported and diffused in manners which differ
from tne ambient conditions of Zone 2. However, at longer distances
downwind, Zone 4 transport and diffusion characteristics converge to-
ward the Zone 2 ambient conditions. Each of the characteristic zones
and the measurements to be performed are described in greater detail
in following paragraphs.
83
-------
ZONE
Zone 1. Tracer source (plume stabilization, etc. as appropriate).
Zone 2. "Ambient" transport and diffusion during approach to elevated terrain.
Zone 3. Plume interaction/impection with elevated terrain.
Zone 4. Downwind wake transport/diffusion/impact and return toward ambient conditions.
Figure 7. Conceptual depiction of characteristic zones of tracer
plume transport and diffusion.
• Source Zone (1). The principal activities required in
the source zone are the release of gaseous tracer and a particulate
tracer. The particulate tracer should be suitable for providing flow
visualization (probably via airborne lidar) at distances beyond 5 to
10 kilometers. The release platform for the particulate and gaseous
tracers must be semi-mobile, with capacity for six to seven hour dura-
tion of tracer release. The source height should be adjustable so the
plume height can be varied from about half-hill height to a modest
amount higher than the hill top. The horizontal location of the
source should be semi-mobile (unless the ideal site was found) so
that it may be easily positioned directly upwind of the hill before
the experiment is begun. Once the experiment is begun the source will
be fixed for the full test period (6 hours). The effective initial
size of the plume shall be on the order of 30 to 50 m in the vertical
and 75 m in width. These minimum dimensions ensure that known plume
sampling methodology may be effectively utilized to define plume
spread statistics, concentrations, and spatial locations within Zone
2 prior to interaction with the elevated terrain. Particulate tracer
and gaseous tracer plumes must be coincident downwind of the release
point(s).
84
-------
The use of multiple release points for pairs of separately
identifiable particulate and gaseous tracers could provide greater
assurance that a successful test was achieved during each attempt.
A method of distinguishing between the various particulate releases
might be required so that flow visualization advantages could still
be retained.
• Transport Zone (2). This zone of meteorological trans-
port and turbulent diffusion is of nearly equal importance to the
actual high terrain zone of plume influence (Zone 3). Within this
zone the atmospheric rate of diffusion, which determines the upper
limit magnitude of plume concentration which potentially may reach
the slopes, is operative. Since neither this rate of diffusion nor
the absolute concentration can be predicted with precision, the con-
centration at a number of successive downwind distances must be
measured; uncertainties in the absolute concentrations will translate
directly into uncertainties in the degree of impaction of plume
centerline concentration.
Documentation of plume transport winds and the observed tra-
jectories will yield insight regarding peak to average tracer concen-
tration statistics and the frequency of occurrence of transport to the
terrain object in question. It is very important that the measure-
ments in Zone 2 are sufficiently numerous so that the details of plume
dilution can be defined before the terrain begins to perturb the flow
and plume concentration. Otherwise, it will be impossible to estabish
what additional dilution results from terrain interactions. In addi-
tion, measurements within Zone 2 will be related to long-term on-site
measurements in order to define the frequency of terrain interaction
in a climatological sense.
There are two major types of information to be observed in Zone
2: (1) plume characteristics and (2) meteorological characteristics.
Plume characteristics will be defined in two ways. First, the spatial
distribution of the particulate tracer will be examined by airborne
and ground-based lidars. The airborne lidar will provide cross-
sections of the plume at various distances downwind from the source.
These measurements will extend into Zone 3, and possibly farther
downwind. The airborne lidar system would provide plume transport
depictions and structure in this zone and in Zones 3 and 4 as well.
The ground based lidar will provide multiple planes of observation of
the plume in a nearly instantaneous sense. A pooling of these obser-
vations will yield averaged time histories of concentration (relative)
and plume spreading statistics. A minimum of two, and preferably 3 to
5 planes (lateral-vertical) of observation are needed.
85
-------
In addition to the particulate tracer observations, an airborne
system must sample the nearly instantaneous plume centerline maximum
concentration of the gaseous tracer by traverses through the plume.
These samples must be at locations which correspond to the cross-
sections of particulate tracer observations so that the particulate
plume (which may be averaged in time) becomes "calibrated" to the
gaseous tracer concentrations.
For meteorological characteristics, three major measurement ob-
jectives are determination of the wind field (u, v, w) at various
heights, determination of the vertical stability, and measurement of
the atmospheric turbulence at various points upwind of the terrain
obstacle. The wind field could be determined by a doppler acoustic
sounder, or by a lidar device, or by tethersonde measurements. The
acoustic sounder might have to be somewhat mobile to be locatable into
the approximate upwind sector before the beginning of each day of
experiments. Vertical stability could be obtained from the tempera-
ture profiles provided by an aircraft, tethersonde, or minisonde.
In addition, a specially instrumented aircraft would provide detailed
measurements of turbulence at various points.
All of these aircraft would also make measurements in Zone 3
approximately near the terrain, and possibly in Zone 4. The doppler
acoustic sounder could continue to make measurements over the pro-
gram study period in order to define the frequency of occurrence and
duration of plume transport conditions believed favorable for terrain
interaction.
• Interaction Zone (3). Zone 3 encompasses that area
starting upwind of the hill where significant distortions of flow
streamlines are thought to begin. It extends downwind from the ob-
stacle an imprecise distance (perhaps at the downwind extent of the
base of the hill or to a few height increments farther). Zone 3 then
indistinctly blends into Zone 4.
The plume effluents transported through Zone 2 toward the terrain
may experience a number of possible special phenomena. The end result
of the joint action of these phenomena may be postulated to be any-
thing, varying from no plume mass impaction upon the elevated terrain,
complete impaction of the maximum possible value of plume concentra-
tion, or something in between these extremes. The problem is further
complicated by the need to know the temporal behavior of what is
little understood in the first place. The flow distortions effects
within the interaction zone must be examined to parameterize their
contributions to the variability of concentrations which may be ob-
served at the terrain surface. These variabilities should be related
86
-------
to meteorological conditions. The resistance (or conductivity) of the
slope boundary layer to mass fluxes of airborne (Zone 2) effluent to
the slope surface must also be investigated; it probably is scale
dependent.
Plume Characterizations. The transport and diffusion
characteristics in the slope boundary layer may be critical to con-
siderations (and understanding) of the plume impaction process.
Requirements for plume measurements in this setting (all of Zone 3)
include measurements of maximum tracer concentrations reaching within
1 or 2 meters of the terrain surface (sequential 60 min. average
values at about 100 locations, plus sequential 10 min. averages at
about 10 of these locations). Several (perhaps 5 to 10) short towers
(on the order ox 10 meters) could be placed at selected sampling posi-
tions to describe the change of tracer concentration with height. As
an initial estimate of the spacing of samplers upon the elevated ter-
rain, the following type of guideline is presented.
a) Horizontal spacing: determined from expected plume width
parameter at the downwind end of Zone 2, e.g., about 0.5
to 0.8 ov .
b) Vertical spacing: determine from expected plume thickness
parameter, in a manner corresponding to ay , about 0.5 to
0.8 ' az.
c) Samples need not cover the entire hill. A band or capping
array could be used, as appropriate to the controlled height
of plume relative to the terrain.
Up- and downwind sides of the hill should be instrumented. Near-
surface profiles of concentrations should be obtained from 2 or more
tracer samplers located on the few short towers installed on the
slopes. The cumulative sampling period for gaseous tracer should be
consistent with the postulated six hour tracer release durations.
Several 3 hour averages may be calculated from the six one hour se-
quential samplings within each individual 6 hour test.
Meteorological Characterizations. Meteorological
measurements in Zone 3 are needed to define the wind field at several
heights and positions about and above the hill. In a similar manner,
temperatures should be measured to specify the thermal stability.
A tower should be located on top of the hill. Measurements of
the gaseous tracer concentration profile and 3 or more levels of u, v,
w, and temperature, plus turbulence parameters, should be collected
87
-------
during the experiments. Wind and temperature sensors (at 2 levels)
should be placed and operated on several of the short (10 m) towers
located upon the terrain slopes. The acquisition of turbulence data
at these towers on the slopes could provide additional information of
importance. If Gaussian formulations are unsuccessful, profile data
might assist in the development of gradient or K-theory models of
turbulent diffusive transfer within the slope boundary layers.
Some portion of the field site meteorological data should be
available to be examined in near-real time (e.g. , 2 minute averages
updated every minute or two) to assist in quality assurance and guid-
ance of the test preparations and conduct.
• Wake Zone (4). The zone downwind of the terrain ob-
stacle tends to lie outside the primary area of interest. However, if
several indirect and mobile sensing systems are utilized in the con-
duct of the experiment, or if atmospheric conditions change the
behavior of the plume and local meteorological parameters, many
potentially interesting and variable observations of opportunity could
be easily collected. These data would either supplement on-going
programmed measurements or substitute for those which would not be
taken due to a departure from test window conditions. Measurements in
this zone would be similar to the data collections in Zone 2. Air-
borne lidar observations of the particulate plume would be of signifi-
cant value to show trajectories and wake diffusion downwind of the
hill. Aircraft turbulence, wind, and temperature measurements could
greatly increase the understanding of transport and diffusion in this
zone (at least relative to the upwind Zones 1, 2, and 3).
3.2.4. Long Term On-Site Documentation. For the complex
terrain situation, it is desirable to have an understanding of
climatology of flows, stabilities, and perhaps other parameters
that control whether an elevated plume will cause high ground-level
concentration on a hillside. It is necessary to determine, for
various time averaging periods, the frequency at which concentrations
occur within twenty-four hour periods and for periods extending out
to the entire year (second-highest concept).
It is also important that we know what other kind of terrain
induced flow and turbulence phenomena might be occurring in the
area, but which were not observed during the study period.
Therefore, it is important to continue some of the more routine
measurements at the study site for some additional time, perhaps
a year, before and/or after the conclusion of the intensive field
88
-------
experiment(s). The most important parameters to continue measuring
may be winds, as a function of height, and the vertical stability.
An acoustic sounder system(s) might fulfill these needs for wind
and vertical extent of turbulent mixing or its layered structure.
3.2.5 Additional Experimental Considerations. Prior to final
experimental design and the physical setup of a field measurement site
(and perhaps before final selection of that site) several screening
studies could be performed. These screening studies should suggest
appropriate sampling resolution (temporal and spatial) and some of the
more important physical phenomena to be quantitatively measured. The
following examples of screening studies may be valuable before de-
velopment of the final field study design. Monitoring data from sites
surrounding sources of opportunity should be screened. If the tem-
poral resolution of the plume(s) of opportunity is on the order of
minutes to an hour or two, meaningful guidance may be gained regarding
peak to average effluent concentrations with some implications for
controlled tracer sampling equipment.
Turbulence measurements previously collected within complex ter-
rain settings should be reviewed for insight about expected phenomena
and their aproximate quantitative values. Recently, turbulence
measurements have been collected by researchers to describe the air-
flow environment for wind energy systems which might be located on the
top of ridges and hills. These data should be studied.
A number of preliminary visual tracer studies should be conducted
on and around the candidate field study sites. Some knowledge of the
size and strength of terrain slope boundary layers and airflows may be
gained. In this way, the bias due to selection of a particular size
of terrain feature (scale related subordination of some phenomena) may
be quantitatively revealed.
Finally, physical modeling of the selected site may provide
valuable insight for the final experimental layout of measurement
devices. Focal points for measurements (characteristic spatial loca-
tions on and about the hill) may also be suggested.
Another approach to the conduct of this plume impaction study is
the separation of the investigations into two component parts, each of
which is conducted somewhat separately. (To some extent, the small
hill (50 to 100 m) study may be an evaluation of impaction behaviors
without the likelihood of significant influences from slope boundary
layers and flows). The phenomena which govern the impaction process
may be separated into phenomena which occur either near the terrain
(within the slope boundary layer and flows) or not near the elevated
89
-------
terrain. Investigation of plume mass transfers through the slope
boundary layer would be expected to be site and site-scale specific;
measurements for several different settings would be required with
detailed, fine-resolution measurements. These measurements would des-
cribe the micrometeorological behaviors within the slope boundary
layer; the plume mass fluxes through this layer would be analogous to
a plume deposition study (usually conducted above flat, simple sur-
faces). The second component of investigation would study the trans-
port and diffusion of airborne effluent to the point of either
impacting upon the terrain (if no significant boundary layer exists)
or to the point of increasing influence of the slope boundary layer
processes. This second component of the study relates to the site
climatology of transport and diffusion in the general vicinity of the
topography; it describes the concentration of plume effluent, the
frequency of its occurrence, and its spatial location. It is analogous
to a potential fjr impaction; the near-slope component describes the
receptivity of the slope boundary layer to a mass flux through it to
the terrain surface receptors. Both components of the study must be
addressed; this alternate consideration simply poses the idea that the
overall study might be decomposed into smaller research modules with
fewer specific focal points within each module. With more but smaller
and specific research modules, the time period for development of an
overall description (model) of plume behavior could be lengthened
unless several modules may be run concurrently with each other.
3.3 Summary and Conclusions
The understanding of impaction of plumes from tall sources
against topographical feature.s is the primary goal of this experi-
mental design. Thermally stable atmospheric conditions are the focal
point, but neutral and transitional conditions should also be under-
stood. In the first stage of experimentation, a hill of simple
geometry and relatively isolated from other elevated terrain is sug-
gested; it should have a size commensurate with real sites about power
plants in the western United States. The essence of the study is a
combination of meteorological and gaseous tracer measurements which
must satisfy the following minimum requirements (at least as viewed by
this panel) during a particular field test.
A set of tracer data points must be collected which clearly des-
cribes the magnitude and rate of dilution of the airborne effluent as
it approaches the elevated terrain. A projected (expected) concentra-
tion value must be obtained, for the position of plume impaction upon
the terrain slope, were no alterations of airflow and rate of diffu-
sion to occur before plume impaction. The observed concentration
90
-------
versus the projected concentration demonstrates the degree of impac-
tion. A null concentration observation indicates no impaction. More
subtle features, such as the concentration gradient in the boundary
layer immediately adjoining the slope, is an interesting facet of the
problem and should clarify the role of the micrometeorology within
the layer relative to the impaction problem.
Sufficient measurements of the meteorology, transport (visualiza-
tions both of the plume and slope flows), and turbulence should be
collected to document the dynamics of the problem. Routine site
monitoring measurements should cover a longer period than the inten-
sive field studies; these data may help resolve the uncertainty about
what magnitudes of plume concentration impact for longer period
averages.
Finally, pre-field test physical modeling simulations may guide a
final design of the field installation. Also plume visualization
should be used more frequently than just during full scale experi-
ments. Many specific phenomena may be qualitatively investigated (and
photographed) by smoke flows around the obstacle, and by small scale
releases within the slope boundary layer flows.
3.4 Budget Estimate
In order to estimate the general feasibility of the cost of the
field experimental design, the following approximate values are pro-
vided. A. study about a 400 to 600 m high hill is postulated. Two 40
day intensive field study periods are assumed, with 50 days of field
sampling and measurement operations. Continuous meteorological data
collection and archiving for an 18 month period are assumed. The
costs of centralized data logging equipment are omitted from this
estimate and should be included elsewhere.
3.4.1 Equipment and Other Direct Costs $ 2,147.600.
3.4.2 Labor 135 man months @ $30/hr (22 days
per montli) 712,800.
TOTAL (Labor, Equipment, and Other
Direct Costs) $ 2,860,400.
91
-------
4. MEASUREMENT TECHNIQUES PANEL
PANEL LEADER: Roy Evans
PARTICIPANTS:
John Eckert Gilbert J. Ferber
Freeman Hall Norm Huey
Rudolf Pueschel Ted Smith
Ivar Tombach
1. SCALE OF MEASUREMENT
Measurement techniques appropriate for three scales of experi-
mental design were discussed by the Measurement Techniques Panel.
The smallest scale considered was appropriate to controlled tracer
releases over a small hill, with plume travel distances of km, as
described in the scenario submitted by Bill Snyder. The next larger
scale discussed was appropriate to the concept of controlled release
over a 500-meter hill, with plume travel distances of about 10
kilometers, as described in a scenario proposed by the Experimental
Design Panel. The largest scale proposed was appropriate to impinge-
ment of plumes on elevated terrain (hills) near an active power
plant.
The chief differences among these three scales are related to
measurement resolution and plume sizes. The small "100-Meter Hill
Experiment" would involve relatively thin plumes with initial
vertical sigmas between 5 meters and 15 meters. The "500-Meter
Hill Experiment" would involve plumes with vertical sigmas of
40 to 60 meters prior to impact on the hill. The full-scale ex-
periment would also involve plumes with initial vertical sigmas
on the order of 40-60 meters; however, these would evolve over longer
travel distances (10 kilometers or more) to widths of more than 2
kilometers.
Investigation on the impingement of a plume on a hill under
stable conditions, regardless of the scale of the experiment, was
considered to have four phases:
92
-------
(1) PLUME RELEASE AND GENERATION (i.e., determining the
plume's source terms and initial conditions);
(2) PLUME EVALUATION (i.e., the growth and changes which
affect plume dimensions during plume transport);
(3) PLUME IMPINGEMENT (the actual process of transport of
pollutant material through the boundary layer); and
(4) FLOW FIELD DETERMINATION (measurement of wind speeds,
wind directions, turbulence, thermal structure and sta-
bility, of the wind field which actually transports
the plume.
Each of the three scales of experimental design involves
measurements of these different aspects of plume generation, trans-
port and evolution, and impingement on elevated terrain. How-
ever, each scale involves plume dimensions and resolution re-
quirements which determine the choice of measurement techniques.
Table 1 is a summary of techniques and their probable applica-
bilities to the experimental scales.
Discussion of these techniques is organized generally in terms
of experimental scale, beginning with the small-scale experiment
and going on to the full-scale (or Real-World) experiment. An
experimental technique is discussed under the first experimental
scale where it becomes useful.
2. MEASUREMENT TECHNIQUES FOR SMALL SCALE EXPERIMENTS
This scale of effort, as discussed by the Measurement Tech-
niques Panel, is similar to the concepts described in the Snyder
Senario for the 100-meter hill experiment. Broadly speaking,
modern technologies (lidar - instrumented aircraft, remote wind-
finding systems) are not appropriate to experiments at this scale,
and researchers must rely primarily on conventional techniques:
instrumented towers for winds and temperatures, integrated samples
(by syringe or bag) for a gaseous tracer (probably SFg), and plume
visualization by smoke release.
The panel did make some specific suggestions for the conduct
of the 100-meter hill experiment, which can be summarized as follows:
(1) Little cost would be added to the experiment by release of
two different tracer species (e.g., SFg and Freon-13) from two
93
-------
CD
r< P
CD H
O Q
rt ,5
CO
G
P
0
G
•H
-P
B
(S
CO
G
•iH
CO
g g
-P -P
O
-H JH
-P CD
CO CO rH
r< S ft
i SB'
H
O
•P
"ft
x_x
CO
-P G
s-3
S
•H
>
B
O -H
•H -P
-p -rl
0
3
G
•S
P
•rH
G CO O
O G -H
•H q -P
•P -H rt
3 -P O
H O O
O CD rH
CO -P
0 CO J3
E O bD
rH O 0
ft v_x.C
94
-------
different points, and dual release would enhance chances of observ-
ing a plume within the sampling network and would very likely improve
knowledge of the wind flow field.
(2) The problem of sampling the plume prior to impact on the
instrumented hill is, at best, difficult. Concentration measurements
can be achieved only by instruments on towers, which are expensive
to install and operate, and which cannot be established with suf-
ficient density to describe the plume prior to impingement. The
use of a ground-based lidar for this purpose is not recommended
by the panel for two reasons: (a) the normal minimum resolution
"cells" for existing instruments are on the order of 3 meters, and
reducing this to 1.5 meters to resolve the small plume as was
suggested in onj of the plenary sessions is pushing the instruments
into operational areas where their performance has not been verified.
Even if the instruments can, in fact, achieve practical resolutions
of this order, a relatively small number of cells would be obtained
within a plume cross-section, and the feasibility of performing
accurate contour mapping of the plume cross-section cannot be
assumed a priori. The feasibility of any such application should
be verified before including it in the small-scale experiment.
(b) The ground-based instruments are expensive to operate ($40,000
to $50,000 per month in the field), and other methods would be more
cost effective. A more effective method suggested by the panel is
the "Laser Illuminator" concept, discussed below.
2.1 Gaseous Tracers
The proposed experiments to study plume impingement on hills
or ridges require the use of a non-reactive, non-depositing tracer
that can be released at precisely controlled rates and measured
accurately at low concentrations. For the distance scales being
consiiered (i.e., a few hundred meters for a "small hill" study
to a few tens of kilometers for a full-scale study at an existing
plant site), sulfur hexafluoride (SFg) appears to be the most
suitable tracer. It has been used extensively as a meteorological
tracer and reliable methods of release, sampling, and analysis are
available. Background concentrations in the atmosphere average
about 0.5 parts per trillion (ppt) and commercially available
electron-capture gas chromatographs can measure SFfi concentrations
accurately down to 1 or 2 ppt. This sensitivity is quite adequate
for tracer measurements out to about 100 km from the release point.
This tracer is readily available at a cost of about $3 per Ib.
95
-------
Other gaseous tracers currently under development, for example,
perfluorocarbons and heavy methanes, with far greater measure-
ment sensitivity (.01 ppt or less), are intended primarily for
use to distances well beyond 100 km. If multiple tracers are
desired, fluorocarbon 12B2 (CF2Br2), 114B2 (C2F4Br2), fluorocarbon
13B1 (CFoBr) are available at somewhat higher costs than SFg.
A potential problem in the use of SF_ in meteorological ex-
periments is the variability of the background concentrations. It
is used in large quantities as an insulating gas in high-voltage
power transformers and switches, hence, there are a large number of
sources that may produce localized concentrations well above the
average background level. A survey of local SFg concentrations should
be included in the site selection process.
2.1.1 Tracer Release. SFfi is supplied as a liquified gas in
pressurized cyclinders and can be released by a gas directly from
these cylinders.
Required release amounts would be on the order of 1 Ib/hr
for the proposed "small hill" experiment; about 10 Ibs/hr for the
"large hill" experiment (distance scale of about 10 km) and about
100 Ibs/hr for the plant site study (scale of about 50 km). In-
dividual release durations of 3-6 hours should be sufficient to
accomplish the objectives of this study. Elevated releases can
be managed with no great difficulty.
Since one objective is to study the effect of release height
in relation to hill height and stabililty profile, it would be
more efficient to use two or even three different tracers (i.e.,
SFg, 12B2, and 114B2) released simultaneously at different altitudes,
at least for some tracer trials. Releasing tracers from different
locations might also maximize the information gained from each trial.
2.1.2 Tracer Measurements in Plume Evolution. It is desirable
to measure tracer plume concentrations and horizontal and vertical
dimensions during its travel from the release point to the terrain
feature of particular interest.
Real-time continuous SFg monitors are available which can be
flown in a small aircraft to provide profiles of SFg concentrations
during a plume traverse.
One version of this instrument, developed at Brookhaven National
Laboratory (BNL), uses the frontal chromatography technique to
provide continuous (but time-delayed) in-flight data for a period
96
-------
of 90 seconds or more. It must then be backflushed for about 2
minutes in preparation for the next 90-sec plume traverse. This
instrument has been successfully flown on many occasions.
Under contract to the NOAA Air Resources Laboratory, Lovelock
has developed a truly continuous real-time instrument by eliminating
the need for a chromatograph column. This instrument can be used
either with SFg or perfluorocarbons and is still under develop-
ment by Lovelock as well as by Dietz at BNL. Lovelock expects to
deliver a flight-worthy prototype by January 1980. Its sensitivity
should extend well below the parts per trillion level, with a response
time on the order of 1 second.
Unfortunately, none of these instruments will have sufficient
resolution to obtain a true profile of the very narrow plume ex-
pected in the "small hill" experiment. They will, however, be able
to detect the plume and provide crosswind integrated concentra-
tions through the plume. The real-time instruments will be able
to provide a great deal of data on tracer concentration profiles as
a function of altitude and distance from the source for the larger
scale experiments.
2.1.3 Plume Impingement. All three scenarios being considered
will require an extensive array of about 100 samplers to measure
tracer concentrations in the vicinity of plume impingement on the
major terrain feature.
Several methods are available for collecting whole-air samples
for SFg analysis. The NOAA Air Resources Laboratories Field Research
Office (ARLFRO) at Idaho Falls has developed a battery-powered
sampler that pumps air into a plastic bag, providing a time-integrated
sample at $100-200 per unit. The units must be switched on and
off manually. Several hundreds of these samplers are on hand
at ARLFRO. Automated sequential whole-air samplers, using bags
or syringes, are commercially available at a cost on the order
of $2,000. An automated sequential sampler, developed at BNL,
collects up to three samples in evacuated steel cylinders. The
system is radio-controlled for simultaneous operation of all samplers.
Twenty samplers are available at present.
Considering the large number of samplers required (100) and the
relatively short sampling intervals desired (10 min to 1/2 hr), a
commercial automated sampler, providing 24 syringe samples, would
appear to be the best choice. The other types of samplers can be
used to augment this network.
97
-------
For the "small hill" experiment in particular, the sheer
number of samples required may be overwhelming (100 samplers x
6 samples/hr x 4 hours = 2400 samples per trial). Meteorological
data or smoke plume information will have to be used for screening,
to select perhaps 10% of the samples for immediate analysis in pre-
paration for the next trial.
An alternative, to provide quicker turnaround, would be to use
analyzers that provide real-time readout (at 2-5 minute intervals
of concentration) to be fed into a centralized computer. The cost
of such analyzers (about $6,000 each) plus the computer hardware
nould have to be weighed against the cost of manual processing
(perhaps $1/sample) and the need for rapid data processing.
2.2 In Situ Measurements of Wind, Temperature, and Humidity from
Towers and Balloons
The measurement of meteorological variables — wind, tempera-
ture, humidity, pressure — by in situ instruments in shelters or on
towers provides the data core for most field experiments. This will
oe true for the proposed complex-terrain plume tests also. A brief
review of the expected accuracies and applicability of such measure-
ments is, therefore, in order.
For measurements of mean wind quantities, propeller or cup
anemometers with vanes as required, are quite suitable if the instru-
ments are properly exposed away from flow obstructions.
Standard cup and vane systems have a general accuracy of +5
in direction and about +0.2 m/s for wind speed. Resolution of the
wind direction is obtained to one degree. Wind speed threshold is
in the neighborhood of 0.25 m/s, but reasonable vane response to
wind gusts is limited to speeds above 1-2 m/s.
Temperatures on the tower are generally recorded with a thermis-
tor or a platinum resistance thermometer in an aspirated radiation
shield. Absolute accuracy of the thermistor is about 0.5°C with
resolution to 0.1 C; somewhat better accuracy and resolution are
available with the platinum resistor. Temperature differences on the
tower are recorded by voltage balancing between the two sensors.
Accuracy of Delta-T with thermistor is about 0.1°C.
Relative humidity is best measured on the tower by a dew-point
hygrometer system. Accuracy is in the neighborhood of 0.5°C.
98
-------
For the accurate eddy-correlation measurement of turbulent
plumes, the distance constant for anemometers and the response time
for temperature sensors must be short enough to resolve all eddies
of significant size. If we are to really understand the turbulent
impact of plumes onto rough terrain through the atmospheric surface
layer, then instruments with at least a ten hertz bandwidth must be
used. This implies hot-wire or sonic anemometers and fine-wire
resistence thermometers. Since three-axis sonic anemometers-
thermometers have demonstrated their accuracy (to small fractions of
a degree and a meter per second) at the NOAA/NCAR Boulder Atmospheric
Observatory, they are a logical choice for turbulent surface layer
measurements. Data processing algorithms compatible with mini-
computers are also available.
2.3. Meteorological Towers
In situ instruments must be appropriately mounted, frequently
on towers, to place the instruments at the desired heights. Tower
costs increase quickly with increasing height. Estimates for
costs of installed towers, with no instruments, were obtained by
asking for bids from a California-based contractor subsequent to
the end of the Workshop. These estimates assume that the towers
will be stressed for ice, wind, and lightning, and that they have
de-icing heaters. Each tower is equipped with four instrument
elevators, with a 19-wire cable to each elevator. Cost estimates were
as follows:
HEIGHT COST
50 meter $30k
100 meter $41k
150 meter $75k
2.3.1 Remote Wind Sensing in the 100-Meter Hill Experiment.
The excellent range resolution and short-range measurement capabili-
ties of FM/CW radar provide important advantages for monitoring the
the small-scale experiment. Should there be fine scale wind shear, as
frequently occurs in stably-stratified valley atmospheres, this system
will be able to measure it. Further, the intensity of the return
delineates these shear zones, thus providing a record of descending
temperature inversions and the ubiquitous gravity and shear waves
associated with such interfaces. Real-time wind readout will also
aid in properly placing the plume sources.
99
-------
With the range of the radar of several kilometers, changes in
the mesoscale or synoptic scale wind structure that might influence
the experiment can also be monitored. Thus, the arrival of warm
fronts, or density currents riding over and eroding the surface-based
nocturnal inversion, can be anticipated.
Supplementing the radar with CW Doppler lidar or Doppler acoustic
sounders is also attractive. These lower cost, more mobile sensors
can significantly expand the volume of the atmosphere being measured
for winds.
2.3.2 Wind Measurements with the FM/CW Doppler Radar. The
FM/CW radar detects radiation backscattered from index of refrac-
tion turbulence structure in the clear atmosphere. A 10-cm wave-
length unit developed at the NOAA Wave Propagation Laboratory employs
phase measurement to detect the Doppler frequency shift from the
moving turbulence that advects with the wind. Good signal-to-noise
ratio is obtained at ranges up to several kilometers, and the
minimum range can be as short as 20 or 30 meters. Range resolution
as small as 1.5 m is feasible, and radial velocity accuracy of 0.1 m/s
has been demonstrated. Ground clutter restricts elevation angles
to 20° or greater.
To measure both components of the wind, u and v, the radar
performs a conical scan about the zenith in a so-called VAD (velocity-
azimuth display) mode. Asymmetry of the VAD velocity about zero
provides a measure of the vertical velocity, averaged over the
diameter of the scanned circle. A minicomputer in the system pro-
vides real-time readout of winds versus height. Separate transmitting
and receiving antennas are mounted on a trailer, and a separate
semi-trailer houses the radar electronics and data processor. Setup
time for the equipment is approximately one day on a level site
20 m x 40 m; three-phase 208V or single-phase 240v, 10 kw is required.
A two-man per shift crew can operate the radar once it is in place.
2.3.3 Wind Measurements with CW Doppler Lidar. A number of CW
Doppler lidars have been developed (at NOAA, EPA, DOT/TSC, NASA,
Marshall) to measure winds by monitoring the frequency shift in 10.6
m radiation backscattered from atmospheric aerosols. These CW units
obtain range information by focusing the infrared radiation at the
desired range, where the backscattered radiation then mixes efficient-
ly with the laser local oscillator on the photodetector. These
systems are quite suitable for VAD wind measurements in the clear
air at ranges to 500 m, although range resolution is quite coarse
at ranges beyond 100 m because of the elongated focus for typical
optical apertures of 20 to 30 cm. The units are all mounted in
100
-------
self-contained trucks. Setup time on a level site 10 m x 15 m is
four hours. Single phase 220V power, 5 kw, is required. A two-man
crew per shift is required.
2.4 Laser Illuminator
• Experimental Use. Plume visualization for the proposed
small hill experiment can be accomplished using an aerosol tracer
and several laser illuminators with suitable recording devices.
The measurement technique is only applicable to a nighttime release.
Several laser illuminators are positioned between the source and the
hill, establishing thin planes of light perpendicular to the plume
path (see Figure 8).
Plume location is readily observable by the experimenters and
plume dimensions can be established through conventional photographs
or video taping with one or more cameras positioned along the axis
connecting the source tower and the hill.
• Equipment Description. All components, are readily available.
Cost of the laser illuminator and camera for each plane is estimated
to be $10,000, with accuracy not limited by the experimental con-
figuration but by the effort expended on edge detection in the images.
Illuminators could be strobed to establish timing. Another approach
using search lights may be possible.
2.5 Detection of Small Plumes by Electric Field Measurement
Electric field measurements in connection with plume trajectory
studies are a novelty and need further exploration. However, if the
artificial smoke particles are charged electrically, so as to perturb
the atmospheric electric field measurably, field measurements can be
a powerful tool to detect a plume or even measure the concentration
of particles in a plume. This could be done remotely from the ground
by measuring the field as function of distance from the centerline
underneath the plume with a field mill mounted on a movable platform
like a car. This could also be done airborne by a mill mounted in
the nose of a model airplane that is remotely directed across and
through the plume. The model aircraft speed is slow enough to achieve
spatial resolution of 10 meters. The technique, therefore, would
be applicable to scales that are expected in the 100- and 500-m hill
studies.
101
-------
Light Planes
Source
Illuminators
Hilt
of Plume Cross Section
Figure 8. Experimental arrangement using laser illuminators,
102
-------
3. MEDIUM-SCALE OR 500-METER HILL EXPERIMENT
In this experimental scale, initial vertical sigraas are on the
order of 40 to 60 meters, with travel distances of the order of
10 kilometers, and plume elevations of 250 to 750 meters. At these
scales, the conventional tower-based wind measurements must be aug-
mented or supplemented by tethersondes, minisondes and remote devices.
The doppler acoustic sounder is particularly well-suited to operations
on this scale, as is the FM/CW radar. Tetroons may be useful for
defining trajectories.
Plume dimensions are large enough to satisfactorily use ground-
based lidar unibs for dispersion parameter measurements; airborne
lidar can contribute to mapping the impingement of the plume on the
hill, provided the plume is tagged with a conservative particulate
tracer and is large enough. In situ sampling with instrumented
aircraft may or may not be useful, depending on the plume dimensions
selected for the release. To effectively use aircraft, plume dimen-
sions must be of the order of several hundred meters to a few
kilometers; plumes less than a hundred meters are too narrow to
effectively measure with aircraft. Turbulence and wind fields aloft,
however, can be effectively measured with instrumented aircraft.
3.1 Flow Field Determination by Doppler Acoustic Sounding
Transmission of an acoustic pulse upward, and measurement of
the doppler shift of the echo which is returned from small scale
atmospheric turbulence, are the basis for one technique for the
remote measurement of winds aloft. Such instruments, which are now
commercially available (at a typical cost of $40 to $50,000) can be
used in a complex terrain study to define the wind and turbulence
field-, aloft, to determine the depth of the mixed layer, and to
illuminate localized phenomena such as shear layers, inversion
layers, and thermal plume activity. Recently-developed instruments
can be mobile and have a useful range around one kilometer, which
makes them effective for intensive or routine field studies of
plume dispersion aloft.
Two different configurations of the Doppler acoustic sounder
are available. Bistatic configurations, based on designs developed
at the NOAA Wave Propagation Laboratory (WPL), are manufactured
by the Radian Corporation. A bistatic unit with a somewhat different
configuration has been manufactured by Xonics, Inc. The NOAA WPL
also has several bistatic units of its own design. (Such instruments
have a maximum range of 600 to 800 meters and a minimum range of from
100 to 200 meters above the ground). Monostatic configurations,
103
-------
based on a design developed at the University of Melbourne (Australia)
are manufactured by AeroVironment, Inc. These instruments have a
demonstrated maximum range, under appropriate meteorological condi-
tions, in excess of one kilometer, and a minimum range of under 30
meters. All of these instruments have different signal processing
techniques for extraction of the wind speed; some of the performance
specification differences cited here depend both on the configuration
and the processing scheme.
Either type of instrument has a wind speed resolution of about
0.2 m/s and a maximum wind speed measurement capability of 25-35 m/s.
Vertical resolution of wind speed data is typically 30 m, although
resolution down to 10 m is attainable.
These resolution scales make acoustic sounders best suited for
studies around larger terrain features, although they can provide
useful data on the higher level winds over smaller (100 meter) hills.
The absolute accuracy of these instruments is not well known.
Tests of acoustic sounders next to the NOAA WPL 300 m tower in
August 1979 should provide some information on this question.
The choice of bistatic versus monostatic configuration depends
on a number of factors. Bistatic units require a substantial area
(several hundred meters on a side) for laying out three or four
antennas; the monostatic antennas are all located at one point,
and mobile units are available.
The bistatic system measures the wind in a single vertical column
of the atmosphere; monostatic units make measurements in several
directions from the antennas, and then require an assumption of wind
field homogeneity to construct all wind components. In either case,
the wind measurement reflects an average value over a substantial
volume of air.
The lower height limit of bistatic systems is substantial, and
depends on the lowest altitude at which the transmitting and receiving
antenna beams cross; monostatic system measurements, on the other
hand, can begin very close to the surface.
The maximum range of bistatic systems is also defined by the
antenna geometry, while signal-to-noise considerations define the
upper limit of monostatic configurations.
Because of the siting considerations, monostatic doppler acoustic
sounders are probably best suited for the proposed complex terrain
104
-------
study. Their versatility probably outweighs the disadvantages of this
configuration.
The routine recording of turbulence aloft by doppler acoustic
sounding should be possible by 1980. With the support of the Electric
Power Research Institute, AeroVironment is currently evaluating
various algorithms for the derivation of diffusing turbulence levels
from doppler acoustic signals. This evaluation includes testing of
these algorithms against in situ turbulence measurements made on an
instrumented tower. The resolution of such data should also be on the
order of 0.2 m/s (or better) for the root-mean-square wind speed
variation. This resolution will be adequate for defining diffusion
under all but the most stable conditions.
Derivation of the mixing height from the turbulence field is
straightforward. In addition, display of the amplitude of the return
signal in the traditional time-height form of a non-Doppler acoustic
sounder record, can illuminate the atmospheric structures (shears,
thermal plume, inversions, local fronts) influencing the transport and
dispersion of the plume.
3.2 Flow Field Determination by Tethersonde
For measurements of the structure of the planetary boundary
layer, a sensor package carried aloft under a tethered kytoon was
developed several years ago at the National Center for Atmospheric
Research. This instrument is now commercially available from
Ambient Analysis, Inc., in Boulder, Colorado, under the trademark of
Tethersonde, for $8,000.00 to $10,000.00, depending on the options
selected.
The Tethersonde instrument package consists of sensors for
wind speed, temperature, pressure, wet bulb temperature, and the
orientation of the package relative to magnetic north (which serves as
an indicator of wind direction since the instrument package orienta-
tion follows that of the kytoon as it points into the wind). An
optional thermal structure function, CT sensor is also available.
A rechargeable battery provides enough power for about four hours
of continuous operation.
3
The package is carried aloft by a small, 3.25 m kytoon. With
relatively light winds aloft (less than 10 m/s) the Tethersonde can
reach an altitude of 1 km above the surface at a cool, sea level
site. To provide additional lift for use in stronger winds or at
higher altitude launch sites, a 4.25 m kytoon is also available.
A motorized winch handles the tether line.
105
-------
A radio-telemetry system cycles through all data channels
once each 30 seconds and transmits the signals to a ground based
receiver. The data is printed on paper tape, along with the time of
each scan. An optional programmable calculator can be attached to the
system to calculate the balloon altitude by integrating the barostatic
equation, using data obtained during the ascent, and can also record
all data on a tape cassette for subsequent computerized analysis.
The Tethersonde is a versatile sounding tool for exploring
the structrue of the planetary boundary layer in all scales of
complex terrain. It is particularly useful for study of the flow
over and around a hill. Since it can be tethered at a relatively
constant height, it can be "parked" at the altitude of a plume and
will then provide a continuous record of the direction and speed of
the plume-transporting wind.
The system is portable, and can easily be operated by one
person with some practice, as long as surface winds are not strong.
The ground station requires 110V, 60 HZ electricity, which can be
provided by an inverter and a car battery. launching and retrieval
in strong surface winds (above 10 m/s) is difficult, as is operation
when winds aloft exceed about 20 m/s.
The response ranges and precision for the various components
recorded by the tethersonde are:
Wind Speed 0.5 - 20 m/s +0.25 m/s
Wind Direction 0 - 360° +5°
Temperature -50 - +50°C +p.5°C
Pressure (Dif- 0 - 100 mb +1 mb
ference Relative
to Surface)
3.3 Minisonde
Simple, expendable instrument packages, similar in concept to
the Tethersonde package, can be carried aloft by pilot balloons to
provide soundings of temperature to heights greater than those
attainable by the Tethersonde. Tracking of the minisonde balloon
by theodolite allows calculation of the wind speed and direction
profiles, provided that the ascent rate of the balloon is known.
In flat terrain and a slightly stable atmosphere, the ascent rate
can be assumed to be constant. In complex terrain, or in unstable
106
-------
or strongly stable atmospheres, a pressure sensor on some makes of
sondes provides a direct record of the balloon height or a second
theodolite is needed to determine the balloon height. Some sondes
also record wet bulb temperature.
The rainisondes are similar in concept to radiosondes, but
are easier to handle and much cheaper. Although automatic radio
tracking of the sondes is not possible, the ground stations for
minisondes often provide for digital data recording on cassettes,
and can contain all of the features of the automated Tethersonde
ground station. Minisonde sensor performance specifications are
generally almost comparable to those of the Tethersonde: The wind
data derived from theodolite tracking is less precise, however, and
its quality depends continually on operator skill.
Minisondes are useful for all scales of complex terrain studies.
Because of the discrete nature of the soundings, and their cost,
the sondes are best used to study slowly varying phenomena, such
as flows well above the surface.
3-4 FM/CW Wind Finding
All that has been said about the FM/CW radar for wind measure-
ments in the smaller scale experiment applies here as well. The
range of the radar to several kilometers is even more important here,
where towers cannot reach into the plume heights.
3.5 Flow Field Determination by Aircraft
Within the framework of the "medium hill" experiment, the light
aircraft can extend the meteorological observation of the plume en-
vironment. This can be best accomplished through horizontal traverses
perpendicular to the plume trajectory. Sampling flights through the
plume can be used for this purpose, although it may be desirable
to extend the horizontal traverses somewhat. Principal objective
of the measurements would be to determine the homogeneity of the wind
and turbulence environment at plume level.
Parameters of concern are:
1. Three-dimensional velocity components (u,v,w) sampled at
100 HZ so that turbulent components as well as average values can be
determined. Turbulence parameters are measured by a gust boom
referenced to gyro-accelerator system which provides a measurement
platform stable to low frequency fluctuations.
107
-------
Turbulence data can be processed into OQ and a. values for a
range of averaging time. Thirty second averaging (about 1 km of
flight path) is probably the longest time acceptable for horizontal
mapping. Accuracy is reported to be + 0.1 m/s (Gilmer, McGavin and
Reinking, 1978). Through these measurements, a horizontal map of
wind and turbulence can be obtained at plume level to be used in
modeling inputs.
2. Flux parameters can also be measured if desired during the
horizontal traverses. Both vertical height and moisture fluxes can
be obtained by cross correlation of temperature and humidity fluctua-
tions with vertical velocity variations. These data form an indepen-
dent measure of the diffusion environment.
3.6 Tracer Measurements Via In Situ Instrumented Aircraft
The space resolution of aircraft measurements depends critically
on aircraft speed and response-times of the instruments. Light-twin
aircraft can be flown at speeds of 50 m sec . Time constants of
instruments and gas measurements must be of the order of one second.
Hence, the space resolution of the sensing equipment is of the order
of 50 m. This is probably good enough for tracer experiments of the
scales in question for this experiment.
Instrumented aircraft can be used to measure in situ the concen-
tration perpendicular to the axis of the plume of tracer gases
and particles (total concentration, flight scatter coefficient).
It is proposed to measure centerline concentration and horizontal
dispersion of particles and gases. Repeated measurements result in a
mean centerline concentration and horizontal dispersion (ay ) of plume
effluents. Repeating the measurements at different heights from the
axis gives the vertical dispersion (pz ).
Performing the measurements at different distances from the
source upwind and downwind of the obstacle (hill) results in the
downwind dispersion at both sides of the hill, indicating the effect
of the hill on plume movement. Aircraft tracer measurements, includ-
ing data processing, will cost about $8,000/day, for most contractors.
3.7 Lidar
• Equipment Description - Several ground-based, van-mounted
lidar systems are available which are suitable for this type of
experiment. Systems are available with calibrated scanning mounts,
a feature necessary for obtaining detailed plume cross-sections.
108
-------
Spatial resolution of current systems approaches several meters,
adequate for all experimental scenarios considered. All lidar
systems have real-time display capabilities and thus can be used
to locate plumes or to direct other experimental efforts into
plumes. Data output is in the form of hard copy line profiles,
density plots, density contour plots, and processed and unprocessed
magnetic tapes.
3.8 Ground Based Lidar
• Experimental Use - The ground-based lidar would find utility
in both the middle-sized hill experiment, and in a full-scale
experiment involving an actual power plant source. Much field work
of this nature has been routinely performed for a number of years,
and the methodology and equipment are well proven. The spatial
resolution of existing devices is comparable to the dimensions of
the small-scale experiment, but other methods of determining plume
location or dimensions may be more cost effective at this scale
(see LASER Illuminator). The ground-based lidar can be used primarily
to determine plume dimensions and, in particular, the behavior of the
plume as a function of time. A workable method of determining source
content is plume sampling by aircraft, with subsequent extrapolation
and averaging of plume concentrations using lidar information taken
over extended time periods. The lidar can also be used to determine
plume impact points with proper safeguards for eye safety of experi-
mental help.
Some features of specific systems might have additional benefit
to the experiment depending on the actual site chosen. For example,
one available system (NOAA) has the capability of measuring depolari-
zation ratios and thus discriminates between spherical particles
(i.e., combustion particles, oil droplets, etc.) and background
fugitive dust. Another system (SRI) can be used in a mobile mode
pointing vertically upward. If extensive road networks exist and
the scale of the experiment is large, this might prove of benefit
over scanning systems. The lidar systems could also be used to obtain
meteorological information when not being used to for plume measure-
measurements.
• Costs - Ground-based lidar systems can be operated for around
$50K/nionth excluding travel cost.
3.9 Airborne Lidar
The airborne lidar maps aerosol distribution beneath the air-
craft. In particular, the airborne lidar can be used to find the
109
-------
dimensions and location of plumes. In the context of the three
scenarios considered, the airborne lidar would only have utility
in the medium hill controlled release and the monitoring of the
effluent from an actual power plant. The airborne lidar and the
ground-based lidar can be used to determine plume location, cross-
sections between the source and the plume, and in addition, the
airborne system can be used to position a sampling aircraft within
the plume for source term determination. Flights over the subject
hill in either the controlled release middle size hill, or full-scale
hill experiments, would yield plume impact information. Figure 9
shows airborne lidar returns several kilometers downwind from a power
plant source. The second trace from the left side of the photo shows
little scattering at ground level, whereas, the third trace shows
considerable scattering due to plume impact.
- 2km-
CLEAR AIR
< 1km-
PLUME
REGION Ground
Surface
Om 60m 120m 180m 300m 360m
Distance Along Ground Path
Figure 9. Downlooking airborne lidar showing plume contact with
ground. Note ground level scattering on center trace.
3.10 System Description
• System Description - At the present time two operational
airborne lidar systems exist and are operated by the EPA Research
laboratory in Las Vegas (EMSL-LV). The system of interest is the
latest device developed by that laboratory; a two-frequency Lidar
110
-------
utilizing a Nd-YAG Laser with a doubling crystal. The device has a
vertical resolution of 10 m and horizontal resolution dependent on
aircraft ground speed and the highest frequency of operation (10
Herz). The system's weight and power requirements are compatible
with small twin-engined aircraft. In this type of platform the
horizontal resolution is approximately 6-10 meters. The device has
real-time display capabilities on board the aircraft, thus enabling
the operator to tell other personnel the plume location and height
and to position other sampling aircraft into the plume. A ground-
based computer system is used to display and to provide hard-copy
output in short time turnaround after the aircraft lands.
Another airborne lidar system is currently under development by
SRI.
• Systems Cost - A rough estimate of total system costs includ-
ing aircraft and ground-based data system is $4,000.00 per day of
operation.
4. MEASUREMENT TECHNIQUES FOR FULL-SCALE TRACER AND ACTIVE POWER
PLANT STUDIES
All of the instrumentation mentioned so far can be used with
full-scale tracer studies and studies of the active power plant;
however, the scales of measurement are now large enough that remote
and aircraft measurements become most economical for most parameters.
The panel recommended that some fixed meteorological tower measure-
ment sites be established and operated to collect a lengthy wind
record in this case also.
The measurement of S0~ or other combustion emission is so ex-
pensive, both in terms of hardware and monitoring operations costs,
that :.s should be considered as a tracer species only if sufficient
funding becomes available for a "second-phase" or "verification"
study where models developed in controlled-release experiments
would be verified on actual power plants. However, an investigation
of existing S02 monitoring data around S0r> sources in complex terrain
may be desirable, and this concept receives further attention below.
Measurements at an active power plant would necessitate continuous
measurement of relevant stack parameters: 302 and/or NO stack-mouth
concentrations, flow rates, and exit temperatures. These data are
available through conventional techniques but are very expensive to
obtain. They are essential to any "real-world" study.
Other techniques become useful in measurements on an active
plant; Van-based correlation spectrometers can be used to locate
111
-------
plume trajectories and estimate horizontal dispersion parameters.
Tetroons should be released to examine valley-flow fields.
Ground-based and airborne LIDAR become extremely useful in
mapping plume cross-sections and impingement of the plume with the
surface.
4.1 Meteorological Measurements by Aircraft
A light aircraft can be used in the "real world" experiment to
supplement the fixed and balloon-borne observations, the principal
objectives are the horizontal mapping of the diffusion and the ex-
tension of sounding data above the 1 km level, as needed. The latter
information, coupled with data from ground-based systems, could
provide the needed wind-diffusivity profile for model calculations.
4.1.1 Horizontal Mapping Data. During the horizontal traverses
made by the light aircraft to sample the plume concentrations, wind
and turbulence data can be gathered to permit horizontal mapping of
the plume environment. Such a map would give average wind and dif-
fusivity inputs into the model and would permit an evaluation of the
spatial homogeneity of the environment.
Wind and turbulence measurements are made by the airpraf,t bv
means of a gust boom, coupled to a gyro or INS system. U , V , W
are recorded at a high data rate suitable for processing into
(U'2 , V'2 , W'2 ) after arbitrary averaging times. Wind measurements
are made with the same system.
In the real-world experiment, marked differences in wind and
diffusivity environment are expected as the result of terrain in-
fluences. The light aircraft offers the potential of exploring
these variations within the limits of safe flying. The mobility of
the aircraft also permits its use to explore up-wind turbulent in-
fluences which may be advected into the experimental area.
4.1.2 Sounding Data. The doppler acoustic radar is expected
to provide wind and turbulence profile data to a height of 1 to 1.5
km. At many sites, wind and diffusivity data will be required to a
higher elevation. The aircraft can be used to extend the sounding
data above the doppler acoustic radar so that a diffusivity-wind
profile can be obtained within the plume environment.
t i t
U , V , W measurements, together with an INS or similar
system, can provide both wind and turbulence information for con-
struction of this profile. It is suggested that this might best be
112
-------
accomplished by a series of horizontal aircraft traverses (of about
30 sec each) at elevation intervals of 100 m above the acoustic
radar. These data could then be processed into profile form and
fitted to the acoustic radar data to form a complete vertical profile
of wind and diffusivity.
4.2 Tetroons
Tetroons are balloons balanced to fly at a constant-density
level. Tracking is accomplished by a tracking radar which records
azimuth and elevation at uniform time intervals. Winds can readily
be obtained over one-minute time periods at the level of balloon
flight.
The tetroon is useful in indicating parcel trajectories and in
providing detailed wind measurements along the trajectory. Wind
speed accuracies of the order of a few tenths of a meter per second
can be obtained.
The tetroon, however, suffers from a few disadvantages. Under
strong heating conditions, the constant density surface tends to
decrease in above-ground-heights as the surface temperature rises.
The balloon trajectory, therefore, may not correspond to a plume
trajectory. In addition, tetroon oscillations may occur as the result
of the attempts of the balloon to find its appropriate density sur-
face. These oscillations may lead to wind measurements and tra-
jectories which do not represent constant elevations above ground.
4.3 In Situ Aircraft Measurement
It is suggested to measure in situ the horizontal distribution
of gases and particulates perpendicular to the plume axis. Typical
gases are S02 and NO (both reactive) and, if desired, SFg (inert).
Typical aerosol parameters are total particle concentration and
light-scattering coefficient. Due to negative electric charges,
which the effluent particles acquire during or before exiting the
stacks, they strongly alter the atmospheric electric field. This
effect can be sensibly and accurately measured with electric field
mills.
Aircraft speeds of 50 m sec of a light-twin and instrument
response times of the order of seconds for direct sensing of particles
and gases result in space resolutions of 50 m. The response times of
field mills of 60 Hz gives a space resolution of 0.83 m.
113
-------
Repeated measurements of horizontal distribution at given alti-
tudes result in mean concentrations and in horizontal dispersion
parameters (°y) of the aerosols and gases. Repeated measurements
at different altitudes give the vertical dispersion (az). Repeating
these measurements at different distances from the source gives infor-
mation of the downwind dispersion (
-------
5. DATA MANAGEMENT AND QUALITY ASSURANCE PANEL
PANEL LEADER: Ronald E. Ruff
PARTICIPANTS:
Jesse Coleman
Robert C. Koch
Donald W. Moon
J. B. Tommerdahl
V. E. Derr
Thomas F. Lavery
Joan Novak
5. DATA MANAGEMENT AND QUALITY ASSURANCE
1. OVERVIEW
Data management and quality assurance, QA, programs are essential
ingredients in the conduct of the field experiment. Though normally
treated separately, we recognize the necessity for their integration
in the manner illustrated in Figure 10. The illustration shows the
major phases of data management and how information is fed-back as a
quality control measure.
The first feedback occurs after the routine data review conducted
in the field. This review is the focal point of data management and
qualify control in the field conducted in a near real-time environ-
ment. The first review consists of analyzing the data for (1)
relevance to the field objectives, (2) planning of future experi-
mental activities, and (3) improper equipment operation. This latter
point serves as a baais for the primary QA feedback wherein mal-
functions are detected on the spot. Immediate repair action can then
be undertaken by the appropriate field personnel.
The second QA feedback occurs during the final checking process.
This is normally accomplished off-site at the end of the activities.
Its purpose is primarily meant to eliminate bad data or flag question-
able data and replace these data with correct values, if possible.
Lastly, an independent data review is normally conducted by some user
other than the field studies contractor.
115
-------
116
-------
In describing the various activities associated with data manage-
ment, we have distinguished field activities from off-site activities.
However, we strongly recommend that both activities be conducted by
the same group through the final data archiving process. One excep-
tion is our recommendation for independent quality assurance audits
conducted by a separate group.
2. FIELD REQUIREMENTS
2.1 Introduction
Requirements for supporting field operations are reviewed for
two extremes. On one extreme, only a minimal number of data are re-
quired during real-time and other data processing is restricted to
rudimentary displays of data. On the other extreme, real-time acqui-
sition, reduction, and display of many parameters are performed; in
addition, other tasks are supported such as initial reduction and
display of remote sensing data, back-up copying of all data, incor-
poration of manually entered data, and selective editing. In
accordance with Workshop recommendations, we have assumed a two-
phase experiment will be conducted. Phase I would be the small hill
experiment; Phase II would be a full scale experiment at a power
plant site.
The determination of the data processing scheme depends upon
(1) the specific requirements of the experiment, and (2) similarities
between the Phase I and Phase II experiments. If the Phases are
similar, the selected field data system and procedures should be the
same for both phases so that Phase I serves as a shake down of the
system for Phase II.
Quality assurance procedures remain essentially the same regard-
less of the complexity of the experiment. However, data review is
more easily accomplished with the aid of adequate data display
facilities on site. Also, the ability to enter manual calibrations,
corrections, or logistical information while still in the field is
desired. A small computer system may not support such field entries.
Hence, these functions must be performed offsite with perhaps some
loss in data integrity.
In the following sections, we describe general requirements for
three field functions.
117
-------
(1) Data Acquisition, Preliminary Reduction, and Archiving —
specifying the forms of data acquisition, what field calcu-
lations must be made, and the data product that is gener-
ated in the field.
(2) Data Review — specifying the on-site information required
to (a) check the validity of the data in a timely fashion,
and (b) to support the planning of future field (experi-
mental) operations.
(3) Procedures and Calibration — describing the pre-field and
field activities required to maximize the quality and
integrity of the data.
Lastly, we recognize the special requirements posed by remote
sensors. Accordingly, we address these in the last sub-section.
2.2 Data Acquisition, Archiving, and Preliminary Reduction
In discussing the requirement for the field data management
system, we have resticted ourselves to generalizations to some extent.
The actual requirements are directly related to the complexities of
the field measurement plan.
2.2.1 General. The following features are described for most
types of field experiments that are foreseen:
• All data should be available in semi-finished form within
12 to 24 hours for experimental planning and review.
• Those data required for decision making and the general
management of experiments in progress should be displayed
in real time. These data should include sufficient
meteorological information to determine if conditions
are favorable for the conduct of other measurements
(e.g., tracer release). It is highly desirable, but not
necessarily cost effective, to display real-time ground-
level concentration values. Graphics display of real
time is also highly desirable but not necessary.
• All data should be archived in the field in raw form
(without necessarily being converted to more usable
parameters such as diffusion coefficients, plume height,
and so forth); this stage of data reduction and conver-
sion will be reserved for a later time.
118
-------
All data should be backed up by duplicate records; data
tapes should be duplicated in the field or off-site as
soon as possible.
Strip charts should not be used for the primary data
record although they are acceptable as backup.
Handwritten records of maintenance and calibration teams
should be carefully maintained for later use in error
flagging, quality estimation, and editing.
The data archival system should include a quality index
technique to flag poor quality data and data that have
been edited.
Certain sensor and system status information (e.g., power
critical voltages, flows) should be routinely acquired
and entered on the raw data tapes.
The data acquisition system should include an automatic
error flagging system for "obviously" erroneous data
(i.e., extreme values or rates of change). This will
help to identify and correct malfunctioning instruments.
2.2.2 Different Types of Data and How They Are Archived.
a. Continuous Data. Continuous data are considered to
be hard wired (or otherwise transmitted) into a data logging computer
that records basic signals representing wind speed, temperature,
continuous ground-level concentrations, and so forth. These data
should be stored on disk or magnetic tape, automatically checked for
errors, and made available for real time display.
b. Batch Data From Self-Contained Systems. Batch data
include those data from remote sensors, mobile or remote systems
(such as aircraft, tethered balloons, pibals, etc.), and batch
processing of bag samples. Some of the data may be available on a
nearby real time basis (such as remote sensing systems). Others may
not be available for hours. These data will be available in different
forms and it is not necessary that they be combined onto a common tape
in the field. Care should be taken, however, to insure that all tapes
can be easily read by the central, off-site computer at a later date.
All data from these various systems should be processed, on-site,
to the extent required for program management. For example, lidar
data should be processed for examination of trajectories and plume
119
-------
cross-sections. All systems measuring the same thing (such as FM/CW
radar, tower wind instruments, and tetroons) should be compared for
agreements.
2.2.3 Data Acquisition and Archival System. The configuration
of a data acquisition and archival system can be outlined in only a
general way until the field data requirements are specified exactly.
As a minimum requirement, key meteorological parameters should be dis-
played and stored on a real-time basis. (If such parameters are to
be routinely monitored at times other than during the intensive field
periods, it may be cost effective for the field computer to remain in
the field. At the other extreme, a more comprehensive computer could
be needed to process more real time data, display auxiliary remote
sensing data, and extensively support data archival. For this more
demanding requirement, we estimate that a more comprehensive field
mini-computer facility will be required. Peripherals could include
a disk, three magnetic tape drives, interactive graphics display
terminal (perhaps two), and a high speed (electrostatic) printer/
plotter.
2.3 Data Review (Field)
Non-real time detailed review of data is expected to be conducted
off-site incorporating final checks of consistency, extreme value and
other automated data checking techniques. Final calibration adjust-
ment of data is also expected to be accomplished off-site.
Real-time and near real-time preliminary on-site field review of
data is limited to that necessary to determine the proper working
order of instrumentation and recording equipment, and that necessary
to operationally plan the successful conduct of the experimental
program.
Preliminary quality assurance of instrument recording, conver-
sion, and transmission may be accomplished by spot check comparisons
of manually reduced analog information and recorded digital informa-
tion. Computer graphics may be beneficial for reproducing the analog
trace for comparison with back-up chart recording. Real-time compari-
sons of wind and temperature field data for consistency should also be
accomplished in the field.
Real-time wind and temperature field data available in the system
provides an exceptional operational planning tool. While it is
important to assess the possible overriding effect of synoptic weather
situations on operation of the experimental program, it is essential
also to assess the diurnal effects of topography and temperature
120
-------
gradients as reflected in the wind and temperature field data recorded
for the past 24-36 hours. The trends established concerning the
timing of drainage flow, the intensity and height of the nocturnal
surface static layer, break-up, and wind direction and speed provide
good indication of whether the day's activity should be devoted to
intensive measurements or perhaps maintenance and calibration and
other activities.
Some meteorological and other measurements essential to meaning-
ful description of the wind and thermal field, such as pilot balloon
observations, by nature require manual entry into the system for near
real-time operational planning. These should be reviewed and compared
with other wind data for consistency.
Additionally, it is assumed that sufficient climatological infor-
mation for the immediate area of concern is made available for
operational planning purposes. Such data should be entered into the
system for recall and display in a meaningful form by the field
program director.
2.4 Procedures and Calibrations (Quality Assurance Program)
This section covers the essentials of the QA program as conducted
in the field. Our purpose is to delineate some general (and a few
specific) procedures which are applicable to the field preparation and
field activities phases of the program which will be useful in assur-
ing data quality. In this regard, it is recommended that a general
quality assurance program be activated which will have as its basic
elements: (1) review of instrumentation and system specifications;
(2) the review of the calibration and operational procedures for the
various instrumentation and data acquisition systems; and (3) the
independent audit of all measurement and data systems, where appli-
cable. The individual elements of such a recommended program are
discussed in the following paragraphs.
2.4.1 Review of the Instrumentation and Data Acquisition System
Specification. After the program design is firm, including a prelimi-
nary selection of instrumentation and data systems, it is recommended
that a review be conducted of the measurement and data system speci-
fications and operational characteristics to determine how well they
meet the performance requirements as dictated by the program plan.
This analysis should include an evaluation of such specifications as
instrument accuracy, resolution, response time, range, minimum
detectible limit, environmental constraints and data output configura-
tions as imposed by the program requirements.
121
-------
Implementation of such a review may be best accomplished by a
panel of several people who have collective experience and expertise
in the areas as dictated by the final program plan. It should include
people knowledgeable in the individual instruments, data systems, and
the utilization of such systems in the field environment.
2.4.2 Instrumentation Calibration Techniques Review. Prelimi-
nary to the actual field operations, a review of the calibration
techniques should be conducted. This should consist of a review of
the calibration techniques and procedures as proposed to be used by
the field unit for each individual type of measurement and data
system. Information should include the frequency of calibration,
type of calibration, traceability to an NBS standard, where appli-
cable, and a functional diagram of sufficient detail to allow
understanding of techniques employed and identification of major
components; a calibration procedure document including an example
calibration formation record form, should be included.
Submission and review of these documents should be sufficiently
early in the program to allow for revision as required.
2.4.3 Review of Operation of Field Systems Procedures. A
field program of the type under consideration may impose several
requirements on one or more of the measurement systems. A proposed
field system operational plan should be developed which would de-
lineate the expected deployment of instruments, indicating the
expected range of environmental conditions, operating modes, use time
or duty factor, primary power supply voltage and frequency range
expected and the calibration and maintenance schedule.
A review should be made of the expected degree of degradation
that these combined factors may have on data quality. Experience in
operation of similar instruments under the expected operational condi-
tions is needed to make this assessment.
The effect of electromagnetic interference should also be in-
vestigated. Such things as high power radio stations and radar
installations can cause noise in the recorded data. This noise may
be coupled into the system via electric field interference in high
impedence circuits and/or magnetic field coupling through long data
or signal lines.
Recognition of this problem and logging of operational infor-
mation can be very useful in identifying these artifacts in the data.
122
-------
2.4.4 Measurement System Audits. In-field audits of all
measurement and data systems are recommended. Audits should be
conducted for both the sensor outputs and key status measurements
(e.g., flow rates for collection devices). The frequency of the
audits depends on the desired degree of independent assessment of
the data quality commensurate with the type of instrument being
audited.
Audit procedures for most of the gaseous analyzers, such as SO,,,
NO-NO , and 0,,, have been developed by EPA. A considerable amount of
information is available to guide the designer in the expected
accuracy and stability of some of the instruments. Audits of meteor-
ological and remote sensor measurements systems are also necessary for
this program aud some reasonable effort should be expended in review-
ing and improving audit procedures for these devices. For some of
these sensors "injected signal" techniques should be used where over-
all performance cannot be verified.
An audit of the data acquisition system should be carried out to
determine the correctness of the transfer of the output signal from
the respective measurement systems through data processing. This may
be accomplished by monitoring the output of the respective sensors
for selected increments of time and then comparing these values, which
have been independently corrected for calibration functions, etc., and
then comparing these data to the output at the central data processor.
This allows also for the checking of the proper applicability of all
calibration equations and correction factors.
If any unit fails to respond correctly to an audit, an appro-
priate notification must be made to the Field Director. Every effort
must be made to ascertain the cause of the failure and to document
the circumstances, duration of the failure (if known), and procedures
to correct the data, if possible.
2.5 Special Considerations for Remote Sensors
The following material is keyed to the outline of the panel
recommendations. It consists of recommendations for the special
consideration that must be given to the data from remote sensors such
as radar, lidar, and acoustic sounders. In general, each remote
sensor has its own data acquisition system and will provide a
computer tape (probably 9-track, phase encoded, 1600 bpi) on which
will appear files containing calibration runs, background information,
and data.
123
-------
In the case of ordinary lidar, this data will be backscatter
coefficients as a function of range, perhaps with separate files for
separate wavelengths and separate polarizations. One or all of these
may be present depending on the files. The lidar data must be pro-
cessed to remove the effect of the inverse square of the range, all
drifts must be removed, the zero point identified, and the calibration
factor applied. Programs exist to perform these functions. They
can be applied by the lidar crew outside regular working hours or the
programs may be adapted to use on the field director's computer.
Because of the differing formats for tape storage, it is probably
most useful to use the field director's computer. Hard copies of
graphs of a few shots can be produced quickly by the lidar crew for
evaluation of plume spread. After the above, processing data may be
displayed, averaged, or otherwise processed.
A Doppler radar (e.g., FM/CW) produces information on back-
scatter coefficients versus range and the radial velocity versus
range. The data is displayed in the field and digitized and placed
on magnetic tape. Data processing programs exist and can be adapted
to the field director's computer. Increased information on three
dimensional velocity fields can be obtained by using dual-Doppler
radars. In this case, the outputs of both radars must be combined in
a computer; probably the field director's computer must be used
because the existing radar computers don't have this capacity.
Acoustic sounders also produce magnetic tapes containing infor-
mation, on repeated scans, on the backscatter coefficient vs. range,
and the Doppler shift versus range. Probably some development of
programs is required here because the field director's requirements
for display must be considered. The acoustic sounder data is often
usefully displayed in a facsimile form to obtain an immediate view of
the structure of temperature inhomogeneities. Extraction of acoustic
Doppler information is available to present wind velocity vs. height.
2.5.1 Objectives. The objectives of data management for the
remote sensors are essentially the same as for other data sources.
However, remote sensors are often called on for data to guide the
field director in the choice of operating parameters at his disposal.
Some of the data must be available in real-time and most of the data
must be available for review within twelve hours in a form from which
good, but not necessarily final, estimates of relevant parameters can
be made.
124
-------
2.5.2 Requirements.
a. Phase I - The Small Hill Experiment. Probably, at the
most, one remote sounder will be present, and only a small computer
facility under the field director (part of the remote sensor system).
In this case the remote sensor computer must produce hard copy graphs
or photos or charts or tables of its output for use by the field
director as well as a complete tape record of its data. A sampling
of theses must be available in essentially real-time to the director.
More complete processing must await transmittal to a central facility.
b. Phase II - Full Scale Experiment. Probably one or two
lidars, one or two radars, and one or more acoustic sounders will be
employed. The additional complexities of this experiment, and its
multiple objectives (depending on weather conditions and where the
plume is going) require the same essentially real-time availability
of remote sensor data to the director as in Phase I.
2.5.3 Data Review Support. Each remote sensor data acquisition
system must possess the capability of automatic rejection of data
(non-recording) when the predictable mishaps occur, such as misfires,
intolerable power fluctuations or other debilitating failures. This
minimizes bad data on tapes. Each remote sensor system must periodi-
cally test for correct operation in such a way that the result shows
up on the data tape and is identified as a calibration file. The
results of these tests must be examined by the computer during
processing. The remote sensor operator must also supply standard
tests to apply to data while processing. Whenever possible, known
physical measurements that test entire systems should be employed.
Error estimates must be supplied by the remote sensor crew.
3. OFF-SITE REQUIREMENTS
3.1 Introduction
The general requirements off-site consist of provision for final
reduction, completion (i.e., merging, organization, storage, re-
trieval), final quality assurance checks, and data products useful
to the model development and evaluation community. The complexity
of this activity is direcctly proporational to the overall complexity
of the experiment and what is accomplished in situ, while it is
preferred that some final reduction (i.e. , conversion and validation)
and some compilation is performed in the field, it is recognized that
125
-------
some may be required at an off-site facility. Clearly, a large off-
site facility (e.g., EPA's Univac 1110) is needed to store the final
archive and thereby serve as a clearing house for future data
requests.
The field team should have the primary responsibility for the
initial off-site activities because, simply stated, they are most
familiar with the data and hence in the best position to validate
it. Their participation should continue through the establishment
of the final archives.
3.2 Compilation
The compilation phase of data management involves accession of
all data collected during the study period and all associated conver-
sion, reduction and error checking. The first consideration is the
type and form of the collected data which may obviously be contingent
upon the availability of an on-site computer during the experiment.
The following is a list of the general data types — (1) continuously
monitored data which can be directly fed into the on-site computer or
collected on data tapes; (2) manually reduced data such as lab
analysis or pibal soundings; (3) self-contained systems such as
Lidar, Sodar or Radar; (4) equipment logs, calibration and audit in-
formation; and (5) preliminary analysis and error detection used
during the field planning review meetings especially if an on-site
computer was available. If an on-site computer was not available
during the experiment, some additional conversion equipment (e.g.,
A-D) may be required for data reduction.
The basic functions during the compilation phase are as follows:
1) Collect all data tapes, logs, etc., and associated documenta-
describing format and content of the data tapes. Enter
all marually collected data onto the computer.
2) Make backup copies of all original data tapes. label and
store these along with copies of associated logistical
documentation. An informative leader at the beginning of
each tape should be included.
3) Perform preliminary automatic error checks. Devise a
system to incorporate status, calibration, and log book
information. Develop preliminary error flagging schemes.
Correct obvious errors.
126
-------
4) Perform appropriate data conversions and reductions to
produce raw data in standard file formats. This results
in the first level of data archive.
5) Perform required time averaging and preliminary data
analysis to calculate intermediate parameters required by
the modelers.
6) Design and prepare second level archive which may include
merging of certain data sets.
7) Acquire or develop a data management system for simple
retrieval and display of the second level archive data.
8) Develop software to produce higher level archives and
data products.
In order to perform these functions it is necessary to receive
from the modelers a specification of the required temporal and spatial
resolution and a list of calculated parameters that are required as
input to the model. In addition to data collected during inten-
sive periods, two other types of data must be managed. First,
climatological data may be collected for a prolonged period. Pro-
cedures must be developed to verify and archive this long term data
set. Finally, an effort should be made to obtain all data available
from other studies which may shed some light on the complex terrain
problem. A method of archiving and retrieving this data should be
developed to provide easy access for the modelers. A special effort
should be made to obtain and integrate data from the DOE and EPRI
studies.
The data management and quality assurance panel recommends that
the contractor provide EPA with all original data and at least two
forms of the archieved data and associated software so that EPA may
act as a clearing house for distributing the data to modelers and
other contractors.
3.3 Final Checks and Analysis^
One of the principal components of the data management task is
to assemble and guarantee a valid, reliable data set to be used by
the model developers. Inherent in these tasks are error checking
and analysis procedures at three fundamental levels in the chain from
field measurement to model development. These levels include
1) Analysis in the field (see Section 2.3), 2) Analysis by the data
manager, and 3) Analysis by the modeler after the data set is
127
-------
final. In this section some automatic, computerized, and analyst-
invoked error-checking procedures contained in the data base manage-
ment system (DBMS) are described. The DBMS should include a modular
set of error checking procedures from simple range checks to sophis-
ticated numerical techniques while maintaining a human decision
factor in any data modification. The system should contain a formal
process for updating the data base and reporting all modifications.
Separate procedures can be developed for single and multiple
sensor error checking.
3.3.1 Single-Sensor Error Checking. There are five basic error
checks for individual sensors:
Range Checks
Rate of Change Checks
Calibration Analysis
Outlier Detections
Time Series Forecasting Techniques
The first two can be computerized to provide an automatic method for
flagging potential errors in single-sensor data. Range checks include
searching for both certain and possible errors: For example, include
negative air quality concentrations. Possible errors can be deter-
mined by comparing the measurements with established climatological or
"reasonable" values. For example, solar radiation data should follow
a diurnal cycle and exhibit a reasonable daytime maximum. Similarly,
maximum daytime temperatures at a particular locale should not exceed
35 degrees C, for example. The temporal rate of change of selected
variables can be compared to probable maximum rates, e.g., a tempera-
ture change of 10 degrees/hour. Similarly, most operational instru-
ments display certain small random changes. If these are not
exhibited in the data, it could indicate a stuck or frozen instru-
ment, excluding those with natural constant states, e.g., zero air
quality measurements.
Analyst invoked (manual) error checks could include an examina-
tion of the calibration history of an instrument. Rapid changes in
calibration constants are probably indicative of malfunctioning
instruments. Unfortuantely, it may be necessary to apply nonlinear
correction factors to recover valid data. Outlier detection pro-
cedures include polynomial or spline curve fitting to the time
128
-------
varying data points. Most real aerotnetric data samples should
reasonably approximate the predicted curve (Figure 11). Outliers from
the predicted curve could indicate incorrect samples which could be
corrected by the "predicted" values. (Such corrected values must be
flagged accordingly in the data record). Time series forecasting
techniques include trace analyses (useful for determining drifting
calibrations) and principal component analysis. Table 2 summarizes
some of the available single sensor error checks.
TABLE 2. SINGLE-SENSOR ERROR CHECKS
Automatic
Range ^ > Q
Rate of Change JZ < 20°C/hr
ot
Manual
Calibration Spline Fits
Outlier Detection Polynomial Fits
Time Series Fourier Fits
Baseline Offset
Trends
3.3.2 Multiple Sensor Error Checks. Again, multiple sensor
error checks can include both automatic and analytical procedures.
Some of the automatic checks include a comparison of pairs of data
to established identical or probable physical relationships, e.g.,
N0x = NO + N02 ; T _> TQ. Statistical relationships derived from
climatological data can be used to isolate low probability events
that can be flagged for further examination. Manual checks of
multiple sensor data can include graphs, tables, scatter diagrams,
line plots, contour diagrams and perspective plots. These simple
exhibits for visual pattern recognition will enable the rapid evalua-
tion of the data.
129
-------
Data Point
Curve Fit
Variable
Time
Figure 11. Example of outlier detection.
130
-------
3.3.3 Update Procedures. The simple automatic analysis-invoked
error checks should be designed to detect possible errors. Only
the analyst should be allowed to actually change a suspect value.
Each change should be identified in a report and in the archive,
as follows:
New Value
Date/Time of Change
Old Value
Reason for the Change
Values that are suspect and not changed in the archive should be
flagged (according to a prescribed code) for the modelers' informa-
tion. The updated report then becomes a permanent part of the data
archive.
3.4 Data Products
There will be three basic data products coming from each experi-
ment including an archive of raw data, a model development (and
analysis) data set and a summary report describing the data. The
data archive will be a corrected and edited compilation of all data
acquired in the field program. It will be available in total and in
subsets without further processing to meet requirements not met by
the model development data set. The model development data set will
include such parameters as are needed for model development, evalua-
tion and diagnosis, which are specified prior to and in the course
of the program. The parameters will be averaged over specified times
and vill include values derived from combinations of archived values,
such as plume spatial distribution parameters a and o and various
stability parameters. The summary report will describe the experi-
ment, the types of data available, the accuracy of the data (as
indicated by audits and parallel measurements), and a quantitative
summary of the data available in each intensive period.
Another data product will be processed data from other field
studies. It is desirable to prepare data from other field studies
at least for model evaluation if not also for model development
efforts.
3.4.1 Data Archive. An archive of raw data should be maintained
so that it is available for special purposes not foreseen in the
131
-------
present study. Selected subsets for designated time periods or ob-
servation systems should be available on request. The volume of
data can be condensed by eliminating useless data, e.g., lidar
scans which do not intersect the plume and which are not required for
background data. The archive will be edited and corrected for instru-
ment calibration and audit results so as to contain the best possible
measured values. The contents and organization of the archive should
be described in the summary report.
3.4.2 Model Development Data Set. A model development data set
should be derived from the archive to meet the specified requirements
of model developers and model evaluators. This data set will be
significantly reduced in volume from the data archive to be convenient
for data handling purposes. The parameters will be averaged and
otherwise characterized over the period of interest. Specified
parameters will be derived; e.g., ground lidar observations could be
transformed into cross-sections of hourly mean back scatter for fixed
distances from the source. Each cross-section could also be charac-
terized by a center of mass position and and values. Similarly,
wind observations can be transformed to mean values and specified
turbulence statistics. Any number of stability indices can be com-
puted and placed on the file. It's desirable that modelers be given
parameters expected to be useful to fit a priori model concepts and
to be useful to diagnose why the expected concepts do not work.
A program must be developed to create the model development
set. In view of the complexity and variety of types of data anti-
cipated for use in the planned programs, it is recommended that on
the order of 6 person-months be allowed for development of this
program. A modification of the program used to generate the model
development data set could be used to process data from other field
studies in complex terrain. The primary modification would concern
interfacing with different data formats.
3.5 Summary Report
A summary report should be prepared to describe the data in order
to help modelers and others who deal with the data to select portions
of the data and to comprehend its scope. The report should include
at least the following
• DESCRIPTION OF THE EXPERIMENT
• HISTORY OF FIELD ACTIVITIES
• TYPES OF DATA RECORDED
132
-------
• SAMPLE DATA RECORDS
• SUMMARY OF INTENSIVE OBSERVATION PERIODS
• SUMMARY OF MODEL DEVELOPMENT DATA SET
• CONTENT AND ORGANIZATION OF ARCHIVE DATA SET
• DESCRIPTION OF QUALITY ASSURANCE, EDITING AND DATA
REDUCTION PROCEDURES
• ESTIMATES OF UNCERTAINTY AND ACCURACY ASSOCIATED
WITH EACH MEASUREMENT
The effort required to summarize data is highly dependent on the
volume of data required. However, the summary is not expected to be
a major part of the data management and quality assurance effort.
Perhaps 5 percent of the overall data management effort should be
allocated to generating a summary.
4. COSTS
Costs are proportional to the complexity of the experiment,
requirements of the field crew, requirements of the model developers
and evaluators, and the number of intensive field periods. For our
purposes, we have assumed an upper-bound of likely requirements for
two field study periods. It should be noted that once the data
management software and QA procedures are established for one ex-
periment, costs are much less for a second similar experiment.
In general, we anticipate that 12 to 13 person-years will be
required for the two phase approach suggested by various panel
members. Personnel and support costs (i.e., travel and subsistence,
report reproduction, miscellaneous supplies) could consume about one
million dollars. Taking into account the need for a field data
facility (about $130,000) the above costs are consistent with general
estimates that data management and QA costs each account for about 10%
of the total budget. If Government Furnished Equipment is available
for the field data facility, the above $130,000 data facility
estimate can be reduced accordingly.
5. OTHER RECOMMENDATIONS OF THE PANEL
During the course of the Workshop the panel became quite con-
cerned that the experiments would not satisfy the primary objectives
133
-------
to improve modeling in a complex terrain setting. Specifically, a
better understanding of plume impaction is desired as a first
priority. Towards this goal, field studies are planned to collect
the necessary data which will enable us to refine existing or develop
new and improved model formulations.
Other panels have presented details and recommendations on the
above topics. They describe alternatives to achieving the stated
objectives. However, our panel would like to address some of the
issues. In particular, we would like to express the need for a
consistent approach in the follwing two areas:
1. The measurements should be consistent with the need to
improve the model in settings of complex terrain. Therefore, we
need measurements to verify model performance (observed air quality
concentrations) and to verify our estimations of certain inter-
mediate derived parameters. In particular, if we are considering
Gaussian models, there is a need for measurements of the vertical
and lateral spread of the plumes as a function of downwind distance
(or time). There is also a need to measure the flow characteristics
of the plume in the vicinity of the terrain obstacle as well as
upwind and downwind.
Flow visualization is one means of describing the above
parameters and is essential in our opinion. If we rely solely on
measurements of ground-level tracer concentrations and conventional
meteorological parameters, we fall into the pattern of many previous
field experiments whereby we obtain answers for a specific regulatory
problem but fail to sufficiently enhance our understanding of the
basic processes that are occurring.
2. The scale of the experiment shoud be consistent with our
ability to measure the required parameters. We support the concept of
first conducting a simple experiment, fixing certain parameters (i.e.,
height of the plume, source strength, and characteristics of the
obstacle). However, having gone to this trouble, it seems only
appropriate that we adequately measure other non-fixed parameters to
the extent practical. In this respect, the scale of the experiment
should be consistent with measurements needed to parameterize the
model input parameters (e.g, meteorological tower measurements),
verify model predictions (e.g. , ground-level tracer concentrations),
and verify parameterizations within the model (e.g., visualization
of plume spreading and flow patterns).
3. It is essential to develop a method of routine meteorologi-
cal monitoring to characterize plume dispersion in rough terrain. In
134
-------
principle, wind speed and direction, turbulence data (vertical and
lateral), and mixing height are sufficient (assuming adequate spatial
resolution). Of course, there is the problem of relating turbulence
data to the plume dimensions. The horizontal wind direction varia-
tion, aQ , may be appropriate for quantifying the lateral plume
dimension, normally quantized as a . For quantifying the vertical
plume dimension, a we may wish to consider measures of the vertical
wind variation such as , or more likely indicies that take into
account the vertical differences in wind and temperature (e.g.,
Richardson number). In complex terrain, the above parameters will
be more difficult to measure adequately. Hence, one of the principal
outcomes of the experiment should be recommendations of the number
and type of meteorological monitors to be used in routine complex
terrain assessments.
135
-------
APPENDIX A
PRE-WORKSHOP MATERIAL PREPARED BY EPA
During the early stages of planning for the Workshop, EPA per-
sonnel in the Meteorology and Assessment Division in Research Triangle
Park began preparing scenarios on various field study approaches.
The first "strawman" plans were prepared by Mr. George C. Holzworth,
Chief, Geophysical Research Branch, and Dr. William H. Snyder, Chief,
Fluid Modeling Section; Holzworth's plan addressing a full scale
measurement program and Snyder's plan a small (100 m) hill experi-
ment. Subsequently, both scenarios were combined into one "straw-
man" plan and distributed to the Workshop participants for their
consideration.
A third document was prepared by Dr. Robert G. Lamb of the EPA
Atmospheric Modeling Sciences Section immediately prior to the Work-
shop. It was distributed during the Workshop for consideration with
the other scenarios.
These "strawman" plans are presented below in the same format
as distributed to the participants for use as reference material to
the panel statements and recommendations presented in Section 5. They
should be considered strictly in light of their original intent,
namely to provide the panel members with specific ideas and sugges-
tions for deliberation and not as documents intended for publications
which have been subjected to editorial reviews and corrections.
136
-------
PROGRAM PLAN FOR DEVELOPMENT OF A MATHEMATICAL AIR QUALITY
ASSESSMENT SYSTEM FOR USE IN COMPLEX TERRAIN
BY
GEORGE C. HOLZWORTH AND WILLIAM H. SNYDER
1. INTRODUCTION
Mathematical models of atmospheric dispersion are relied upon
heavily to evaluate compliance with air quality standards and preven-
tion of significant deterioration. Although the economic consequences
of model results can be very large, it has been difficult to demon-
strate that available models have a reasonably high degree of relia-
bility, especially in regions of complex terrain. This chronic prob-
lem is becoming acute as energy development and utilization expand in
mountainous areas. The total problem is overwhelming, not only from
its complexity of meteorological conditions and terrain configura-
tions that occur, but also from the expense of obtaining adequate
experimental data.
The Meteorology and Assessment Division (MD) of EPA's Environ-
mental Sciences Research Laboratory has been charged by the EPA with
the development and execution of a five-year, multi-million dollar
program designed to produce atmospheric dispersion models that are
applicable to large sources in complex terrain and that have a demon-
strated higher degree of reliability than existing models. Current
(preliminary) plans for carrying out this charge are contained in this
document, which will be presented to a Workshop of invited experts
during July 16-19, 1979. The Workshop is directed to make recommenda-
tions for improving this plan. Some components of the program plan
have been adopted from the recommendations of the ERDA Workshop on
Research Needs for Atmospheric Transport and Diffusion in Complex
Terrain (Albuquerque, 1976).
2. PROBLEM
Reliable mathematical formulation of transport and diffusion of
atmospheric pollutants has proven to be an extremely difficult task.
For example, for individual plumes emitted under the relatively
simple and optimum situation of flat terrain and steady meteorological
and emission conditions, model validation of ground-level, maximum,
one-hour average concentrations at downwind distances to several
kilometers is generally within a factor of two. But for complex
137
-------
terrain settings, where many large individual sources are locating,
simulation model results sometimes are not within a factor of ten of
measured concentrations. This happens because of (1) the distortions
in air flow, (2) the distortions in turbulence that occur in the
vicinity of terrain obstacles of all sizes, and (3) because terrain-
level concentrations are highly sensitive to the height above under-
lying terrain of the maximum concentrations in the plume (i.e., at
plume center line). These three terrain effects are basic to the
overall problem.
The first basic effect of irregular terrain on atmospheric dis-
persion (i.e. , transport and diffusion) deals with the transport,
knowing where the plume is located in three-dimensinal space and
time. While it is obvious that air must go around or over obstacles
(the latter typically during stable atmospheric conditions and the
former more readily during unstable/neutral conditions), terrain
features also commonly influence the local circulation through the
diurnal cycle of heating and cooling that occurs at the ground-air
interface. During daytime, upslope flow is enhanced over sun-facing
terrain and downward motion is favored over shadowed areas; during
non-cloudy nights, downslope flow is enhanced over elevated terrain
in association with the development of strong temperature inversions.
Under weak large-scale flow conditions these ground-based flows,
which vary in depth from tens to hundreds of meters, induce counter-
flows aloft. With moderate large-scale conditions the flow aloft
resembles that of the large-scale, and with sufficiently strong
large-scale conditions, the low-level, thermally-induced flows are
likely to be overwhelmed or markedly modified by the large-scale
flow. In addition, during certain conditions with critically fast
winds, terrain features can induce aerodynamic effects which distort
the usual three-dimensional flow in the vicinity of the terrain
obstacles.
The second basic effect of irregular terrain on atmospheric
dispersion concerns the extent to which turbulence and diffusion are
altered from their values over uncomplicated terrain. Recent limited
investigations in mountainous regions have demonstrated that the
Gaussian diffusion parameter values, which are often expressed as a
function of the Pasquill-Gifford stability classes, are generally
larger over complex terrain than over flat terrain, especially
during stable conditions and at relatively short downwind distances
from the plume source. But these differences are not necessarily
in the same proportion for the horizontal and vertical components
of diffusion. The appropriate diffusion parameter values likely
depend upon the details of the complex terrain, not only over the
138
-------
terrain beneath the plume but also upwind of the plume source. Fur-
thermore, the diffusion parameter values are expected to vary with
height above the terrain, at some level decreasing to the corres-
ponding values for flat terrain.
The third basic effect of irregular terrain on dispersion is in
reality a combination of the first two; it concerns the small-scale
features of plume interaction with a terrain obstacle in the immedi-
ate vicinity of the obstacle surface; it involves transport and
diffusion, but for practial purposes at some close distance of
plume approach to obstacle surface the transport becomes indistin-
guishable from diffusion. This effect is particularly important
because of the frequency with which plume center lines (where con-
centration gradients are relatively large) abruptly approach terrain
obstacles. For example, a fundamental issue among modelers is the
extent to which plumes, that are well above the ground where they
are emitted, interact with higher terrain at downwind locations.
Various models (e.g., Valley, Cramer, ERT) that have been adapted/
developed for use in complex terrain handle this situation by main-
taining the plume center line at 1) a constant height above sea
level, or 2) a constant height above the underlying terrain, or
3) a variable height above the terrain that is some fractional value
of the height difference between the terrain at the source and beneath
the plume trajectory; the use of these techniques usually depends
upon atmospheric stability. For the case of a plume within a tempera-
ture inversion, technique (1) results in relatively very high concen-
trations, technique (2) ignores the terrain, and technique (3)
produces intermediate results. The technique that is most appropriate
remains controversial because adequate, full-scale field measurements
have not been made.
Some additional questions that may be posed about terrain effects
on dispersion are:
1. What are the maximum concentrations on a terrain obstacle
that is in the apparent path of and at a higher elevation
than a plume embedded in an unstable or neutral layer, with
or without an inversion aloft?
2. How do terrain features affect "diffusion coefficients",
a and a ?
y z
3. In what manner does complex terrain affect mixing heights?
4. How does complex terrain affect plume rise?
139
-------
Answers to those questions are likely to depend upon the wind fetch
over irregular terrain, the details of terrain features (e.g.,
variously inclined hills and ridges, individually and in combination),
atmospheric stability, wind speed, and plume height with respect to
that of terrain features.
Clearly, the total problem of understanding and modeling atmos-
pheric transport and diffusion over complex terrain is extremely
difficult, but not unsolvable. With the resources that are emerging
to support such a research program and the wisdom of the scientific
community, we believe that a very significant impact on the problem
can be achieved over the next few years.
3. SOLUTION STRATEGY
At this time it seems inappropriate to tackle the total problem
in detail, but rather to focus on modeling individual phenomena. In
this approach each phenomenon modeled may be viewed as a module of a
Mathematical Air Quality Assessment System (ultimately, it would be
desirable to have a single equation appropriate to all phenomena
modules, but such an approach here appears premature).
As its first priority module the TEB proposes to model the one-
hour-average, ground-level concentrations that result from plume
impaction on elevated terrain. More specifically, we mean "fanning"
plumes (i.e., within a temperature inversion) that encounter a
terrain obstacle(s). This phenomenon has been difficult to model
satisfactorily and has occurred often in assessing the satisfaction
of air quality requirements. The focus is on maximum terrain-level
concentrations since typically they occur relatively closer to
pollution sources than lesser concentrations, and therefore are
easier to study and model. The model that is to be developed must
have demonstrated reliability and specified applicability; it must
be reasonable to apply in terms of computer resources and required
meteorological input information. However, because meteorological
conditions are more variable over complex than flat terrain, it is
likely that for a given degree of reliability, complex terrain models
will require more meteorological input than comparable flat terrain
models, and than currently is available in most real situations.
We propose to develop this modular model from the results of
field experiments which are essentially on two scales and which are
briefly described here:
140
-------
1. A small hill about 100 meters high in otherwise uncompli-
cated terrain would be highly instrumented to measure
meteorological variables and the concentrations from a
movable tracer source. This experiment essentially would
expand into the real atmosphere the scale of laboratory
fluid modeling that has been done by Hunt et al. (1978).
2. Full-scale field measurements would be made of meteoro-
logical variables and pollutant concentrations from an
operating power plant in complex terrain. Tracer studies
on terrain obstacles in the vicinity of the power plant
would be employed also. This experimental strategy is
intended to provide a direct link between idealized experi-
ments and actual operating situations.
It should be emphasized here that model production is the ob-
jective of this effort. While the required model(s) clearly cannot
be developed with inadequate experimental information, the necessary
experiments must be kept within bounds that assure adequate resources
for model development and evaluation. We believe that our modular
approach to the total problem provides a means of achievability within
available resources and it provides a logical basis upon which to
build if additional resources develop.
4. IMPACTION MODULE DEVELOPMENT PLAN
Since model development is paramount in this effort, the modelers
must be active throughout the project. They must specify requirements
for measurements early in planning experimental operations; they must
be involved in archiving and documenting the data to assure that
their requirements are being met; and they must be immediately in-
volved in the experimental data analyses in order to thoroughly under-
stand the results for effective incorporation into future experiments
and inclusion in the mathematical models that are under development.
A. Field Study Design
Experimental field studies will be carried out in three designs
in order to quantitatively establish the process of plume impaction.
1) Small Hill Tracer Study. This study will provide a tie-in
between the real atmosphere and the laboratory, scaled, fluid modeling
studies that have been done by Hunt et al. (1978). Meteorological
conditions and the impact of an inert tracer plume will be measured
on a three-dimensional hill, no higher than 100 meters and located in
141
-------
otherwise flat terrain. The tracer (e.g., ethane, SFg, etc.) will be
dispensed from a portable tower or tethered balloon. About five 150-
meter towers, instrumented at several levels, will be placed around
the hill and a shorter tower will be positioned on top of the hill.
All monitoring data will feed directly to a central minicomputer for
immediate analysis and real-time graphical display so that source
location, height, and strength can be altered to achieve the desired
impaction characteristics. This study is designed to enhance the
acquisition of data by simplifying the logistics and terrain com-
plexities; it differs from the laboratory studies of Hunt et al. by
necessarily including the turbulent boundary layer next to the hill
surface that occurs in real situations.
Meteorological measurements would be made of the three-dimen-
sional wind, temperature, and other variables that the modelers may
specify. It is estimated that tracer samples will be required at
about eight levels on the hill.
The hill size required for this experimental study and for
general applicability of the results is not believed to be important,
except the hill should be completely immersed in a very stable layer,
the hill should be large with respect to the surface boundary layer
depth, and the ratio of hill base-diameter to height should not exceed
eight. Theoretically, the flow structure will be dependent upon the
Froude number
where U = wind speed, h = height, g = gravity, and T = temperature.
If the vertical temperature gradients and velocity profiles are
linear, which frequently occurs, the Froude number is independent of
hill height. Thus, the entire range of atmospheric stability varia-
tions may be obtained by performing the experiments on a single hill,
one perhaps even smaller than 100 meters high.
We believe that a large and adequate amount of experimental data
can be obtained during about three months of field operations after
the equipment is set up.
2) Full-Scale Plume Study. This study is designed to link
the idealized, small-hill, experimental results with those focused
on the plume of a large operating power plant. The principal pollu-
tants to be measured will be S02 and total particulates, including
sulfates, which will permit calculation of the sulfur budget. SOp
transformation over travel t^mes up to a few hours, however, is not
expected to be significant. While pollutant transformations and
142
-------
depletions are important factors to be included in comprehensive
models, they need not necessarily be studied in complex terrain,
and are under study elsewhere. The measurements of total par-
ticulates and particle-size distributions in the plume will be
especially useful in conjunction with lidar measurements of plume
configurations. Measurements of meteorological variables, S02 and
particulate concentrations will be made from ground-level and aloft
through the plume, using sophisticated (but currently available) and
conventional sensors carried on mobile and fixed platforms as neces-
sary to define plume configurations and meteorological conditions.
For ease of operation it is desirable that potential plume impaction
terrain obstacles be relatively close to the power plant, preferably
within 25 kilometers. Candidate power plants for this full-scale
experimental study should be screened carefully for the desired
terrain features and meteorological conditions. In addition, the
operators of the power plant(s) selected for study must be willing
to cooperate completely since pollutant emission rates, effluent
temperature and volume-flow rates, other chimney/plant character-
istics, and meteorological conditions in the immediate vicinity of
the plant must be known continuously during field measurement
experiments.
S02 may be measured remotely by the Barringer correlation
spectrometer method and directly by readily available, EPA reference
methods; particulate aerosols can be measured indirectly by lidar
and directly by light scattering (nephelometer) as well as by con-
ventional filtering techniques. Although accurate measurements of
particulate concentrations over short-time intervals are difficult,
the main reason for requiring them is for tie-in to lidar measure-
ments of plume configuration.
Meteorological measurements will require several towers ex-
tending to about 50 meters. Besides the usual measurements of trans-
port and turbulence, the fluxes of heat, momentum, and moisture would
be determined. Indirect measurements of wind fields would be made
by doppler acoustic or laser techniques. Frequent double-theodolite
pibal and temperature soundings would be required. Consideration
would also be given to the use of acoustic sounders to record growth
and decay of nocturnal temperature inversions. Several aircraft
would be used not only for plume sampling and meteorological measure-
ments, but also for servicing and transporting mobile ground-based
equipment.
This full-scale study will be conducted intensively during four-
to six-week periods of at least one winter and summer season and at
143
-------
other power plants in succeeding years as necessary and dependent on
available resources.
In conjunction with the full-scale plume measurements, tracer
studies will be conducted in order to optimize the measurements of
impaction on terrain obstacles in the vicinity of the power plant.
These experiments will also be used to resolve complex, site-specific
questions about modeling assumptions and to broaden the range of
applicability of the results. For the most part we expect these
tracer experiments to be carried out when transport conditions are
inappropriate for full-scale plume measurements, although in some
situations they may be extremely useful in augmenting the full-scale
measurements.
B. Laboratory Simulation
Careful consideration needs to be given to the further use of
laboratory scaled (physical) modeling techniques to complement the
field studies and model development. This relatively economical tool
can provide essential detail and data for correlation with and exten-
sion of computational model development, field investigation, and
definitions of basic aerodynamic characteristics. For example, it
would be extremely useful to model specific terrain features in the
vicinity of the plant(s) selected for study, as well as in the
vicinity of other plants. Such laboratory modeling would also be
used in developing mathematical modules of other important plume
phenomena such as "fumigation" and "trapping".
This effort could be performed at ESRL's Fluid Modeling Facility
at Research Triangle Park, especially in view of that group's contri-
butions and on-going interest in the overall problem as well as the
administrative convenience of being co-located with the MD.
C. Potential Field Sites
From practical considerations it will be desirable to conduct
the small hill tracer study in the vicinity of the power plant
selected for full-scale study, although this approach should not be
pursued to the extent of compromising the basic requirements for the
small hill experiments. In any case, it is believed that an appro-
priate site can be found without great difficulty.
For the full-scale plume study, the specific details of the
experiments would be designed around the power plant that is selected
for study. A prime factor in this selection must be the willingness
of the plant operators to cooperate. At the present time, because
144
-------
of anticipated energy resource development and utilization in the
West and because of presently operating plants that are relatively
isolated from one another, a western plant is favored for the initial
study.
As points for discussion, two possible power plant sites are
described. However, their potential for plume impaction on elevated
terrain should be studied further. The Navajo Power Station near
Page, Arizona, is an example of a power plant on a plain with nearby
prominent terrain features. With a capacity of 2250 MW, Navajo has
three 236-meter stacks that emit about 200 tons/day of 302. The plant
elevation is 1300 meters MSL, and there is a variety of terrain ir-
regularities within 30 km. At their closest point about 22 km west,
the steep Vermillion Cliffs and the plateau behind them rise to around
2000 meters; the cliffs trail off to the southwest and northwest for
10-15 km. There are numerous singular peaks fairly close to the
plant, e.g. , Lechee Rock reaches 1800 m at 7 km to the southeast and
Tower Butte is almost 1600 meters in elevation at 10 km to the north-
east, as well as other obstructions at greater distances. The terrain
in this region is so rugged that many locations cannot be reached by
surface vehicles. This site has good potential for studying plume
impaction as well as other types of plume interactions with individual
terrain features.
The Mohave Station, on the Colorado River near Bullhead City,
Arizona, is an excellent example of a plant in a very deep and broad
valley. This plant has a capacity of 1500 MW and emits about 44 tons/
day of SOg from a single 150-meter stack. The plant elevation is 200
meters MSL and the terrain generally slopes up to the east and west
of the north-south Colorado River, reaching 900-1200 meters within
15-25 km. There are also numerous smaller terrain features within
30 km of the plant. The weather at this site is dominated by a
diurnally varying mountain-valley circulation, which could serve as
the focus for field experiments. With suitable large-scale flow,
aerodynamic effects on the plume are expected. During summer, tem-
peratures well above 100°F are common at Mohave and likely would
cause some problems for field operations.
It should be mentioned that both these sites are in the southern
Rockies where moisture is sparse so that precipitation depletion of
the plume and SOg tranformation (as enhanced by high humidity) would
be small. Furthermore, the infrequent occurrence of precipitation
is highly desirable for field operations.
The question arises whether to conduct the complex terrain plume
study intensively at only one location for the study duration, or
145
-------
whether to progress to a second (and maybe third) site based on
terrain peculiarities. While it would be highly desirable to operate
at two or three different sites in order to increase the general
applicability of the models developed, resources might not be suffi-
cient to allow this portability at the anticipated funding level.
D. Model Development and Testing
For emphasis, it is mentioned again that the models to be de-
veloped must be reasonable to apply in terms of required computer
resources, input meteorological and plant operations data, and other
information, and the models must have demonstrated reliability. Since
production of mathematical models is the basic objective that drives
this project, the modelers must be active in all phases. They must
prescribe the initial modeling approach and the required input vari-
ables (including any new but practical concepts); they must partici-
pate in field study designs so as to build upon earlier results; they
will be responsible for field data analyses/interpretation and for
incorporating them into the modules of the air quality assessment
system that they are developing. The models will be tested for sensi-
tivity to dependent parameters, validity of calculated concentrations
and transferability to other complex terrain settings. In addition,
comparisons will be made to "off-the-shelf" models (e.g., EPA's Valley
and CRSTR models and Environmental Research and Technology's variable
height model), as well as to other models under development, in order
to demonstrate modeling accomplishments.
At this point in time it seems unlikely to expect the emergence
of basic new modeling concepts. Rather, the Gaussian distribution of
time-averaged concentrations within the plume is expected to persist,
but with distortions and modifications, due to terrain, somewhat as
modeled now but supported by experimental data. Perhaps the concen-
trations can be calculated in successive phases of dispersion as
effected by elapses/types of underlying terrain. For example, plume
impact ion on elevated terrain is likely to be a function of the maxi-
mum terrain-level concentrations within the plume just before it
significantly interacts with the terrain. Thus, the plume centerline
concentrations could be calculated in the first phase much as in
the usual manner and the terrain-level, impaction concentrations in
the second phase by consideration of the terrain characteristics.
Because of terrain effects, it seems clear that considerable
effort will be required to adequately model plume paths, particularly
at greater distances from sources. It also seems likely that more
input information will be required to satisfactorily apply the models
than is normally available in complex terrain. In terms of input
146
-------
data, consideration should be given to using them to generate reli-
ability factors in the model applications. For example, wind fluctua-
tion and/or other meteorological measurements, taken over appropriate
averaging times, may be indicative of the reliability of the estimated
plume location as well as the diffusion rate. Diffusion rates, as
functions of terrain features, height, and stability probably will
have to be determined experimentally as in the past.
E. Complex Terrain Modeling Workshop
During July 16-19, 1979, a Workshop (conducted under an EPA con-
tract with North American Weather Consultants) will be held in
Raleigh, North Carolina for the purpose of commenting and developing
further recommendations on .the program plan presented in this docu-
ment. About 45 participants have been divided into five panels:
1. Model Development and Analyses
2. Model Evaluation and Application
3. Experimental Design
4. Measurement Techniques
5. Data Management and Quality Assurance
The final report on the Workshop recommendations is due from the con-
tractor in September 1979, for subsequent publication and distri-
bution. An initial draft of the document is expected at the close of
the Workshop.
F. Schedule and Budget
A schedule of research milestones appears on the following page.
Notice that the modelers will be active throughout the project. The
small-hill impaction study will be conducted at only one site, as
presently conceived, although such studies of other plume configura-
tions might be warranted at a later time. Full-scale plume impaction
studies would be continued as necessary, although we believe that
after one year sufficient progress will have been made to focus on
other modules (plume phenomena) of the assessment system. The first
year's (FY-80) resources are expected to produce fundamental results
based on the small-hill tracer impaction study. Continuation into
the full-scale plume measurements will depend upon continuing re-
sources. Assuming funding of approximately 2 million dollars per year
for three years (beginning in FY-80; to be expended over about five
147
-------
SCHEDULE OF ORD RESEARCH MILESTONES
FOR DEVELOPMENT OF DISPERSION MODELS
IN COMPLEX TERRAIN
(ASSUMING RESOURCES BEYOND FY-80)
JUL 1979 - CONDUCT WORKSHOP TO GENERATE RECOMMENDATIONS FOR RESEARCH THAT
WILL PRODUCE SCIENTIFICALLY SOUND DISPERSION MODELS FOR SOURCES
IN COMPLEX TERRAIN
SEP 1979 - WORKSHOP REPORT FROM CONTRACTOR
DEC 1979 - ISSUE RFP BASED ON WORKSHOP RECOMMENDATIONS
APR 1980 - SIGN INITIAL CONTRACT CWITH OPTIONS TO BE EXERCISED DEPENDING
ON PROGRESS AND RESOURCES)
PLAN MOppy PFVFl OPMI^NT, FjFLD STUDIES. AND SCALED PHYSICAL
MODEL DEVELOPMENT AND TESTING
MODELING: DO
APRIL 1980
1985
CONDUCT SMALL-HILL IMPACTION TRACER STUDY
JAN 1980-DEC 1980
PUBLISH RESULTS (PERTINENT TO PHYSICS ASSUMED IN VALLEY
AND OTHER MODELS
JUN 1981
INITIATE FULL SCALE FIELD STUDIES PROGRAM
JUN 1980
CONDUCT FUL
-SCALE PLUME IMPACTION FIELD STUDIES
JAN-FEB 1981
JUL-AUG 1981
JAN-FEB 1982
PUBLISH INITIAL
RESULTS OF MODELING IMPACTION
APR 1982
JUL-AUG 1982
PUBLISH REFINED RESULTS OF MODELING IMPACTION
DEC 1982
CONTINUE FULL-SCALE FIELD STUDIES IN 6-MONTH INTERVALS. POSSIBLY
AT ADDITIONAL SITES AND FOCUSING ON OTHER IMPORTANT PLUME-
TERRAIN INTERACTION PHENOMENA
PUBLISH MODELING RESULTS AS SOON AS POSSIBLE TO COMPLETE
148
-------
years), we believe that significant progress in the development of a
complete model assessment can be demonstrated.
G. Management and Implementation
Responsibility for directing this complex-terrain, plume-
modeling study has been assigned to ESRL's Meteorology and Assess-
ment Division. A high-level steering committee will be formed to
provide overall guidance toward program goals, to monitor progress
at frequent intervals, and to provide a liaison between the program
and the user community. In addition, an in-house working group com-
posed of modeling and measurement subcommittees will be available
to MD personnel to help assess the validity of the modeling effort
and to assist in reviewing technical proposals and reports.
As presently foreseen the entire program plan will be carried out
under contract, probably with a prime contractor responsible for the
overall modeling effort and with provision for subcontracting.
Monthly progres reports from the contractor will be required and will
form the basis for reports/presentations to keep the steering
committee informed of program accomplishments and immediate plans.
5. DEVELOPMENT OF OTHER PLUME-TERRAIN INTERACTION MODULES
Although this mo del-development program plan has focused on the
important plume-dispersion phenomenon of impaction on elevated
terrain, it is appropriate to point out for completeness that models
for other plume phenomena may be developed through the same basic
approach; that is, through laboratory scaled physical modeling (as
appropriate), tied in with full-scale plume measurements.
Data collected in the full-scale field study will be useful for
this purpose. Reference to the schedule shows that the program is
designed to implement the small scale hill study first and to initiate
detailed planning for the full scale field study at the same time.
The full scale field study will be initiated in late FY-80 and will
overlap the conduct of the small scale studies.
The site selected for the full scale field study will be selected
to maximize the opportunity to incorporate the findings of the small
hill study and to maximize the opportunity to collect data useful for
developing other modules of the final model equation.
Two major areas of uncertainty by the presence of complex terrain
are the transport wind field which may vary significantly over small
149
-------
distances, particularly in the vertical direction and an altered tur-
bulence field which may drastically affect the dilution rates. Thus,
the long-term goals of the modeling program must consider a satis-
factory description of the transport wind and turbulence fields as
well as the joint frequency of occurrence of a wide range of meteor-
ological conditions and topographical configurations. The frequency
with which certain events occur, are of course pertinent to the
question of whether or not the Air Quality Standards are violated and
must therefore also be studied. A separate document has been prepared
by Dr. Robert Lamb which addresses the frequency question and will
be available to participants at the Complex Terrain Workshop.
150
-------
COMMENTS AND SUGGESTIONS ON THE PROPOSED COMPLEX
TERRAIN DISPERSION MODELING PROGRAM
BY
DR. ROBERT G. LAMB
1. INTRODUCTION
Regardless of whether one is concerned with flat or complex
terrain, the objective of air pollution modeling efforts is to develop
a means of predicting the second highest concentration that will occur
in any year from a source of given location and specifications.
Actually, no one can say what the highest or second highest concen-
tration will be at any site in any future year because the physical
processes that control concentrations, principally the weather, are
not deterministic. At best one can estimate the probability that a
particular concentration will occur based on projected source emis-
sions and historical weather data. Thus, in the discussions that
follow, we will view the pollution modeling problem from the pro-
babilistic standpoint.
One reason for doing this is to get away from the prevalent
notion that understanding the dynamics of plume impaction on hills is
the key to the complex terrain modeling problem. We will discuss
below two other equally important aspects of the problem that must be
resolved before the above stated goal of diffusion modeling can be
achieved.
Another motivation is to illustrate the magnitude of the overall
modeling objective. There are 8760 hours in a year so the task of
estimating the second highest hourly averaged concentration in a year
is essentially that of finding the value C2 that satisfies the
equation
f
(
p(C)dC ==
where p(C)dC is the probability that the hourly averaged concentration
at the given site is in the range C to C + dC.
It is evident from this expression that the stated objective of
pollution modeling is an ambitious one in that it requires a rather
precise knowledge of the form of the upper tail of the concentration
probability density p(C). Under the best of conditions they are
151
-------
limited to the accuracy with which the probabilities of the extrema
of any process can be estimated. Consequently, we should devote part
of any modeling effort to the question of whether, given the inherent
uncertainties with which we can describe the phenomena that effect
pollutant concentrations in complex terrain, we can ever achieve a
model of the second highest concentration that provides estimates
reliable enough to form the basis of highly costly emissions control
procedures. Stated another way, we should compare the magnitude of
model uncertainties, translated into dollar costs of control equip-
ment, with the cost of alternate methods of estimating the second
highest concentration, such as long-term tracer monitoring studies
conducted at prospective plant sites. The results of these types of
analyses might lead to the abandonment of complex terrain modeling
altogether.
In any case it is desirable to determine as soon as possible
(if it is determinable at all) the inherent limits on the predict-
ability of the probabilities of concentration extrema in complex
terrain. This knowledge gathered early in a multi-year model develop-
ment program could guide the redesign of those studies planned for
subsequent years and thereby circumvent the expenditure of large
amounts of money and manpower pursuing goals that are unattainable.
Thus, one criterion for the design of a long-term complex terrain
modeling program should be that the information required to analyze
the predictability question be gathered during the initial phase of
the study.
The remaining design criteria can best be formulated by examin-
ing the information needed to solve the equation that gives the pro-
bability density p(C).
For this purpose it is convenient to collect the physical pro-
cesses that affect transport and diffusion into several groups as
follows:
Firstly, we isolate the geostrophic (or synoptic scale) meteoro-
logical conditions over the given site at a given time. The important
variables are the wind speed and direction, temperature and dew point,
all measured at the geostrophic level (say 700 mb); cloud cover; and
elevation of the subsidence inversion base. We will denote this set
of variables by the symbol G and refer to it often in the probabilis-
tic analyses as the "event" G, to signify the occurrence of a specific
set G of these variables.
152
-------
Secondly, we isolate the micro-mesoscale meteorological condi-
tions at the site. The important parameters here are the three-
dimensional flow and temperature fields in a layer roughly 1 km deep
based at the source elevation and extending over the entire horizontal
region in which significant concentration levels are likely to occur.
We will represent these fields by the symbol M and refer to them
occasionally as the "event" M.
Thirdly, we consider the plume rise and buoyancy generated plume
spread processes. The three-dimensional coordinates of the plume
centerline; the plume's width, depth and material content as functions
of distance along the centerline; and the frequency and lateral extent
of plume meander will jointly be denoted by the symbol (and event) P.
Finally, we have the processes that govern plume impaction. We
are not interested in the details of these processes themselves, but
rather with the ground-level concentration they produce for given P.
Thus, we define the event C as the occurrence at the given site of a
particular concentration value, say C. The impaction processes are
implicit in C.
The four components of the problem we have just defined are
interlocked: a particular set G of geostrophic conditions occurs at
the site; these acting in concert with physical properties of the
local terrain give rise to a particular micro-mesometeorology M; in
this flow the power plant emitting material at a given rate and tem-
perature from a stack of given height and diameter produces a plume
of specification P; boundary layer processes bring this plume into
contact with the ground causing a concentration C. Thus, the task of
estimating the probability that a particular concentration C will
occur at some point is simply that of determining the total pro-
bability that all combinations of these four events that produce
the value of C will occur.
The probability that M, say, is in the range N to N + dM is
p(M)dM. Since M is dependent on G, p(M)dM is different if we are
given that a particular G has occurred. We represent this condi-
tional probability by p(MJG)dM. Similarly, we can define the condi-
tional densities P(PJM) and p(C|p,M).
Using these functions and basic statistics theory, we find that
p(C) that enters in Eq. (1) is given by
p(C) = |/p(C|P,M)p(P|M) / p(M|G)dGdPdM. (2)
153
-------
Equations (1) and (2) are the basic form of the "second highest"
concentration model that will serve as the focal point for the dis-
cussions that follow. We wish to emphasize here that (1) and (2) are
not an esoteric set of equations of academic value only. Rather, they
embody in highly condensed form all of the physical processes that
are believed to be important in the complex terrain dispersion problem
and therefore any model that is to yield reliable estimates of concen-
tration extema probabilities is reducible to the form of (1) and (2).
In order to provide a better understanding of the physical mean-
ings of each of the probability kernals in (2), we shall discuss each
below. From these discussions will emerge a picture of many of the
current needs in complex terrain modeling.
A. The Impingfeaient Kernel, p(C|P,M)
Recalling the definitions of P and M, we see that p(c|P,M) is
essentially a relationship among the concentration at some ground-
level site, say a hillside, and the height, width, depth, and material
content of the plume and the ambient flow speed, direction and tem-
perature stratification. It is precisely this relationship that has
been the subject of nearly all the laboratory and field studies of
complex terrain dispersion that have been conducted to date. In keep-
ing with the established terminology, we shall call p(CJP,M) the
impingement kernal because it embodies only the processes that bring
a given plume into contact with the ground.
Over flat terrain, which incidentally is also within the scope
of applicability of (1) and (2), this kernal is frequently assumed to
have the conventional Gaussian form. In this expression, P is re-
presented by the effective source height, a , a and source emission
rate Q; and M is represented by the wind speed u and the atmospheric
stability, that establishes the magnitudes of a and a . In this
instance, the mean flow is everywhere parallel to the ground and hence
material impinges on the surface only as a result of the action of
vertical velocity fluctuations, or eddies, in the turbulent boundary
layer. It is the universal character of steady-state turbulence over
a flat surface that is manifested in the universal Gaussian expression
assumed for p(C|P,M) in smooth terrain.
In complex terrain the mean flow is not everywhere parallel to
the earth's surface and therefore it can act in conjunction with tur-
bulence to transport plume material to the ground. In this case the
form of P (C|P,M) is influenced strongly by the size and shape of the
hill, by its surface roughness, and possibly by the nature of the
154
-------
terrain just upstream of the hill. It is speculated that if time and
length variables are non-dimensional by the proper combinations of
flow variables and characteristic length scales of the hill, universal
forms for the relationship p(C|P,M) exist that will be applicable in
complex terrain just as the Gaussion plume formula is assumed to apply
over flat areas. However, if we consider in detail the physical pro-
cesses that affect concentrations on the side of a hill, we see that
previous laboratory studies are not adequate in themselves to estab-
lish the true form of p because they do not account for the effects
of turbulent boundary layers.
No matter how large or small a hill may be and no matter how
stably stratified the flow may be, there exists a turbulent boundary
layer adjacent to the hill's surface. The turbulence is generated
in part by the shear in the. velocity field; partly by obstacles such
as boulders, trees, etc. that may dot the surface; and partly by
heating and cooling of the hillside. The last process is probably
very important on hills the size of those that are likely to exper-
ience plume impingement. (Considering the fact that power plant
stacks are typically 200-300 meters high and that plumes commonly rise
300-500 meters in stable conditions, we estimate that the character-
istic hill size of interest is 500-800 meters). On the sunlit side
of a hill, heating generates upslope flows that produce a convective
boundary layer that increases in depth from the base to the top of the
hill. Conversely, on the shadowed side and at night, surface cooling
generates a downslope flow that is deepest at the base of the hill
and very shallow at the hill top. On the windward side of a hill
this downslope flow would accentuate the surface layer wind shear
and possibly enhance turbulence levels.
Furthermore, hills several hundred meters tall usually have
smaller hillocks superposed on them and ravines cut in their sides.
These features generate turbulent wakes that are shed and advected
along the hill's surface to further augment the depth and intensity
of the turbulent boundary layer.
The essential point here is that when a plume impinges on a hill-
side, it must travel through this turbulent surface layer to reach
the ground. In the process plume material is diluted by mixing with
clean air in the boundary layer; and depending on the character of
the turbulence in the surface layer, for example, the extent to which
it is composed of wake turbulence, the plume material may meander
across the surface of the hill and thereby produce quite low time
averaged concentrations.
155
-------
Figures A-l and A-2 depict the processes we have just described.
The main purpose of these figures is to illustrate that the process of
dispersion within the immediate environment of the hill is practically
identical to that of plume fumigation from a stable layer aloft.
They also illustrate that concentrations at the surface of the
hill are dependent upon the depth of the hill's boundary layer (HBL)
at the point of plume entry, the intensity and scale of turbulence
in the HBL, the mean flow speed in the HBL, and the rate at which
plume material is fed into the HBL from the laminar layer above. The
latter is dependent upon the angle that streamlines in the laminar
region make with the top of the boundary layer. Since laboratory
simulations have purposely suppressed boundary layer formation by
using very smooth hills and approach planes (they have also ignored
thermal boundary layer effects), it would seem that those studies
cannot produce meaningful expressions for hill surface concentration
values. However, they should be able to yield quantitative informa-
tion on the manner in which hills distort the mean steamline patterns
in the laminar flow regions around them. Even this information must
be used with caution until it is demonstrated that the turbulent
boundary layer does not significantly alter the flow in the laminar
region.
In summary, we have argued that to obtain the functional form of
the impingement kernel p(C|P,M) under stable conditions in complex
terrain, we require both a quantitative description of the mean steam-
line patterns in the laminar flow regions around hills and certain
characteristic features of the surface turbulent boundary layer. The
specific quantities required are (1) the depth of the boundary layer
as a function of distance along the hill, (2) the intensity and scale
of the turbulence and the mean flow speed in the boundary layer as
functions of location, and (3) the rate at which fluid from the
laminar region enters the boundary layer at each point. As we stated
earlier, the last quantity is dependent upon the angle that steamlines
in the laminar region make with the outer surface of the boundary
layer. This is information that laboratory and specially designed
field studies can provide.
Such studies can also provide the boundary layer information out-
lined above and with it the diffusion experiments necessary to relate
that information to the rate of dilution of plumes that enter the
boundary layer and impact the ground. As we noted earlier, this dilu-
tion process is fumigation-like in nature. Having all this informa-
tion as a function of M, the physical characteristics of the hill
(including its surface roughness and that of immediate upstream
areas), and cloud cover and albedo from which we can estimate surface
156
-------
Turbulent Boundary Layer
wm^^^^ffi
Figure A-l. Diagram showing flow pattern of stack plume - nighttime.
Turbulent (Convective)
Boundary Laye
Figure A-2. Diagram showing flow pattern of stack plume-daytime.
157
-------
heat flux, we can formulate the kernel p(C|P,M) required to predict
the desired concentration probabilities. Later we outline field
studies that can address these subjects.
By virtue of its definition, p(C|P,M) tak«s into account only
those processes that affect a plume during its transport across the
surface boundary layer to the ground. The point where the plume con-
tacts the outer boundary layer, the plume's dimensions at this point,
its rate and envelope of meander, etc., are variables that are in-
cluded in the P event that we consider next.
Before proceeding, we should add that although the discussions
above dealt exclusively with the stable flow case, the same sets of
information are required for neutral and unstable flows as well. It
has been observed in the Geysers area of California, for example,
that maximum ground-level concentration occurs under stagnant, con-
vective conditions. This situation may be typical of sources in long,
deep and relatively narrow canyons or valleys where all of the stably
stratified flows transport material approximately parallel to the
contours of high relief features.
B. The Plume Rise/Transport Kernel, p(PjM)
The event P was defined as the occurrence of a specific, detailed
plume structure from the point where material leaves the source to
the point where it just enters the boundary layer of a terrain
feature. The probability density p(PJM) of this event is a function
of M and also of the stack height and diameter, and the temperature
and velocity of the exhaust. Thus, loosely speaking, p(PJM) is a
plume rise and transport model for complex terrain.
Over flat regions, empirical expressions are available for esti-
mating plume rise under stable and neutral conditions. Under these
conditions, it is assumed that once a plume reaches its equilibrium
level, it follows a straight path indefinitely far downwind from the
source. Neither this assumption nor the flat terrain plume rise
formulas are applicable a priori to complex terrain. One reason is
that low-level stable flows in complex terrain are not generally
horizontally homogeneous like they are over uniformly flat ground,
and thus plumes are often observed to wind along irregular paths and
to meander horizontally. Furthermore, in rough terrain regions,
vertical profiles of wind and temperature often exhibit extreme vari-
ations with height. The extent to which these variations affect
plume rise and spread is not known, but it must be established in
158
-------
order to obtain reliable estimates of p(P|M). In addition, the magni-
tude and frequency of plume meander must also be determined and
related to M and characteristics of the topography. For example, the
meander phenomenon may be due to vortex shedding behind large hills
or it may be caused by "sloshing" of cold air in a valley under the
influence of the wind stress exerted by the geostrophic flow aloft.
In these cases it might be possible to establish at least bounds on
the meander magnitude and frequency based on hill size, flow speed,
valley width, geostrophic wind speed, etc.
We will suggest field measurements to investigate these phenomena
later.
C. The Micro-Mesoscale Meteorology Kernel, p(M[G)
It has long been known that the differential heating and cooling
of the sloping surfaces in complex terrain generates local circulation
regimes much like the differential heating and cooling of adjacent
land and water surfaces gives rise to lake and sea breeze circula-
tions. It is the three-dimensional structure (speed, direction and
lapse rate) of these local flows that is embodied in our definition
of M.
One of the problems that complicates the modeling of dispersion
in complex terrain is that M does not normally exhibit the horizontal
uniformity that it does over flat areas well inland from large bodies
of water. In other words, over smooth, inland sites the smallest
horizontal spatial variations that exist in flows have length scales
of several tens of kilometers; whereas, in complex terrain the corres-
ponding scales can be an order of magnitude or more smaller. Thus,
over flat, inland terrain meteorological data from a rather coarse
network of stations are adequate to piece together a reasonable three-
dimensional description of the flow, i.e., M; and if sufficient
quantities of historical data are available, the probability density
p(M) can be estimated.
By definition
P(M) = lp(M|G)p(G)dG, cf Eq. (2) (3)
In complex terrain the density of meteorological stations is
generally much to small to determine M. However, there is sufficient
information to obtain G and p(G). This suggests that there are two
avenues of approach to the gathering of data needed to determine p(M).
159
-------
One is to undertake an extensive study of the physical processes
that govern micro-mesoscale flows in complex terrain with the objec-
tive of developing a model of p(M|G) applicable to all types of
terrain. The processes involved are almost totally dominated by the
terrain features, which incidentally have been documented by the
U.S.G.S. with extreme resolution for the entire country. This study
would involve field projects, physical modeling in the laboratory, and
numerical simulations. At this juncture, there is evidence that
a usable model of p(M|G) cannot be developed.
The second approach to gathering the requisite data is to deploy
at every proposed site of a power plant a network of meteorological
stations dense enough to resolve M, and to collect data for a long
enough period (at least two years), to determine p(M). This approach
would be expensive and time-consuming, and unless the measurement
network were designed properly, the information gathered might not
be useful at any other locations. One might argue that if this type
of an approach were taken, it would make more sense to bypass the
meteorological measurement altogether and to perform the tracer
studies necessary to determine p(C) itself. For example, one could
erect a tall, tower, at least as tall as the lowest anticipated plume
elevations, and release a different inert tracer steadily from
various levels of the tower. Automatic syringe samplers deployed
over the surrounding area and collecting hourly air samples on a con-
tinuous basis would provide the information required to determine
p(C) directly. No aircraft observations or mobile measurements of
any sort would be necessary. " The obvious advantage of this method
is that it circumvents the uncertainties that attend model calcula-
tions and thus it provides a much more reliable basis for making
control decisions.
The preliminary design, planning, site selection, etc., necessary
for large power plants require several years to perform. Thus, if a
monitoring study such as this one were begun at, say, the three most
preferred sites, by the time plant construction was ready to begin,
adequate concentration data would have been gathered to determine the
degree of emissions controls required. One problem with this method
is that a plume of passive tracer material does not expand as much
under stable conditions as that of power plant emissions because it
lacks the buoyancy. For this reason, measurements of wind and tem-
perature on the tracer release tower would be needed to estimate the
degree to whch buoyancy would enhance dispersion.
The purpose of this digression was to point out that there are
alternatives to modeling dispersion in complex terrain. We believe
160
-------
that a critical evaluation of these various methods should be con-
ducted as a part of the proposed modeling program, weighing the un-
certainties that attend model predictions, translated into dollar
cost of control equipment, against the cost of monitoring studies
where uncertainties are greatly reduced. Results of predictability
studies like those suggested earlier in this report would be of value
in this task.
Returning to the discussion of M and p(M), we can view M as the
resultant of a local, thermally induced circulation superposed on a
terrain augmented geostrophic flow. In instances where the geo-
strophic flow is weak, the local circulation will dominate, and hence
M will reflect the characteristics of the local terrain. Since these
features are constant, we would expect that for this set of "weak"
geostrophic flows and given cloud cover conditions and hour and day
of year, there is negligible variation within the corresponding set
of M. It is likely that in many areas the highest ground-level con-
centrations occur during the "weak G" situations. Where this is the
case, being able to describe the local circulation regime and knowing
the probability of "weak G" are all that are required with regard to
p(M) to solve (2).
Another simple case is that where the local circulation is so
shallow that plumes are able to break out of it and enter the geo-
strophic flow above. In this case we replace p(M) in (2) with p(G)
(see also Equation (3)) and M in p(C|P,M) with G. Recall that p(G)
is usually known.
In situations where plumes are frequently confined to the local
flow; i.e., the layer described by M, and neither the terrain nor the
geostrophic flow dominate M, the modeling problem is much more com-
plicated. It is this class of problems in which the highest modeling
uncertainties will exist.
2. SUGGESTED FIELD EXPERIMENTS
We have shown that there are three basic components of the
complex terrain modeling problem and have emphasized that all three
must be understood to meet current modeling objectives. One of these
components, namely the impingement processes, has already received
a great deal of attention. The other two have not.
We have also emphasized the uncertainties in model calculations,
especially in complex terrain environments, and have stressed the need
for studies aimed at assessing quantitatively the inherent limits on
the predictability of concentration extrema in such areas. The
161
-------
results of this study should be used to perform cost/benefit calcula-
tions to determine whether there are possibly methods other than
modeling that can best meet existing regulatory needs.
A. Studies Relevant to the Impingement Kernel, p(C|P,M)
The major information required to formulate p(C|P,M) is the
following:
1. A characterization of the three-dimensional streamline
patterns in the laminar flow region around hills during
stable conditions, expressed in terms of Froude number, hill
height and aspect ratio, surface roughness, surface heat
flux, etc.
2. Depth G.J the boundary layer adjacent to hills as a function
of distance along the hill, expressed in terms of mean flow
speed and stability in the laminar region, surface roughness
and heat flux.
3. The intensity and scale of turbulence in the boundary layer
as a function of distance along the hill, expressed in terms
of the same parameters as (2).
4. Mean flow speed in the boundary layer and fluid transport
rate into the boundary layer from the laminar region, both
expressed as in (2).
5. A mathematical relationship among the parameters listed in
(2) - (4) above and the rate of dilution of material as it
moves across the boundary layer from the laminar region to
the ground.
6. The range of variation in the dilution rate parameter (see
(5)) and the laminar region streamline patterns (see (1))
resulting from upstream random arrays of hills and depres-
sions of heights and depths less than that of the study
hill under various conditions of stability, heat flux, etc.
(This information is needed in the predictability studies,
to formulate an estimate of the magnitude of uncertainty
in model calculations.) These bounds would be expressed
in terms, say, of rms height of upstream terrain, Froude
number of study hill, etc.
There have already been a number of laboratory studies conducted
to investigate (1). However, the applicability of those results to
162
-------
complex terain is questionable because the Reynolds numbers achievable
in the laboratory are many orders of magnitude smaller than those
characteristic of rough terrain. Also, previous laboratory simula-
tions have not included upstream turbulent boundary layers or non-zero
heat fluxes.
Below we outline an experiment that can address all six of the
subject areas listed above and, in addition, provide virtually a con-
clusive answer to whether the results of the laboratory studies
mentioned are applicable to the atmosphere.
The basic experimental design involves a simulated hill con-
structed of flexible material such an canvas, mylar, or other suitable
material stretcned over metal hoops or rods to achieve particular
shapes. The hill is hung from a cable stretched between two towers
and, being flexible, it can be raised or lowered to any height
desired (see Figure A-3). The hill would be erected on a dry lake
bed; e.g., Lake Bonneville, where the ground surface is very flat for
many kilometers around and stably stratified winds are frequent.
Tracer material would be released from desired elevations and
distances upwind of the hill and sampled on the hill's surface by
instruments accessible through ports in the fabric that composes the
hill. All monitoring and recording of concentrations and flow pa-
rameters would be performed at a central facility located inside the
hill that is connected to all instruments by cables.
The boundary layer of the flow approaching the hill would be
controlled by raising fences of fabric panels as shown in Figure A-3.
These "roughness" elements could be replaced by or used together with
arrays of small hills, perhaps inflatable structures, to study dis-
persion on a hill surrounded by very rough terrain in its immediate
vicinity.
Positive and negative heat fluxes could be achieved on the hill
using perhaps fine water sprays on the exterior surface and gas space
heaters exhausting into fabric ducts on the inside hill surface.
Instruments to measure concentration and flow variables in the
hill's boundary layer could be attached to poles 2-3 meters long and
placed in position from ground-level inside the hill as the hill is
raised in place. These poles would be secured to the metal framework
that supports the fabric.
Some of the advantages of this experimental design over others;
e.g., instrumenting a real hill, are:
163
-------
^sp'^^
164
-------
1) It provides a laboratory with complete control over hill
size, shape, roughness and heat flux. For example, the hill
height could be changed gradually during a single experiment
to examine Froude number effects. Equally important is the
control possible over the characteristics of the turbulent
boundary layer of the approach flow. None of the parameters
just tested is controllable with a real hill.
2) Wide scope of experiments. All six of the subject areas
that require study to formulate the impingement kernel can
be investigated.
3) Having the monitoring center located inside the hill mini-
mizes the logistics problems of servicing the instruments
and relaying data to the monitoring site (e.g., cable rather
that radio data transmission is used). The flat terrain
site and the support towers that suspend the hill facilitate
measurements of the approach flow and filming operations.
The maximum hill sizes and flow speeds used in the EPA Fluid
Modeling Studies are 25 cm and 50 cm sec , respectively. The fluid
used in water, v - .01 cm sec . The hill size envisaged in
the present experiment is 25 meters (larger ones are possible) and
flow speeds of 4 m sec are possible. Thus, Reynolds numbers 100
times larger than those achieved thus far in laboratory studies could
be obtained. Favorable comparisons of results obtained over this
range of Reynolds number would provide strong justification for
continuing and expanding laboratory simulations.
The experiment just described is restricted to studies of stably
stratified flows. Modeling dispersion under free convective condi-
tions may be possible using numerical models currently in development.
To apply those models to complex terrain, we require at least the
following information:
1) Probability density of vertical and horizontal velocity
fluctuations over heated hill surfaces of various heights
and widths and the same information over shadowed surfaces.
2) Horizontal scale and characteristic vertical velocity of
thermals as a function of underlying hill and ravine sizes.
3) The parameters listed in 1 and 2 as functions of sun angle,
cloud cover, albedo, and surface slope and height of hills.
165
-------
These three sets of information could be gathered from aircraft
measurements of the 3-D velocity components taken at various altitudes
above complex terrain. The instantaneous data samples would have to
be correlated quite precisely with the horizontal position of the air-
craft so that the velocities could be analyzed as a function of the
terrain features.
The case of forced convection; i.e., strong surface heat flux
together with moderate horizontal winds, might possibly be modeled
numerically, but there are no studies of this sort in progress now.
The theoretical problems to be surmounted in developing such a model
are significant, and we must conclude that large uncertainties in
concentration predictions under these conditions will always exist.
B. Studies Relevant to the Plume Rise and Transport Kernal, p(PJM)
The major information requirements are:
1. The effects of vertical variations in wind speed and direc-
tion, temperature stratification, and turbulence energy on
the plume entrainment rates of heat and mass.
2. The effects of these same parameters on plume meander; i.e.,
spatial oscillations in the plume centerline.
The first studies conducted should be an exhaustive analysis of
the ability of current plume rise models to predict accurately plume
heights, widths, and thickness as functions of travel time in complex
terrain. Adequate data for this purpose exist in the Mohave, Navajo
and Four Corners data sets. These studies should correlate model per-
formance with the stack height and diameter buoyancy flux, etc.; and
the characteristics of the flow, such as wind shear magnitudes, verti-
cal variation in Richardson number, etc.
This study would expose the weak points in current models. For
example, the current methods of parameterizing entrainment may not be
adequate. Subsequent field studies using lidar probes of the plumes
that models simulate most poorly could gather the data necessary to
refine the rise and spread models. The error levels still present in
refined models would be used in the predictability studies.
C. Studies Relevant to the Micro-Mesometeorology Kernel p_(MJG)
To formulate this function requires a fundamental knowledge of
the processes that govern the three-dimensional structure of flows
within the first kilometer over valleys in complex terrain. There
166
-------
are two complementary approaches to formulating p(M|G). One is
physical and numerical modeling of the physical processes involved.
The second is an empirical approach based on extensive measurements
of key variables in complex terrain sites of various types. The
latter approach is necessary in any event because only it can supply
the data required to validate numerical and physical models. An
empirical model developed from field data would be used primarily to
estimate the frequency, given the climatology of the local geostrophic
flow G, with which a plume from a given plant might impact a parti-
cular terrain feature; and the type of vertical wind shear and tem-
perature stratification profiles, necessary for performing the plume
rise and spread estimates (i.e., P), that would occur in these cases.
We recommerd that physical and numerical flow models be developed
in the event that empirical formulations prove to be unacceptable.
Some of the questions that a field program should attempt to
answer are:
1) How deep are valley drainage flows and how does the depth
vary during the night? Can the depth be correlated with
the depth of the valley, distance down the valley, etc.?
2) Are the speed and direction of drainage flows at any level
functions of the slopes of surrounding terrain features of
a particular scale?
3) Is the temperature stratification within the drainage flows
strongly dependent on terrain features, or it is roughly the
the same from one site to another and one hour to another?
4) How strong must the geostrophic flow be at a particular site
to eradicate the drainage flow regime?
5) During weak geostrophic flows, does the same drainage cir-
culation regime occur at a given site (given the same time
of year, cloud cover, snow cover, etc.)?
6) Is there evidence that wind stress on the top of the cold
drainage flow produces a sloshing phenomenon that could
cause plume meander, and if so, is the sloshing frequency
relatable to geostrophic flow speed, valley dimensions,
etc.?
167
-------
7) Are the radiation inversions and vertical wind speed pro-
files that occur over high mesas that contain deep gorges
similar to those found over uniformly flat terrain?
8) To what horizontal distance from the base of an isolated
ridge does the drainage flow penetrate? How does the depth
vary along the ridge face?
To answer these questions and to gather the data required for
model validation, we recommend the following experimental approach.
1) Locate a geographical area that contains terrain features
of various types within an area small enough for the geo-
strophic conditions G to be assumed constant at any instant.
For example, valleys of various widths, depths and lengths;
mesas with river gorges; isolated ridges; etc. (One such
area exists near Prescott, Arizona.)
2) Within the various terrain types deploy a network of instru-
ments to measure vertical profiles of wind speed and direc-
tion and temperature up to about 300 m AGL continuously
through nighttime hours.
3) Release and track groups of tetroons at various levels with-
in and above the drainage flows (these data are necessary to
check the accuracy of particle trajectory predictions based
on fixed site wind observations).
4) Monitor meteorological variables on high terrain, make
radiosonde measurements, acoustic sounder probes, etc.
The experimental setup is shown in Figure A-4. Gathering the
vertical temperature and wind data continuously at numerous sites
would be very costly using conventional methods. In the insert of
Figure A-4, we have sketched a system that utilizes a captive balloon
to gather these data at unmanned sites. The concept of the design
is based on existing technology, but as far as is known, such a system
does not now exist. We recommend that some portion of the modeling
effort be devoted to developing low cost means of gathering the data
required.
3. OUTLINE OF PROPOSED MULTI-YEAR PROGRAM
The suggested work is divided into two programs operated in
parallel: One develops the required model and the other monitors
the model development progress and steers the overall effort.
168
-------
-*-TETROON
Tracking Aircraft
Control Center
(Radio Communication with Unmanned Remote Stations)
Typical Remote Station
Figure A-4. Experimental design for studying the micro-
mesometeorology of complex terrain.
169
-------
3.1 Year 1
3.1.1 Model Development.
A. Select site and construct simulated hill. Perform
hardware tests and begin preliminary experiments.
B. Comprehensive analyses of ability of current plume
rise and dispersion models to predict accurately plume
rise and expansion rates (a (t),a (t)) in complex
terrain. Exercise models using all available data
from Navajo, Four Corners, Mohave, etc. Correlate
model errors with magnitude of wind speed and direction
shear, inversion depth, stabililty, etc.
C. Investigate possible techniques for inexpensive routine
ground-based measurements of vertical wind and tempera-
ture up to about 300 m. Begin instrument development
program if necessary.
3.1.2 Monitoring of Model Development Progress.
A. Begin theoretical studies of predictability of second
highest concentration.
B. Investigate alternatives to modeling that can meet
regulatory needs, such as long-term tracer studies.
3.2 Year 2
3.2.1 Model Development.
A. Full scale operation of simulated hill facility. Experi-
ments to include:
(1) Flow patterns around hills of various shapes at
various Froude numbers for comparison with laboratory
simulations,
(2) Develop empirical relations between plume dilu-
tion rate and turbulent boundary layer characteristics,
(3) Establish the range of dilution rates caused by
random distributions of upstream terrain irregularities.
170
-------
B. Based on outcome of plume rise studies conducted during
first year, perform lidar probes of power plant plumes
at sites and under met conditions where plume rise and
spread models were found to perform worst. Measure
entrainment constants, energy dissipation rates and
other parameters needed to refine the model.
C. Field test instruments for vertical wind and temperature
measurements. Begin analyses of existing complex terrain
data looking for influence of surrounding terrain on
depth of surface inversion and flow speed and direction
profiles at night; threshold of geostropic flow for
micro-meso regime breakdown, etc.
D. Begin development of physical and numerical models of
flow in complex terrain sites to be studied in the field
next year.
3.2.2 Model Development Progress.
A. Formulate impingement kernel for cases studied to date
and test against Mohave, Navajo, Four Corners, etc.,
data.
B. Compare refined plume rise-spread model predictions
with existing complex terrain data.
C. Using results of tasks A and B above and first year
predictability study, estimate modeling error assuming
perfect meteorological data, i.e., perfect p(M). The
outcome will be:
• Modeling objectives are achievable if p(M) can
be determined with accuracy E. (This will lead to the
procedure in Year 2, Model Development, Section D).
• Modeling objectives unattainable even if p(M) is
known precisely.
1. Examine cost-effectiveness of alternatives to
modeling investigated last year, such as long-term
tracer study. If viable method found, abandon major
modeling effort.
171
-------
2. Reformulate air quality standards to achieve
predictability — resume model development at Task D
of Year 2.
3.3 Year 3
A. Continue studies at simulated hill experimental center.
Examine heat flux effects on flow patterns and boundary
layer turbulence.
B. Final refinements of plume rise and spread model,
culminate in final form of p(P|M).
C. Begin full scale study of micro-mesometeorology at sites
chosen previous year.
D. Begin validation and refinement operations on physical
and numerical flow models using results from field study
of Step C.
E. Begin aircraft studies of free convection for use in
formulating p(C|P,M) for unstable conditions.
3.4 Year 4
3.4.1 Model Development.
A. Concluding impingement studies at simulated hill site
culminating in final form of p(C|P),M) for stable cases.
B. Develop empirical formula for p(M|G) using field data.
C. Formulate expression for p(M|G) from refined physical
and numerical models.
D. Continue developments of p(C|P,M) expression for free
convective case. Begin numerical model study for forced
convective case.
3.4.2 Model Development Progress.
A. Test these expressions for p(M|G) against data from
Mohave, Navajo, Four Corners, etc. The outcome will
be:
172
-------
• p(M) can be specified with required accuracy E,
in which case, continued model development.
• p(M) cannot be specified with accuracy E using
model only, in which case:
1. Prepare guideline for performing necessary
meteorological measurements at any proposed plant site,
i.e., measurements needed to determine p(M). Resume
model development.
2. Investigate alternatives to modeling, such as
long-term tracer studies — if viable method, abandon
modeling effort.
3. Reformulate standards — resume model develop-
ment.
3.5 Year 5
A. Amalgamate all studies into final form of model of second
highest concentration and specifications for input data
requirements.
173
-------
APPENDIX B
PREWORKSHOP MATERIAL PREPARED BY NAWC
On June 13, 1979, NAWC forwarded a Workshop package to all the
participants with information pertinent to the tasks facing each
panel. Of particular concern was the need for the panels to quickly
focus upon the critical issues and to facilitate the flow of infor-
mation between the panels in order to achieve the goals of the Work-
shop. The information package included the "strawman" document by
Holzworth and Snyder (Appendix A); publications by Barr, et al. ,
(1977) from the Albuquerque Workshop on Complex Terrain Dispersion,
by Argonne National Laboratory (1977) from the Chicago Specialist
Conference on EPA Modeling Guideline; and by Hanna, et al., (1977)
from the Boston Workshop on Stability Classification and Dispersion
Parameters.
In addition, NAWC prepared material for distribution to the panel
leaders on June 28, 1979, which is presented below. Subsequently,
each panel leader contacted their respective panel members with both
•written and oral instructions prior to the start of the Workshop in
Raleigh.
174
-------
INFORMATION FOR EPA WORKSHOP PANEL LEADERS
PREPARED BY NAWC
1. CONCEPT OF PANEL'S ROLE
1.1 Model Evaluation and Application/Model Development and Analysis
Panels
These panels must generate the basic approach of the Workshop as
a whole during the first session (Tuesday AM). They must review and
select from the basic scenarios which approach EPA should follow in
their overall program.
Specifically, the goals of modeling must be considered and es-
tablished and the relative importance must be given on such topics as:
• Centerline plume impaction
• Maximum concentration value
• Type of models to be considered and their relative merits
• Measurement needs
• Scale of experiments -
Simple controlled (Snyder) versus full-scale power plant
studies
• Required resolution
• Priority of various parameters
• Physical modeling
• Plume rise/initial dilution
Furthermore, the Model Evaluation and Application Panel must address
such topics as peak/mean ground-level impacts and locations of im-
pacts; i.e., the overall goal of a model development program relative
to the users.
175
-------
1.2 Experimental Design Panel
Initially this panel should discuss in general the ramification
of carrying out simple, controlled tracer experiments versus large
scale power plant experiments. They must address the relative pro-
grams and budgetary requirements and provide interaction with the
Measurement Techniques Panel on what is required to carry out dif-
ferent programs.
Once the scale, resolution, and emphasis of the experiment have
been decided, discuss:
• Type of tracer and gas measurements
• Ambient measurements
• Potential locations
• Logistics and costs
• Surface measurements and instrumentation
• Aircraft instrumentation
• Tracer samples
• Physical modeling techniques
1.3 Measurement Techniques Panel
Again, this panel will have to start off with general assessments
of measurement requirements for different types of field programs. Of
particular concern is the area of turbulence versus meteorological
parameter measurements. A lack of knowledge of relative rates of
diffusion for different elevations and stability, as well as different
terrain, has resulted in difficulties with model verification.
Once the experiment design parameters have evolved, discuss
methods of data acquisition and recording. Discuss accuracies and
calibration techniques for meteorological measurements, gas analyzers,
and tracer analyzers. Discuss the applicability of various types of
techniques and potential for using newly developed remote sensing de-
vices for augmenting experiment usefulness.
176
-------
Discuss practicality and costs. Interact with
• Data Management - Discuss data reporting formats, data
control, instrument accuracies.
1.4 Data Management and Quality Assurance Panel
Before this panel receives clear indications of the basic goals
and concepts which the Modeling Panels will zero in on, this panel
should address such questions as how to utilize the wealth of data
collection in the past by government, utility industry, and private
firms.
• Discuss potential methods of collecting and utilizing
data from past tracer and measurement programs
• Discuss quality assurance methods in view of experiment
emphasis
• Discuss methods of data management which allow easy display
and correction
• Consider costs, material requirements
Interaction
1) Discuss best data formats for data from past and future
experiments with model application and evaluation panel.
2) Discuss reporting methods with measurement techniques
and experiment design panels.
2. INTERACTION OF PANELS
Guidance provided by George Holzworth, the EPA Project Officer
for the Workshop, suggests that both source-oriented (as per example
in Holzworth's "strawman"), and receptor-oriented (as per example in
Snyder's "strawman"), can be accomodated in the long-term EPA research
plans. However, specific experiment design, scheduling, priority of
research, and interaction with other programs (i.e., EPRI, DOE, and
NOAA) are major considerations for the Workshop.
The present Workshop plan calls for alternate panel and plenary
sessions arranged so that after an initial plenary session on Monday,
there are at least three separate feedback loops between individual
177
-------
and assembled panels on the succeeding days: Tuesday through Thursday.
Separate panel leaders meetings are part of the loops, with their
closing session on Friday. Figure B-l diagrams the information
flow in each loop. Arrows pointing both ways . indicate flow be-
tween panels at plenary sessions. The single direction arrows,
on the other hand, are meant to emphasize the natural flow of develop-
ing ideas from panel to panel. The MD&A Panel heads the column be-
cause it has to be the original source of a well-defined quanti-
tative demand on the other panels for development and critiquing
of the MD&A panel's particular idea. It seems natural for MD&A to ask
ED how they would test the idea in an experiment; then ED asks MT
about possible measurements techniques, and MT passes the idea in its
current stage of development to DM & QA to acquire their input on how
the data can be managed. Finally, ME&A plays some conceptual simula-
tions to see whether the particular idea, at this stage, will fly,
and if so, whether it meets their set of applicability criteria. In
succeeding loops the particular idea is either refined or discarded.
Finally, on Friday, the panel leaders polish a draft report covering
the surviving ideas and fit them together as a consistent set of
program recommendations.
Probably not all ideas finally developed will appear in the first
loop, but may be appended as inspiration produces them along the way.
The inspiration for a new idea may come from any panel and should be
inserted into the pipeline at the first available plenary session.
At the initial plenary session on Monday many ideas will be formulated
and these will serve as the initial "seedbed". Some ideas not meeting
wholehearted support at this session could still be subject to
processing at least once through the loop to make sure everyone has
a chance to think the matter through.
A more specific example of how the Panels will interrelate is
shown in Figure B-2, which is a flow chart showing how some specific
recommendations would evolve. This would begin with listing of the
general objectives and parameters to be measured by the Model Develop-
ment and Model Applications Panels. Their ideas would be passed on
to Experimental Design, who would devise an instrumentation array
design and recommend sampling modes (time of day, time of year, dura-
tion, etc.). Their list would go to the Measurement and Data Manage-
ment Panels, who would recommend specific instruments and list data
handling/QA-QC, respectively. Those recommendations would go back
to the other panels again, and each would refine the original design
and develop a summary of recommendations.
A suggested schedule for the week is shown in Figure B-3. The
specific recommendations, as shown in the flow chart, are connected
178
-------
MODEL DEVELOPMENT
& ANALYSIS
(MD&A)
EXPERIMENTAL
DESIGN
(ED)
MEASUREMENT
TECHNIQUES
(MT)
DATA MANAGEMENT
&
QUALITY
ASSURANCE
(DM & QA)
MODEL EVALUATION
&
APPLICATION
(ME & A)
s
•^ /
s ^
/• s
*v f
s- -s
X '
/ •>)
V S
>s ^
x V
^
y
's.
s-
Figure B-l. Flow diagram for Workshop,
179
-------
MODEL DEVELOPMENT
& ANALYSIS
MODEL EVALUATION
& APPLICATION
1. GENERAL OBJECTIVES
2. PARAMETERS TO BE MEASURED
TUES. A.M.
EXPERIMENTAL
DESIGN
1. INSTRUMENTATION ARRAY
2. SAMPLING DURATION/AVERAGING PERIOD
WED. A.M.
MEASUREMENT
TECHNIQUES
1. RECOMMEND SPECIFIC
INSTRUMENTS
WEb. EVE.
DATA MANAGEMENT
&
QUALITY ASSURANCE
1. HOW TO INCORPORATE DATA
FROM OTHER STUDIES
2. DATA REDUCTION/STORAGE WED. EVE.
3. QA/QC
TO OTHER PANELS
Figure B-2. Flow diagram for specific complex terrain model study
recommendat ions.
180
-------
#1 #2 #3 #4 #5
DAY/ MODEL EXPERIMENTAL MEASUREMENT DATA MODEL
TIME
MON.
PM
TUE.
AM
TUE.
PM
WED.
AM
WED.
PM
WED.
EVE.
THUR.
AM
DEVELOPMENT
1
G E r
PARAMETERS
\
N
SELECTED
PANEL
TOPIC(S)
DESIGN
PLENARN
1 E R A L C C
SELECTED
TOPIC(S)
;<>
V1 P I
\
\ ^
INSTRUMENT
ARRAY
SAMPLING N
DETAILS \
TECHNIQUES
' S E S S I [
INSIDER;
SELECTED
TOPIC(S)
. E N A R Y^
>^^
^
SELECTED
PANEL
TOPIC(S)
\
\\
MANAGEMENT
) N S
^ T I 0 N S
SELECTED
TOPIC(S)
^
*f^
*s^
HOW TO
INCORPORATE
DATA FROM
OTHER STUDIES
EVALUATION
PARAMETERS
S*
SELECTED
PANEL
TOPIC(S)
PLENARY "v//Vs
COST
ESTIMATES
FOR #1
^--
REFINE
ORIGINAL
DESIGNS
COST
ESTIMATES
FOR #2.
&2^3i
>^ ~y
REFINE
ORIGINAL
DESIGNS
\^
V
\
RECOMMEND
SPECIFIC
INSTRUMENTS
••*
COST
ESTIMATES
FOR #3
s^
>^
^
DATA
REDUCTION
& STORAGE
AND QA/QC ^
COST
ESTIMATES
FOR #4
COST
ESTIMATES
FOR #5
l/*s
/
REFINE
ORIGINAL
DESIGNS
Figure B-3. Suggested schedule for Workshop as detailed by Panel Leaders in
their letters to Panel Members.
181
-------
with clashed lines. In addition, each panel should have a list of
selected discussion topics for the periods noted on the schedule.
The Panel Leader should select the day's topic(s). Also, each panel
will address the anticipated costs for the program recommended by
that panel and determine the percentage of total program funds which
should be allocated to each aspect of the study (e.g., Model Develop-
ment 25%, Experimental Design 15%, etc.). Each Panel Leader would
be responsible for summarizing the consensus of recommendations of
his panel and interfacing with other Panel Leaders. Continuing
dialogue between the leaders will allow the panels to continually
update and refine their design recommendations.
Already a variety of ideas have been proposed and appear in
various memos and letters that have been distributed to Panel Leaders
and participants. These ideas represent responses to questions about:
1) Deformation of streamlines and parallel deformation of
the concentration pattern in passing over complex terrain,
under different stability regimes.
2) Changes in dispersion over complex terrain from that for
flat-land stability types.
3) Changes in entrainment and plume rise factors over complex
terrain.
4) The nature of slope flows in complex terrain, and the role
of radiation in producing them.
5) The nature of flow separation and its effect upon dispersion.
6) The validity of using stagnation point concentration in
the case of direct impaction of a plume on a cliff (and
how thit type of flow can come about in nature, if at
all).
7) The problem of transport and dispersion of stagnation
periods.
The preparation of a final smooth report on this Workshop will
be greatly facilitated if there is an orderly recording of ideas and
their development, including reasons for their discard, if that is the
action taken. It would also be useful to name the persons in the
notes responsible for initial introduction of an idea and those
responsible for various seminal amplification of an idea. It is not
182
-------
necessary to record al 1 the nuances of a. debate, but careful record-
ing of technical details (especially with instrumentation) would be
helpful. It is easy to assume that everyone is familiar with code
names and jargon. The principle on which an instrument works should
be given along with its standard specifications.
It is most critical to the successful conclusion of the Workshop
that there are adequate opportunities throughout the Workshop for
interchange of information between the panels.
Since the Experimental Design, Measurement Techniques, and the
Data Management Panels are very much dependent on the specific
directions to be generated by the Modeling Panels, these directions
must be conveyeu early, no later than at the Tuesday Evening Executive
Panel Leader Session.
183
-------
APPENDIX C
POST-WORKSHOP COMMENTS BY PANEL MEMBERS
During a two-week period immediately following the Workshop,
the panel members had an opportunity to review and comment on the
written Panel Recommendations to EPA. Appendix C contains specific
comments, received by Panel Leaders and the Project Director, which
are presented for the purpose of amplifying the viewpoints by panel
members which they believed required further clarification.
184
-------
3 August 1979
Mr. Maynard E. Smith
Meteorological Evaluation Services, Inc.
134 Broadway
Amityville, New York 11701
Dear Maynard:
As we agreed in our telephone conversation concerning the
report of our panel, I am writing you about a potential problem that
was not discussed by the panel. This potential problem is the
generation of turbulent wakes in the vicinity of the terrain interface
during the field experiments by the hardware (masts, sampling devices,
support equipment) and by certain activities within or near the
sampling arrays that would significantly enhance turbulent mixing or
otherwise alter the boundary layer characteristics during the stable
regime. These wake effects are most likely to be significant for
the small scale experiment but may be significant for the intermediate
and large scale experiments as well, depending on the relative height
and density of the natural roughness elements that are presented. It
is clear that the potential wake effects or flow disturbances
attributable to the measurement techniques, hardware and support
activities should be carefully evaluated as part of the experimental
design process and that these effects may limit the scale of the
experiment.
Sincerely,
Harrison E. Cramer
!-IEC:bjs
185
-------
REFLECTIONS ON THE SMALL HILL STUDY
by
William H. Snyder
As originator of the small hill study concept, I take this
liberty to defend the plan. I feel that many aspects of the plan
have been misunderstood, indeed, misconstrued, perhaps only because
all have not had the opportunity to perceive the plan as first
proposed. Numerous questions were raised at the Workshop; I will
attempt below to answer the most frequent and serious ones.
1. Cost of Meteorological Towers: I have been assured by Mr.
Michael Fleissner of Rohn Mfg. Co., Peoria, 111., that the erected
cost of a 152 m (500 ft) meteorological tower, complete with guys,
lights and ladder, would range between $32,000 and $35,000. An
upper estimate of the cost of installing anemometers is 6 man-weeks
or $6000 (this estimate does not include the cost of the anemometers
themselves). It is evident that substantial savings would be achieved
by constructing 5 towers at essentially the same site and the same
time. Hence, the cost of 5 meteorological towers is well within the
constraints of the budget (~ 5%).
2. Scaling to a Large Hill: The primary goal of the small hill
study was to understand the physical mechanisms involved in the
impingement and/or diffusion of plumes to a hill surface. It was
definitely not conceived as a study whose results would be applied
ipso facto to the larger hill or "real" terrain. (From that point
of view, however, it is interesting to note that often the results
of fluid modeling studies conducted at scale ratios of 300:1, 1000:1,
even 10,000:1, are applied to full scale situations, whereas,
scaling the small hill results by factors of 2:1, 5:1, at most 10:1,
appeared to be an insurmountable obstacle!). The point is that
once we understand the physical mechanism involved in plume impinge-
ment/diffusion onto a sloping, three-dimensional surface, we will
then have a much more clear idea of how to construct a mathematical
model to handle the problem.
An example may help to clarify the point. Laboratory studies,
guided by fundamental principles, have shown that, under strongly
stable conditions, fluid is constrained to move in horizontal planes.
A plume with an elevation lower than a nearby hill, therefore, cannot
go over the top, but must, irstead, travel around the sides. If the
wind vector at the source is aimed precisely at the centerline of
186
-------
the hill, laboratory studies have shown that resulting surface con-
centrations will be essentially equal to what they would have been
at the centerline of the plume in the absence of the hill. The
laboratory studies have also shown, however, that surface concentra-
tions are drastically reduced if the wind direction is changed
only slightly. What is obviously lacking in the laboratory studies
is the natural wind meander, which can only be obtained from field
experiments. Once the physical mechanism, i.e., wind or plume
meander factor, is understood, it will be a relatively simple matter
to extend the concept to larger hills, where the standard deviation
of wind direction fluctuations has been measured. The small hill
study was conceived as a project wherein detailed and extensive
measurements con Id be made on a manageable scale. The size of the
hill (100 m) was chosen to be large enough to frequently be much
higher than the very thin (sometimes absent) jLurjcml^nt^ surface
layer under very stable atmospheric conditions, yet small enough
to eliminate the horrendous logistics problems of large scale
experiments.
Plume meander is obviously not the only missing factor in
laboratory or mathematical models. We also need to know the effects
of anabatic and katabatic winds; how closely the plume centerline
approaches the hill top as the stability is reduced from very
stable through neutral; how frequently does "strong" stability
occur and how is wind meander related to stability; how are the
rules changed as the wind profile shape or the density profile
shape vary, etc. If these kinds of questions can be answered from
a small hill study—and I believe they can from a properly conducted
one—we will have progressed a long way towards our goal of construct-
ing a model suitable for use in complex terrain.
3. Real-Time Data Collection: At least two panels at the
Workshop indicated that time averages, not to exceed 10 minutes in
length, were essential to understand that some real-time data would
be very useful, but dismissed the idea of total real-time data collec-
tion because of a perceived excessive expense. I contend that a
real-time feedback system can be established that will be relatively
simple, reliable, and of quite reasonable cost—even more economical
than any other system when all factors are considered over the total
duration of the small-hill study. There are three questions to be
answered here: (1) What tracer/sampling system could be used?
(2) What time resolution is necessary? (3) How much would a real-
time feedback and recording system cost?
(a) Tracer/Sampling System: I propose to use ethane
(COHC) as the tracer and flame ionization detectors (FID's)
^ O
187
-------
for concentration measurements. Ethane is neutrally buoyant,
sufficiently inert chemically and readily available at reasonable
cost. FID's (total hydrocarbon analyzers) are we 11-understood,
reliable, readily available, fast response (~ 1 sec), linear
in the range of 0.5 to 10,000 ppm (i.e., can measure dilutions
on the order of 2x10 ), and are not terribly expensive (~ $2500/
unit). They produce an analog output voltage that is readily
digitized. Adequate surface concentrations (i.e., above back-
ground) may be obtained with reasonable release rates of ethane
(calculations are easily made). It appears quite feasible to
rig a tubing network such that samples from 10 locations on the
hill surface are continuously pumped past each of 10 FID's.
Through a switching network, then, the FIDs are able to repeti-
tively sample each of the 100 locations on a once-per-ten-second
basis—conceivably much faster than that—as laboratory studies
have rather easily obtained frequency responses of 300 hertz.
(b) Time Resolution Required: To fully understand the
physical mechanisms, resolution of concentration fluctuations
as well as wind speeds on time scales much shorter than 10
minutes is essential. The necessity for time resolution of the
wind fluctuations is obvious. The necessity for resolution
of concentrations is not so obvious, but laboratory studies
and some large scale field studies have shown that semi-coherent
vortex shedding may occur in the lee of hills under strongly
stratified conditions. This vortex shedding in the lee causes
an oscillation of the plume on the upwind side of the hill, which
can markedly reduce surface concentrations. The vortex shedding
period is expected to be on the order of a few minutes, so that
this physical mechanism would be completely obscured by observing
only 10 minute averages. Similarly, any other physical mecha-
nisms taking place on time scales smaller than 10 minutes would
be obliterated by 10 minute averages.
(c) Data Feedback and Recording: The above discussion
points towards 10 second averages at each of 100 locations for
a presumed study period of 3 hours, leading to a total of
100,000 samples/day. This sampling rate and number of samples
obviously points to a mini-computer recording on magnetic
tape. A system to handle this job would cost under $100K.
It can easily be set up to automatically and periodically
sample each analyzer, to record each sample on magnetic tape,
and to display real-time and/or running-average information
at all sampling locations simultaneously. At the end of each
study period, all information is immediately available for
comparing with models and for user distribution. This system
188
-------
is to be compared with something equivalent to syringe sampling,
where a new syringe is filled every 10 minutes at each of 100
locations for a period of 3 hours, yielding 1800 samples that
must be individually labeled, handled, analyzed, recorded,
flushed, etc., each day. The choice is obvious. A comparable
system for wind measurement and recording is likewise in order.
4. Source Configuration and Management: If the vertical width
of the plume is too large compared to the hill height, we will be
unable to learn much about the flow structure over the hill, because
surface concentrations would be essentially uniform. If its vertical
width is too small, it will not be possible to resolve the plume
structure as it approaches or encounters the hill, e.g., lidar
measurements would be unable to resolve the plume as it approaches
the hill, and similarly, too many samplers would be required on the
hill surface. An appropriate number appears to be a = 10 to
15 m for a 100 m hill. Such plume widths could be obtained oy placing
the source an appropriate distance upwind for the existing stability
class. From Turner's workbook, indicated distances are 1 km for
F-stability and 400 m for D-stability (this is another reason for
having a mobile source). An alternative would be to configure the
source to obtain the desired initial plume widths, for example, by
emitting the tracer through an array of tubes spaced in a rectangular
matrix in the y-z plane.
Several Workshop participants expressed the opinion that once
the source was fixed and sampling began, the source should not be
moved again for the duration of the study period, i.e. , 3 hours.
I agree with this concept in principle, but, in practice, it may
be entirely appropriate to move the source. Suppose, for example,
after collecting samples for a one-hour period, there is an obvious
and permanent shift in the wind direction of 180°. This could be the
case (and possibly, quite predictably) shortly after sunrise if the
100 m hill were located in a broad sloping valley. It seems to me
fruitless and wasteful of resources to continue the scenario another
two hours; it would be far better to move the source 180° around
the hill and begin a new scenario, obtaining useful data. It would
be obvious from the wind records, in any event, that the three-hour-
average concentration for the first scenario would be one-third
the one-hour-average concentration. Of course, strict protocols
would be mandatory and decisions would have to be based on prior
experience, but a rigid specification of an absolutely fixed source
position is too restrictive.
5. Anabatic and Katabatic Winds. The lack of knowledge concern-
ing the nature, strength, and thickness of anabatic and katabatic
189
-------
winds was apparent at the Workshop, but the general consensus appeared
to be that they would be small to insignificant on the 100 m hill,
yet predominant on a 500 m hill. Because there are numerous
physical mechanisms to be understood before constructing an adequate
mathematical model (the nature of the impingement process, the close-
ness of approach of streamlines to the hill surface, the wind meander
factor, anabatic and katabatic winds, etc.), it is logical to
eliminate as many variables as possible in order to fully understand
the remaining processes. If anabatic and katabatic winds are insig-
nificant on the 100 m hill, it is therefore a logical first step.
In any event, a few smoke studies before any equipment is set up
would provide very useful knowledge on the existence and nature of
any such thermally generated surface winds, so that we may be prepared
to quantify them as necessary.
190
-------
August 9, 1979
Dr. Bruce A. Egan
Environmental Research & Technology, Inc.
3 Milata Drive
Lexington, Massachusetts 02173
Subject: Your memorandum AQC-833 of July 25, 1979, concerning EPA
Rough Terrain Workshop
Dear Bruce:
I was away from the office until your deadline for comments so
I am sending a copy of this letter to Einar Hovind.
I find no fault with the panel report. It appears to reflect
most of the concerns that were expressed at the panel sessions.
Upon reflection and a plotting to scale I am concerned that 100
tracer samplers will not be enough to define the ground level concen-
tration pattern and concentration gradient normal to the hill. For
examp1e, an oval hill that peaks at 120 meters with a six to one
slop,, .vould require 25 samplers at 20 meter spacing along the 50
meter contour level for just one quadrant. The recommendation should
be 200 samplers.
191
-------
Dr. Bruce A. Egan page 2 August 9, 1979
The Workshop was enjoyable and I look forward to seeing the
report.
Very truly yours,
TRC - THE RESEARCH CORPORATION
of New England
Norman E. Bowne,
Vice President & Chief Scientist
NEB/mfr
cc: Einar L. Hovind
North American Weather Consultants
192
-------
August 13, 1979
Mr. Einar L. Hovind
Vice President, Air Quality
North American Weather Consultants
600 Norman Firestone Road
Goleta, California 93017
Dear Einar:
I am enclosing a copy of my letter of August 1 to Maynard. Note
the circled third paragraph*. If such a distance range is not already
covered under the group that waas working on project design, I think
it is important to show how difficult it will be to have the tracer
"hit the hill" as the release point is moved from two km to four km
up wind.
For your general information Gerard DeMarrais telephoned Montana
Power last week to inquire as to the availability of the small hill
upon which the meteorological tower is located for field experimental
work. Unfortunately that hill is not 100 meters high and it is more
conical in shape than the 3 to 1 aspect ratio recommended by the Work-
ing group of which Maynard Smith was the head.
*"Perhaps an item 8 under the Small Scale effort should indicate a
distance range up-wind from the hill be recommended for tracer re-
lease. I personally would like to see some data collected at both
2 and 5 km up-wind from the first possible impingement point on the
hill. Such an investigation will clearly illustrate the less precise
air flow pattern for a greater distance. Perhaps it could be better
stated as not less than 2 km up-wind with a second release effort
not more than 6 km up-wind."
193
-------
Mr. Einar L. Hovind page 2 August 13, 1979
I appreciate the opportunity to join with others at the Workshop
in North Carolina. However, I remain unconvinced that the findings
of a detailed study project using a hill in the center of a valley
will answer the real life problems of the imagined impingement on the
shoulder terrain many kilometerss away from a power plant in the
center of the valley.
Sincerely yours,
Loren W. Crow, CCSJ
LWC:dd
Enc.
194
-------
TECHNICAL REPORT DATA
(Please read Instructions on the reverse before completing
1 REPORT NO.
EPA-600/9-79-041
3. RECIPIENT'S ACCESSION-NO.
4. TITLE AND SUBTITLE
WORKSHOP ON ATMOSPHERIC DISPERSION MODELS
COMPLEX TERRAIN
IN
5. REPORT DATE
November 1979
6. PERFORMING ORGANIZATION CODE
7 AUTHOR(S)
E. L. Hovind, M. W. Edelstein, and V. C. Sutherland
|8. PERFORMING ORGANIZATION REPORT NO.
9. PERFORMING ORGANIZATION NAME AND ADDRESS
North American Weather Consultants
600 Norman Firestone Road
Goleta, California 93017
10. PROGRAM ELEMENT NO.
1AA603 AB-050 (FY-79)
11 . CONTRACT/GRANT NO.
68-02-3223
12. SPONSORING AGENCY NAME AND ADDRESS
Environmental Sciences Research Laboratory - RTP, NC
Office of Research and Development
U.S. Environmental Protection Agency
Research Triangle Park, North Carolina 27711
13. TYPE OF REPORT AND PERIOD COVERED
Final 3/79 - 10/79
14. SPONSORING AGENCY CODE
EPA/600/09
15. SUPPLEMENTARY NOTES
16. ABSTRACT •
During the period of July 16-20, 1979, an EPA-sponsored Workshop was conducted
in Raleigh, North Carolina, to address problems associated with plume dispersion
modeling in complex terrain. This Workshop was intended to aid in the design of
a proposed EPA-funded research program dealing with this topic. Workshop participants]
represented a cross-section of environmental organizations, control agencies,
industry and the scientific community with technical background and expertise in
complex terrain modeling and field studies.
The Workshop was organized into five panels: Model Development and Analysis;
Model Evaluation and Application; Experimental Design; Measurement Techniques; Data
Management and Quality Assurance. This report contains the unabridged recommenda-
tions by each panel as summarized by the Panel Leaders. Also included are presenta-
tions by invited speakers who presented summaries of related complex terrain dis-
persion programs currently being sponsored by industry and by government agencies
other than the EPA.
17.
KEY WORDS AND DOCUMENT ANALYSIS
DESCRIPTORS
Air pollution
*Plumes
^Atmospheric diffusion
^Terrain
*Mountains
^Mathematical models
*Meetings
b.IDENTIFIERS/OPEN ENDED TERMS
c. COSATl Field/Group
13B
21B
04A
08F
12A
05B
13. DISTRIBUTION STATEMENT
RELEASE TO PUBLIC
19. SECURITY CLASS (This Report)
UNCLASSIFIED
21. NO. OF PAGES
213
20. SECURITY CLASS (This page)
UNCLASSIFIED
22. PRICE
EPA Form 2220-1 (9-73)
195
------- |